Ted Gournelos - Joshua R. Hammonds - Maridath A. Wilson - Doing Academic Research - A Practical Guide To Research Methods and Analysis-Routledge (2019)

Doing Academic
Research
Doing Academic Research is a concise, accessible, and tightly

organized overview of the research process in the humanities,
social sciences, and business. Conducting effective scholarly
research can seem like a frustrating, confusing, and unpleasant
experience. Early researchers often have inconsistent knowledge
and experience, and can become overwhelmed – reducing their
ability to produce high quality work.
Rather than a book about research, this is a practical guide to
doing research. It guides budding researchers along the process
of developing an effective workflow, where to go for help,
and how to actually complete the project. The book addresses
diversity in abilities, interest, discipline, and ways of knowing
by focusing not just on the process of conducting any one
method in detail, but also on the ways in which someone might
choose a research method and conduct it successfully. Finally, it
emphasizes accessibility and approachability through real-world
examples, key insights, tips, and tricks from active researchers.
This book is a highly useful addition to both content area
courses and research methods courses, as well as a practical
guide for graduate students and independent scholars interested
in publishing their research.
Ted Gournelos is Associate Professor at Old Dominion
University, USA, specializing in cultural studies, digital media,
and strategic communication. As CEO of the consulting firm
Story | Strategy, he works with organizations to fuse success
and social progress. His broad background in media, culture,
and business helps him bridge the gaps between education and
application.
Joshua R. Hammonds is Assistant Professor at Rollins

College, USA, specializing in interpersonal and professional
communication. His teaching and consulting work focuses
on the management of relationships through both qualitative
and quantitative methods. He currently develops research
interventions to improve communication competency in the
medical industry and increase employee engagement in the
corporate world.
Maridath A. Wilson is Assistant Professor and E-Resources

and Serials Librarian at Rollins College, USA. She specializes
in electronic resources management and information literacy
instruction to students of the humanities and social sciences.
She is interested in issues of constructed authority, as well as the
work of demystifying the research process.
Doing Academic
Research
A Practical Guide to
Research Methods and
Analysis
Ted Gournelos
Joshua R. Hammonds
Maridath A. Wilson
First published 2019
by Routledge
2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN
and by Routledge
52 Vanderbilt Avenue, New York, NY 10017
Routledge is an imprint of the Taylor & Francis Group, an informa business
© 2019 Ted Gournelos, Joshua R. Hammonds, and Maridath A. Wilson
The right of Ted Gournelos, Joshua R. Hammonds, and Maridath A.
Wilson to be identified as authors of this work has been asserted by them
in accordance with sections 77 and 78 of the Copyright, Designs, and
Patents Act 1988.
All rights reserved. No part of this book may be reprinted or reproduced
or utilized in any form or by any electronic, mechanical, or other means,
now known or hereafter invented, including photocopying and recording,
or in any information storage or retrieval system, without permission in
writing from the publishers.
Trademark notice: Product or corporate names may be trademarks or
registered trademarks, and are used only for identification and explanation
without intent to infringe.
British Library Cataloguing-in-Publication Data
A catalogue record for this book is available from the British Library
Library of Congress Cataloging-in-Publication Data
A catalog record for this book has been requested
ISBN: 978-0-367-20791-5 (hbk)
ISBN: 978-0-367-20793-9 (pbk)
ISBN: 978-0-429-26355-2 (ebk)
Typeset in Galliard
by Apex CoVantage, LLC
Visit the eResources: https://www.routledge.com/9780367207939
Contents
List of figuresvii
Preface: two professors and a librarian
walk into a bar . . . viii
Part I
Get the party started right 1
1 Learn to love, not dread, research 2
2 Where do I start? 14
3 Now what do I do?

What’s a lit review? 26
4 Choosing a method 42
Part II
Methodology overviews 71
5 Close textual and thematic analysis 72
6 Content and discourse analysis 81

vi
Contents
7 Ethnography, interviews, and

focus groups 98
8 Surveys 119
9 Observational methods in practice

(participatory-action research,
experimental research, theory,
and metasynthesis) 135
Part III
Data analysis and the final push 145
10 Analyzing qualitative data and

making sense of detail 146
11 Analyzing quantitative data and

making sense of numbers 156
12 Persuasion, presentation, and publication 173
Glossary181
Index185
Figures
1.1 From topic to paper 6

1.2 Choosing a method 9
3.1 From topic to literature review
(samples from popular culture and society) 33
4.1 Comparing methodologies 44
4.2 IRL: How to do what your boss asks 67
7.1 Types of questions 108
7.2 Primary and secondary questions 110
8.1 Populations vs. samples 121
8.2 Categorical question 125
8.3 Likert-type scale questions 126
8.4 Semantic differential question 127
8.5 Validity and reliability 128
8.6 Bad questions 131
11.1 A “normal distribution” 159
11.2 Negatively skewed distribution curve 161
11.3 Positively skewed distribution curve 161
11.4 Tests and variables 165
11.5 A quick guide to statistics terms 171
Preface
Two professors and a librarian walk

into a bar . . .
Research doesn’t usually start in an office, or as an assignment, or

while reading some giant textbook. It often starts at the moment
you see the need for something, or you take a step back and wonder
why something happens the way it does. In other words, research
is all about the world we live in and how we understand that world.
In our case, this book began, as many books have before it, in a bar.
Happy hours are a long tradition in work cultures, and univer-
sities are no exception. In this case, Maridath, Josh, and Ted got
together after work to have a couple beers and decompress after
a long week. While a good happy hour doesn’t dwell on work for
very long, it usually starts with a bit of complaining. Ted ordered
a pint of some thick stout that looked like an unholy mixture of
bread and coffee with some alcohol in it, and told the others how
frustrated he was that students do so little research on campus.
“They always look at me like I’m crazy when I assign a real
research project where they actually gather data. What else would
we be doing?”
Josh took a sip of his hipster-approved, hop-forward IPA
before replying. “Even in research methods classes they aren’t
ix
Preface
prepared for it, and they always end up taking that at the last
possible moment because it terrifies them.”
“OK, but that’s a problem, right? They’ll be doing this for the
rest of their lives, or at least they will if they want to get a decent
job. That’s what being a good manager means: research.”
Maridath, the librarian that works with Ted and Josh to intro-
duce students to research methods and how to find sources,
walked in predictably late and ordered something from Brook-
lyn to match her artfully distressed leather jacket. “They always
come to me for help for exactly that reason, but I don’t know
what to tell them. It’s one thing to learn about research. It’s
another thing to learn how to actually do research.”
They chatted for a while and realized that it wasn’t just their stu-
dents. It was them, too. Even graduate students have often never
learned how to conduct various methods and why you’d choose
one over the other for any given question or context. Students,
graduate students, and even faculty often just use the same meth-
ods over and over again; sometimes they even avoid reading work
applying different methods. It was a reality check, as Maridath
reminded them. “We can’t blame our students, right? It’s not like
there’s a book out there that does this, either. They’re all huge,
dense and often boring, and still don’t give you a practical guide.”
They all sat for a little while, and then Ted looked up and
said, “Um, well, why don’t we just write it?”
*****
That all actually happened, more or less. We probably sounded a
lot dumber, and probably added in a lot more complaining. But
that’s why memory is so great; you can avoid thinking about
unpleasant truths. This book is obviously that book, born in a
bar and written in offices, coffee shops, libraries, at home, and,
again of course, the occasional bar.
What we’re going to do here is take you through a few steps.
Part 1 (Chapters 1 through 4) is about understanding what
x
Preface
research is, how to choose a topic, and how to get the back-
ground so that you actually know what you’re talking about and
can write a “literature review.” Part 2 (Chapters 5 through 9)
takes you through a few of the main methods that you’ll encoun-
ter and shows you what they do and why you’d choose them.
Part 3 (Chapters 10 through 12) shows you how to actually
analyze the data you gathered through your method, and then
how to put all of that together into a finished product. While
this book is suitable for anyone doing academic research, it’s
intended mainly for undergraduate and graduate students. Basi-
cally, it’s what we wish we’d read when we were starting out.
We’ve added in a couple of things to make this even more
useful as a practical guide. We offer some tips and tricks that we
use to make things manageable, and we hope those will help you
along the process. We also offer “tips from your friendly neigh-
borhood librarian,” in which Maridath takes you through how
librarians can help you along each stage of your research project.
Chapters also feature “IRL” (in real life) sections that show you
how each part of the process might appear in the workplace, in
fields ranging from marketing and restaurant management to
human resources and customer experience monitoring. Finally,
we supply a set of online resources for each chapter. We’ve
included articles from many different disciplines that use the
methods described in the book, videos that guide you on how
to use technology that will make your life a lot easier, and links
to citation guides, tutorials, and even paid services that can do
everything from transcribing your data to keeping your sources
organized to arranging for people to take your survey.
We hope you enjoy the book and that it makes the whole
process a lot less intimidating. If you’re not interested in your
research project, and if it’s not at least a little fun, then you’re
doing it wrong. This isn’t something you should dread; it’s a
rare opportunity to figure things out, to find out what makes
you you, and what makes the world what it is. Run with it.
PART
Get the party
started right I
Learn to love,
CHAPTER
1 not dread,
research
OVERVIEW AND OBJECTIVES
This chapter goes through what research actually is

(answering interesting questions), rather than what we
usually feel it is (agonizing work). We’ll discuss how
to think about your research and your topic, as well as
how to find research questions and choose methods.
We’ll expand on topics and research questions in the
next chapter, but this is the general overview of how to
do it without panicking. Using examples of sample top-
ics drawn from a recent class, along with their possible
research questions and possible methods, we give you
a brief overview of the various kinds of methods you
could choose from and that are highlighted in depth in
later chapters.
1: Learn to love, not dread, research 3
Starting a research project always sounds a lot harder than it actu-

ally is. Most people would rather hide under their desks (or beds,
the choice is yours) and hope it goes away than begin a new proj-
ect. The thought of starting something big, whether a report for
your boss, a term paper in a class, or a new book, is often terrify-
ing. We understand that, and we want you to understand why it’s
so scary. Research is scary because there’s a whole big wide world
out there, and for every piece of that world there are dozens, if not
hundreds, of ways to study it. What’s going to work? What will
people understand? What won’t drive me crazy to study? What’s
interesting about this project, or (scarier) what’s new about this
project? We’re getting nervous just thinking about it.
Calm down. That’s step one. The good news is that while
research is hard, it’s also a lot easier to start than you might
think. Because there is so much to study, and in so many differ-
ent ways, you should be able to find something that interests you,
that’s possible to accomplish, and that isn’t completely repeating
someone else’s work. That might not seem very reassuring, but it
should be. For instance, when Ted began his dissertation for his
PhD (not much is scarier than that, we can assure you), he went
to a professor with a twelve-chapter book outline that he was
really excited about. It studied twelve different media examples
(from graffiti to internet memes to The Simpsons). The professor
took one look at him, shook his head, and said “you’re insane.”
It was true. The project would have taken years and been thou-
sands of pages long if it was done right. So Ted went home, really
thought about it, and decided to choose one chapter, the one on
South Park, to be the focus for his dissertation and then his first
book. This is the question we want you to ask yourself: What can
I study that I don’t think will bore me and my readers, or make
me hate it? Never start a project because you think it’ll be easy; if
you do it right, it won’t be. Start a project because you’re fasci-
nated by it, and then figure out how to make it possible to do in
whatever timeframe you have been given.
4 Part I: Get the party started right
IRL: A BOSS THAT WANTS THE WORLD
It’s Thursday afternoon and your boss has just come back
from a meeting with her boss. She asks to chat with you
for a second, which is always scary, and tells you she needs
a comprehensive report on how your company is doing
around the world. And she needs it tomorrow. At this
point you’re more than nervous; you’re worried about
keeping your job if you can’t complete an impossible task.
So here’s what you do (or what one of us did, since this is a
real story): First, ask what your boss really needs to know.
In other words, if she just wants numbers that you can
find on a company database, that’s not so bad. If she wants
those numbers along with detailed information about cus-
tomer satisfaction, you’ve got a big problem. Second, ask
who she needs the research for, and why they want it. If
the CEO just wants to be able to show the Board of Direc-
tors that you’ve got stores in 20 countries as one tiny part
of a presentation, no problem. If they’re evaluating where
to invest in new marketing campaigns and where to hire
staff to support an international branch, it’s going to be a
long, long night and you still won’t finish. Remember that
research is a process, and there is always room for more
depth or more reach. You have to know what questions
you need to answer, and why, before you can do it well. Be
careful, because you can get in over your head really fast.
Time is an important factor in research because, in theory, all

research projects could go on forever. Every topic is like a huge
room that you can’t see into, and each research project just adds
a little peephole in the wall that you can look through. Maybe
eventually you’ll be able to see everything, but it’s going to take
you a while, and you’ll have to shift your perspective, depth of
focus, and sight angle to see much. And that’s if nothing in the
room changes. Good luck with that. People, media, minds, cul-
ture, politics – everything is changing all the time. The trick is to
get a perspective on what’s happening in a way that meets your
goals as well as your timeline. Sometimes that’s a matter of how
much you study, and sometimes it’s a matter of how deeply you
study it. But time, like money, is never infinite.
Once you’ve chosen a general topic, you need to figure out
how you want to do the study (see Figure 1.1). Researchers
often split our work into three categories or perspectives, or
what we call “paradigms”: the Empirical, the Interpretive, and
the Critical. Empirical scholars tend to believe that the topic
they’re studying is discoverable, and often through measurable
data with numbers. They want certainty, and so they tend to
focus on large amounts of data that they can extract and analyze
in a very methodical way that leaves little room for variation. In
other words, they’re looking for a snapshot through that keyhole
without drawing conclusions about anything they can’t see. For
example, in a project Josh conducted on healthcare, he wanted
to know how often patients did what their doctors told them
to do. He decided to study whether or not the amount of time
doctors spent with patients (the independent variable, or what
we use to explain changes) predicted how satisfied the patient
was with the experience (the dependent variable, or what actu-
ally changes). There are several ways to tackle how time affects
patient satisfaction, but for his study, he decided that the length
of time and degree of satisfaction told the best story.
Interpretive researchers, on the other hand, look at their
research slightly differently. They’re not looking for a snapshot,
they’re looking for what’s moving. More specifically, they’re
looking at how things might be portrayed, received, understood,
Topic 1: Topic 2: Quality Topic 3: Fighting
Cognitive healthcare hunger in the USA
disabilities in
the media
Focus: Focus: Doctor/ Focus: Nonprofit
Portrayal of patient organizations’ use
autism and interaction of social media
dementia in the (interpretive to communicate
media (critical paradigm) (critical/empirical
paradigm) paradigm)
RQ1: How do RQ1: How RQ1: Who do
plays portray/ do doctors nonprofits tend to
include people communicate target in their social
with different with patients media, and how?
ways of about health How often? With
thinking? issues? what response from
viewers?
RQ2: How do RQ2: How do RQ2: What stories
portrayals patients connect do nonprofits tell,
of autism in their interaction and what visuals do
children and with doctors to they use to tell those
dementia in the their perception stories? How often?
elderly differ? of the quality of With what response
care? from viewers?
RQ3: What RQ3: What might RQ3: How might
kinds of be improved approaches by
problems in setting various nonprofits
and solutions expectations differ, and how
do the plays and establishing do they align with
focus on, and effective ways of industry “best
who causes/ communicating practices” for
solves those during effective website
problems? interactions? development?
Figure 1.1 From topic to paper

Method: Textual Method: Method: Content

analysis of two Observations analysis of
scenes from of twenty twenty websites,
two plays (first interactions and looking at specific
introduction post-observation categories of
to a character interviews with visuals, storytelling,
with cognitive twenty patients structure, and
disability, other types of
climax of content drawn
conflict within from industry and
the play) scholarly sources
on effective
websites
Figure 1.1 Continued
and reacted to by the people in that room. For example, you

might study how an organization’s staff decides which content
to put on their website or social media, or how various groups
of people feel about that content. Interpretive scholars are often
more concerned about how we think and feel, the unique fac-
tors of what drives our behaviors and reaction, than they are
about measuring finite patterns and outcomes. For them, the
world is full of complex concepts, variables, and variations, and
numbers just can’t represent what’s really happening.
Critical scholars simply add a layer to empirical and interpretive
research by asking why things are happening in relation to the
world more broadly. They’re concerned with relationships involv-
ing power, ethics, and institutions (including their own power
and influence), and are constantly questioning even their own
analyses. They wonder why the room is closed off, who is trapped
or allowed to go inside, and what it means, not just to them but
also to society (or other groups, or the planet, etc.) as a whole.
Choosing a paradigm is important because it allows you

to start forming some research questions (RQs). This is use-
ful, even if your questions start off vague. What do you want
to know? What questions can be answered by other schol-
ars, and what questions do you specifically need to be able to
answer through your research? Part of that means choosing a
method, of course, and that method is in part dependent on
your paradigm. For instance, most people start hearing about
the “scientific method” in elementary school; you Observe,
Question, Hypothesize, Experiment, Analyze, and Conclude.
This is no different, except we suggest you keep hypotheses
to a minimum; focus on the questions. What’s important is
how you’re going to do that experiment (method), and this is
completely dependent on your resources (time, money, and
people to help), perspective (paradigm), and questions. Fig-
ure 1.2 gives you an overview of various kinds of methods, and
the questions they might try to answer. We’re going to talk
you through many of those throughout this book, so don’t
freak out.
We suggest that before you choose a method, start with some
questions. What do you want to know? If you want to know how
something is portrayed, discussed in the media, and/or made part
of an organization or institution, you might want to choose a
text-centered method. You would analyze the message, artifacts,
and context that surround them. If you want to see how people
respond to something, are influenced by something, or discuss
something among themselves, you might want to go with a
people-centered method. If you want to determine reactions that
people might have to a specific stimulus (specific enough that it
can be both repeated and contained in a controlled environment),
then you’re probably going for an experiment-centered method.
Once you know which type of method you’re probably going
to use, you can think about number and depth. Because resources
are limited, you need to go back to the research questions. What
do you want to know? If you need to see how a large number of
• Textual analysis (in-depth reading of one or a few
texts, with emphasis on their physical structure, e.g.,
shooting and editing for film, layout for design, rhet-
oric for a speech)
o Number: low
o Depth: high
o Sample: intentional
• Content analysis (brief reading of a large number of

the same sorts of texts, usually with the aid of explicit
“coding,” in which commonalities and differences are
examined across all texts and charted, usually in part
for statistical significance)
o Number: medium
o Depth: medium
o Sample: intentional but systematic; coding is as
comprehensive as possible
• Discourse analysis (brief reading of a large number

of usually different kinds of texts in order to discuss
the ideological landscape surrounding a central topic,
including journalism, popular media, educational sys-
tems, state institutions, family systems, etc.)
o Number: high
o Depth: low
o Sample: Intentional but wide ranging, surrounding
a topic/focus
• In-depth interviews (structured, semi-structured, or

unstructured, interviews are often done with fewer
participants but at a high level of depth, usually from
Figure 1.2 Choosing a method

1–2 hours in length, and try to get participants to
tell stories and explore concepts in detail, often with
follow-up questions or examples)
o Number: low
o Depth: high
• Focus groups (small groups of individuals that are tar-

geted to gain specific information about a topic; usu-
ally random groups or groups that are intended to get
a range of opinions, focus groups are widely used in
industry studies of media, like whether a new TV show
will be popular, or in social sciences to gain perspec-
tives on things like body image, etc.; focus groups are
great because you can use a semi-structured interview
guide but encourage conversation among individuals,
which you record and then analyze; however, they can
also result in “groupthink” or silenced voices)
o Number: medium
o Depth: medium
• Surveys (a questionnaire given to a sample of individ-

uals to gain their perspective on a topic)
o Number: high
o Depth: low
o Sample: varies from “convenience” to “snowball”
to “random”
• Ethnography (observation and investigation of the

discourses, events, rituals, and communications within

a community or a (set of) individual(s) within that

community; although auto-ethnography is absolutely
valid as a method, it must be chosen with great care
and attention to the method as appropriate for the
sort of impact and evidence necessary for the project)
o Number: low
o Depth: high
o Sample: intentional but wide ranging to encom-
pass many variables
• Metasynthesis (brings together a bunch of scholarly

sources to examine their results all together, almost
like a qualitative content analysis of scholarship)
o Number: high
o Depth: low
o Sample: impartial and wide ranging; comprehen-
sive if possible
• Experiment (within a controlled environment, manip-

ulating variables in one group in order to determine
changes)
o Number: low
o Depth: high
o Sample: intentional with a control group
people think about or react to something, you need a high N (the

population size being analyzed). You’ll probably need a survey in
that case. If you are more concerned with the nuances of how
people think or feel about something, you will need them to tell
stories or discuss the topic in detail with you. In that case you
might want to do interviews with a high D (depth) that could take

several hours each. If you’re shooting for something in the middle,
maybe how a group of people will respond to a new advertisement
or feel about their romantic partner one year after getting mar-
ried, then focus groups are better bets. The same thing goes for
text-based and experiment-based methods; in general, the higher
the N, the lower the D (see Figure 1.2). That doesn’t mean that
high-N methods don’t have depth. It just means that with limited
resources you’ll have to focus on one over the other.
Be careful, when choosing the sample that you’re going to
study (which texts, which people, which places, etc.), to avoid
what we call “cherry-picking.” Cherry-picking means limiting
your sample to the sources that will support your argument,
which can be avoided by focusing on questions rather than
answers. For instance, one of Ted’s students had a problem
last semester with studying body image on Instagram. A good
topic, certainly, but when she tried to study it through a content
analysis, she pulled images from celebrities and focused on their
“sexy” photos. What if some or all of those celebrities posted
just a few sexy photos, but more often posted pictures of nor-
mal life: without makeup and in ugly pajamas, pictures of their
pets and kids, and random pictures of food? Her research would
be skewed because she just chose primary sources that suited
what she thought were going to be the answers to her questions,
rather than letting the data speak for itself.
The thing to remember is that you don’t have to do any of this
alone. Other scholars have struggled with finding a topic, creat-
ing research questions, and choosing methods since . . . well, since
people started doing research. And because analyzing the world
around us is fundamental to the human experience, if not life itself,
that’s a long, long time. Reading other scholars is one way to keep
yourself from getting overwhelmed. But so is talking to people,
including your friends, your family, your professor, your colleagues,
your boss, and of course, your friendly neighborhood librarian.
A NOTE FROM YOUR FRIENDLY

NEIGHBORHOOD LIBRARIAN:
Librarians know that your first (or tenth!) research project

can be a really daunting process. Every semester, Maridath
works with dozens of students who are overwhelmed by
even beginning their research assignments. Often, your
professor will spend a great deal of time explaining the
requirements, but they often forget how hard it is to come
up with a good topic. We can help! Unlike the boss that
wants to know everything about “how satisfied the cus-
tomers are,” we want to help you both with the broad
questions and with how to do the research without driv-
ing yourself crazy. We take you through the process step
by step, from getting to know you and your interests, to
creating and focusing a topic and the research questions
suitable to your project, to actually finding the resources
to make sure you’re successful. We’re great at helping you
find sources for your research assignments, but we really
enjoy the process of helping you find a topic that’s inter-
esting to you so you won’t get bored and hate us and your
professor (or boss) later. You may not realize that we also
work closely with your faculty and often have the inside
scoop on what they’re looking for. It’s important to also
come to your session with your librarian with any work
you’ve already completed on the assignment, including
preliminary sources, ideas around your topic and focus,
and, of course, the assignment prompt itself.
Where do
CHAPTER
2 I start?
This chapter teaches you how to effectively and efficiently

retrieve and use scholarly sources for your project. This
process allows you to draw on existing research, making
your job a lot easier. It also avails you of the support of
those who have been working on this stuff for a long time,
often in ways that you aren’t going to work on it. It always
helps to have multiple perspectives. We give you an idea of
what a good source is, but also what a bad source might
be, how to avoid them, and some tips and tricks for how
to find good sources without driving yourself nuts. Finally,
we talk about how to create a good annotated bibliography
to keep yourself organized and on the right track.
2: Where do I start? 15
Now that you have a topic that you don’t hate and that you can
feasibly accomplish, when do you get to the actual really real
research stuff? To be honest, you’ve already started. Choosing
the topic, asking the questions, and finding people to help you
are part of the research. In fact, they are the most important
parts. Everything else is comparatively easy, even if it’s hard work
(yes, we know that sounds weird). Your librarian in particular
can help you here, because librarians are excellent at helping you
find stuff out, no matter what “stuff” means. They don’t do that
through a research method the way we’re describing it; instead,
they help you by giving you a step up. To use a common saying
in English, they help you “stand on the shoulders of giants.”
So what does that mean, realistically? It means librarians help
you find other people who have tried to answer similar questions,
or find ways in which other people have studied similar topics,
so that you don’t try to do everything (or that you don’t waste
time doing something that someone else has already done). The
trick is that you need to be able to keep this process methodical.
If you just start randomly choosing articles or books that have
to do with your topic, you can run into some major problems.
Problem 1: Bad sources
The main reason you might find “bad” sources is if they don’t
pass the “CRAP” test for a source’s “currency” (how recent it
is), “relevance” (how related to your topic and its particular dis-
cipline), “authority” (how credible the publisher or writer is,
and if they have authority in that particular field), and “purpose”
(are they just trying to persuade you of their opinion, or do
they want to inform you?). In general, for secondary sources
you’ll want to stick with scholarly, peer-reviewed journals. In
some cases, like with the project on nonprofit organizations, it
can also be important to get sources from established “trade” or
“industry” publications. These might give you current perspec-

tives on what works or doesn’t work in an industry or practice
(that could be website design, but it could also be doctor-patient
interactions or adapting a novel to the theater). Even major
newspapers like the New York Times or the Washington Post (if
they aren’t editorials) can be useful, but use them sparingly.

What’s peer review and why should

I care?
So who are these “peers?” And what exactly are they reviewing?
Librarians love to talk about “scholarship as conversation” –
seriously, get one going on this for a minute and see how
excited they get. Articles that appear in peer-reviewed aca-
demic journals are written by experts for experts. After per-
forming their research, scholars submit their work to these
journals, which then undergo a review process. Their work
is checked for style, substance, and quality by other schol-
ars in the field, usually in a double blind method where no
one knows the identity of anyone else. Articles and books
are then either accepted for publication in the journal,
accepted with revisions (where the scholar must make
often substantial changes and then re-submit), or outright
rejected. Usually, from start to finish, this process can take
a long time – sometimes years. It’s important to note the
length of this process, because this notion of “currency”
(the “C” in the CRAP test above) really differs between
scholarly journals and newspapers; it also depends on dis-

cipline. If you are researching something that happened in
the news last year, there might not be much in the scholarly
literature on it. What you will find is an ongoing discus-
sion of sorts – scholars citing and commenting through time
around the overarching themes and theories of their par-
ticular discipline – thus, the scholarly conversation on, say,
the news industry will be more likely than something on a
specific current event. Newspaper articles also go through
review, but usually by an editorial staff (if that), and rarely
by people who know anything about the content or broader
facts/context. Journalism is in many ways “normal” people
speaking to other “normal” people in an easily understand-
able way, and they often interview scholars for the context
they need. Ted, for instance, goes on the news often to dis-
cuss culture, politics, social media, and communication in
general (his first appearance was for a spot on online dat-
ing, which he’s still embarrassed about). It’s important to
note that scholarly sources are not necessarily “better” or
“worse” than newspaper articles or more popular texts, but
they serve different purposes and wildly different audiences.
Problem 2: Out-of-date or irrelevant sources
While this depends on what your project is and who has assigned
it (your boss, your professor, yourself), usually you want to make
sure your sources are current and useful. While not necessarily
“bad,” you might find that out-of-date sources are easier to find
because they are “kind of” about your topic, but not really useful
in the end. Sources that are “kind of” about your topic probably
don’t pass the “R” (relevance) part of the CRAP test. In some
cases, older sources may be appropriate (e.g., if you need to trace
the origin of a problem or track its historical trajectory). That
said, it’s perfectly fine to consult sources that inform you about
the topic more generally. Just be careful to keep them to a mini-
mum. For Ted’s project on disability and theater, for instance, it
was useful to look at sources on mental disabilities in the media in
general, and ageing in the media in general. However, looking at
medical studies on autism or on Alzheimer’s probably wouldn’t
be useful for a paper on how it’s portrayed in the theater, or how
audiences react to those portrayals. Similarly, be sensitive to how
old a source is. Old doesn’t mean bad, but it does mean that the
author(s) weren’t able to react to conversations that happened
since. Remember, scholarship is a constantly evolving discussion.

But are Google and Wikipedia BAD?

I thought they were BAD?
Many researchers choose to begin their research on Goo-

gle Scholar (not Google!), or even Wikipedia. We like to
do this because Google Scholar allows us to do a couple
of unique things: see the impact factor or how many times
a particular journal article is cited by other researchers,
and quickly find related articles on a particular subject.
Often, Google Scholar is indexed in your library’s main
catalog via complex metadata sharing which will allow
you to follow the link to the full-text PDF of the arti-

cle. A word of caution, however: you can usually only
do this from a computer on campus. You should never,
under any circumstances, pay the fee for a source you
find on Google Scholar. What you should do (especially
when off-campus), is take the title and/or author’s name
and plug it into your library’s main search bar and see if
they provide access. If they don’t, you can often request
that the source be sent to you by “Interlibrary Loan,” or
ILL, which often means a librarian from another univer-
sity will scan the article and email it to you, or even send
you a book in the mail. As you progress further into your
research and your topic narrows, it’s a good idea to look
through the more discipline-specific databases to which
your library provides access, but starting either at Google
Scholar or the library search engine is the best place to
begin. Google Scholar also has a wonderful set of tools
to narrow down your search, including “related sources”
and “cited by,” which will allow you to find useful mate-
rial that you might not see in a regular search. Similarly,
Wikipedia allows you to get a sense of what people have
said about a topic’s history, background on associated
content (e.g., the filmography of a director whose film
you’re analyzing, or the biography of a historical figure
whose speech you’re comparing to national policy), and
even news press or scholarly sources. Wikipedia can also
be useful for giving you a primer on the terminology
associated with your particular topic, which can aid you
in crafting quality search terms. Like with other source
material, make sure you look for citations and give those
the CRAP test!
Problem 3: Cherry-picking
“Cherry-picking” is a pitfall of one’s method and use of primary

sources, as we mentioned above. It’s also important to avoid in
your literature review. In the secondary sources for your anno-
tated bibliography and literature review, cherry-picking means
that you’re just choosing sources that agree with you. Don’t
do that. The more diversity you find, the better. The “litera-
ture” isn’t monolithic. Instead it’s a series of people just like you
(maybe with a few more letters after their names) that are ask-
ing similar questions, and answering them in many ways. Often,
that means they disagree, and that disagreement is important for
your research.
One of the best ways to keep these problems under control is
to make sure that you’re searching for things strategically; but
you should also make sure to constantly critique the sources.
Keep in mind the notion of authority as well. Ted has exper-
tise in several fields, but medicine isn’t one of them. Medical
journals on autism weren’t exactly the best for his research on
autism and theater, and he shouldn’t make medical claims when
he writes. In other words, don’t get overwhelmed by the fact
that an author has a bunch of degrees; you should still be criti-
cal of their work. Similarly, you should keep track of how you’re
searching and what you’ve found, even if it’s irrelevant stuff.
After all, if you have to come back to this topic you’ll want to
know what not to look at, and if you’re handing the project to
someone else (another student, an assistant, a professor, a boss,
etc.) you want to make sure they don’t have to do all that work
over again either.
IRL: RESEARCH CAN GET YOU

PROMOTED, BUT MAKE SURE
YOU DO IT RIGHT
One of the most common pieces of feedback we get from

students is how surprised they were when the research
they did in our classes is the skill they use most often in
work (and what often gets them better jobs). It doesn’t
matter if we taught them to analyze scenes from a film,
conduct surveys, or interview their peers; the process of
choosing and conducting high quality research is market-
able and translates across industries. Our students conduct
focus groups and interviews with employees and custom-
ers to figure out how to make their experiences with a
company better. They create surveys to measure how
professionals work most effectively, how doctors perceive
their patients or their pharmaceutical representatives, and
how donors and volunteers relate to a nonprofit. They
compare the social media used by competitor brands in
content analyses, and examine that along with analytics
to see what works to achieve the company’s goals and
what doesn’t. They do textual analyses of their company’s
advertisements or press releases to avoid public relations
crises or unintended messages. But what connects all of
these is that they have to do it right. Making mistakes like
cherry-picking in a sample, ignoring key research, or using
bad sources might not make you fail a class, but it can cost
your company millions of dollars. If you do it right, on the
other hand, the sky’s the limit.
Keeping a consistently formatted bibliography of sources will

help you a lot with the project. People complain about this all
the time, including us. It always feels like it slows down the
process, because when you start your research you want to go
quickly, and nothing feels slower than formatting sources. How-
ever, trust us that by the end you’ll be glad you did. Every min-
ute you spend here might save you two or three (or more) later
down the line, usually when you’re pressed for time.
We create bibliographies with consistent citation styles such as
APA, MLA, and Chicago so that we can organize our sources con-
sistently and methodically. Why does this matter? First, you save
time by ensuring that you and your readers (e.g., your professor)
can quickly and easily find the sources you are referencing, or spe-
cific pieces of the source like the journal or the year it was published.
You also prevent plagiarism or a dreaded visit to your school’s
Honor Council, which could ruin anyone’s day. But the main issue
is that this is useful as you gather more and more sources, and
especially if you’re doing this for work, when they might be also
used by your boss or a team. For most books, for instance, you
consult not dozens, but hundreds of scholarly sources. Without
keeping a bibliography, you not only risk forgetting the sources
but also leave out key pieces of information and background.

Citation exporters
There are many ways to cite sources, but they’re all system-
atic. You’ll most likely encounter the styles from the Asso-
ciated Press, the American Psychological Association, the
Chicago Manual of Style, and the Modern Language Associ-

ation. You’re probably aware of a few tools that allow you to
select a citation style and click a button to “export” the cita-
tion into a word document. These have names like EasyB-
ibb, EndNote, Zotero, or NoodleTools, some of which are
free and some not. Zotero is one of Ted’s favorite tools, not
because it formats things correctly, but because you can add
it to Microsoft Word and to browsers like Google Chrome.
It lets you grab citations from articles and even websites
and newspaper articles quickly as you find them, and then
go back to download or read them later. There is often a
citation exporter button in the search results page on your
library’s website, too. Zotero is free and usually works the
best, so it’s the reference management tool we recommend.
Any of these tools will get you most of the way there, mean-
ing they’ll help you capture the elements that are required
in a properly formatted citation. But the reality is that they
often make mistakes. Use them, but fix what they produce!
Many sources, like the official citation guides from each of
the major style manuals, make this an easy process (espe-
cially for formats like Chicago that have two different ver-
sions). One other suggestion is to make sure you keep things
organized, and on the “cloud.” Using OneDrive, Google
Drive, or iCloud, keep your sources (like PDFs with high-
lights) and every element from your paper stored off your
computer, so you don’t risk losing all your work. Trust us.
We’re all still traumatized by losing huge amounts of work
when computers crashed or we forgot to save a file before
the cloud was there to save us from working after too little
sleep and too much coffee.
Annotated bibliographies
You also want to make sure that you don’t choose some sources,
read them, and then forget what you read and have to read it all
over again in a week (or a month, or a year). The best way to
prevent that is to create an “annotated bibliography,” which is
a bibliography to which you’ve added notes about each source
(including whether it was useful or a waste of time to read). What
we recommend is keeping this organized for yourself, either with
bullets under each source or in a short paragraph. While there are
various way to do this, there are a few things you want to make
sure to hit, some of which we discussed in Figures 1.1 and 1.2:
1. Topic (what’s their general concept?): Don’t try to get too

specific here. Think general search terms for Google Scholar
like “gender and advertising” or “disability and media,” or
even “doctor patient communication.”
2. Focus (what specifically are they looking at?): This is where
you will want to be more specific about what you’re studying.
Is it a set of places (hospital waiting rooms, elementary school
architecture), texts (poems, films, speeches), people (patients,
millennials, students), concepts (deception, humor)? How
specific you want to get depends on how far along you are in
the project. Eventually “humor” probably won’t be enough.
At some point, for instance, you’d study how college stu-
dents respond to humor in a specific episode of The Office
regarding the intersections of class, race, and gender.
3. Method (how did they study it?): This varies, but you’ll
want to specify sample (and sampling technique) as well as
the general method (see Figure 1.1).
4. Results (what did they find out?): Connecting your method to
your focus, how did the study answer its research questions?
Be aware that the questions might not be obviously laid out,
even in a thesis statement. You might have to look for them.
5. Implications (why does it matter?): Implications are both

your judgment about the source’s relevance to your proj-
ect and also how the author(s) argue the study is import-
ant. Often this is in a “discussion” section, but usually
it’s towards the end. Limitations are also a good thing to
include here: what should happen next? Where did the
study fall short?
For example, in Ted’s paper on disability in theater (topic), he

examined the portrayal of cognitive impairments in two plays
(focus) through a textual analysis of two scenes each (method).
He found that the plays emphasized difference through a mix-
ture of disconnection from society and empathy with the cog-
nitively “different” characters that encouraged viewers to be
critical not of the impaired, but rather of the “normal” characters
and an inaccessible society (results). He suggested that media is
thus a potentially effective way to broaden social recognition of
a cognitively diverse society, reducing stigma and encouraging
empathy even if its portrayals are fictionalized (implications).
Once you have your sources laid out, categorize them and
ask yourself two questions: First, have you addressed all of the
aspects of your topic that are necessary for your project, and
in enough detail (with enough sources)? Second, what sources
did the studies that you read cite that you might be able to use?
Once you’ve got that figured out, you’re ready to start your lit-
erature review. This is also why you want to spend as much time
as you can on the annotated bibliography; if you start with the
literature review, you might end up spending a lot of time on
sources you don’t really need, or you might miss out on more
helpful sources. More importantly, it’s harder to categorize and
see themes when you’re immersed in the sources. The annota-
tions help you take a step back and see the big picture.
Now what do
CHAPTER
3 I do? What’s a
lit review?
This chapter takes you through the process of making

a disconnected set of sources into a coherent, persua-
sive “literature review.” Basically, that means you’re
going to tell the story of your project, through the
scholars who have written on the same or similar top-
ics before you. We provide some suggestions for how
to find categories of literature to look for before writ-
ing the literature review, which involves drawing from
and expanding on your annotated bibliography. This is
sometimes a tricky process, so we’ll take you through
how each of us approaches it in detail to give you some
guidance on process.
3: Now what do I do? What’s a lit review? 27
Once your annotated bibliography is done, it might seem like

you’re ready to dive into your paper. In many ways, you’re
right! This is where the project truly begins to take shape, and
it’s important that this happens before you ever start to do your
analysis, or even before you choose a method. That might seem
strange. After all, what is research if you aren’t actually doing
anything? Why aren’t we trying to answer the questions we spent
all that effort framing? The answer is simple, really. We don’t
know enough yet. We might have an idea of what we want to
talk about, and we might have thought about it a lot. We might
have even read a bunch of scholars talking about our topic, and
so we know its history and how people have approached it over
the years. But if you start writing now, you’re going to run into
problems. Why? Because the first part of writing the paper is
making sure you know what’s going on, categorizing it for your-
self and, more importantly, contextualizing it for your readers.
It’s important that they know that you know what you’re doing.
IRL: WHY WOULD I EVER HAVE TO DO

THIS AFTER I GRADUATE?
It might seem like a literature review is one of those things

you just do because your professor makes you. You’re not
wrong. But that doesn’t mean you won’t have to do it in
some way after you graduate, or that it’s not a useful skill
that will be valued by your company (and might even get
you promoted). The reality is that while you’ll rarely do a
traditional literature review in a job, you’ll almost definitely
have to synthesize complex information from a bunch
of sources that might not be directly about your project,
compare various authorities, and insert yourself into a con-

versation that you aren’t yet a part of. That’s really what
it means to be a manager. Knowing how to do a literature
review means you know how to see what you’re doing as
an extension of what’s already been done. It means you can
look at what competitors and people that disagree with your
perspective are saying, and take what’s valuable from them
before respectfully suggesting that there’s another solution.
Imagine your company is merging with another company,
for instance. What are the different ways that’s been done?
What causes employees to stay happy and become hopeful
rather than fear the change? What causes shareholders to
increase the stock price, customers to keep buying the prod-
ucts your company creates, and the research teams to work
together? There are differing views on all of these things,
and each is a different research question. Knowing how peo-
ple have approached the problem, as well as what worked
and why, will help you suggest and provide a solution that is
grounded in research and has some backing when colleagues
or bosses are skeptical or upset. For instance, Ted once told
a client who was worried about low profits that their biggest
problem wasn’t their employees, who were blamed because
they weren’t looking for new business. Instead, he said it was
the bosses that had the problem, and every employee that
left not only cost them money, but scared away potentially
great new hires. They didn’t want to believe it, but when
they saw the data and the literature on the topic, they started
to rethink how they approached all sorts of things, from
meetings to salaries to benefits. It worked; but he couldn’t
have done it without the research.
This is where the literature review (or “lit review”) comes

in. A literature review is pretty much what it sounds like: it’s a
review of the literature on a topic. But this isn’t literature like
“English literature.” It’s the scholarly literature, the conversa-
tion around your topic in its various forms. How deep you want
to go here depends on your topic. But there are a couple of
things that can guide you along the way other than a professor
saying “you need ten sources” or something along those lines.
We do that to our students, don’t get us wrong. However, we
don’t care so much about the number as much as we do about
getting you to expand beyond cherry-picking your sources and
just going along with what they say. Here are some ways you can
approach the lit review that might be helpful:
1. It’s about the conversation

You’ll never be able to read everything ever written about
your topic. If you’ve chosen it right, your topic is broad
enough that lots of people have already written about it
in some way. Be careful about limiting yourself to your
more narrow focus, even if there’s been a lot written
on it, because that conversation is only part of a much
larger one. In some cases, there will be tens of thou-
sands of pages written on your topic. Every semester we
hear students say “but no one has written about this!”
or “but I can’t find anything on my topic!” They’re
always wrong; they just don’t know what to look for.
So here’s the thing. Your topic isn’t a set amount of infor-
mation. It’s a conversation about an issue. That means
it’s a bunch of people thinking about, studying, and
writing about something connected to what you’re
interested in. The hottest new thing in music, technol-
ogy, or science is never really new. So you just have to
find out who’s talking about it and listen in. It’s kind
of like going to a party where you don’t know anyone,

and wandering around until you find a group that’s
speaking about something interesting. You probably
shouldn’t just jump in and start speaking, even if you’re
an expert. You have to ease into it.
Think about this in terms of music. When you listen to
music, you aren’t just listening to the artist themselves;
you’re listening to the people that influenced them over
time. It’s one thing to enjoy music, and another to be
able to appreciate it through its history. To truly under-
stand what’s going on with Kendricks’ “Damn,” you
might need to look at how similar emcees have dealt
with issues of race, power, and money. So that might
be Jay-Z with “99 Problems,” Biggie Smalls with
“Mo’ Money Mo’ Problems,” or Wu Tang Clan with
“Cream.” You could also look at lyrical style and the use
of musical instruments, tracing the artists back through
early rap, jazz, blues, and so on.
At the same time, you don’t have to wait forever, because
there’s always more to find. Every project is like a rabbit
hole that has no end. What you want to do is get a feel
for what various people are saying and what the pieces
of the conversation are before you can contribute in a
meaningful way. While that might be enough for many
projects, especially as a student, when you’re trying to
do a more comprehensive job it might require what we
call “saturation.” Saturation basically means that you’re
not hearing anything new, and everything someone
talks about has already been discussed by someone else
in some way. Practically, this means that most of the
citations in any article you read reference other articles
you’ve read. Eventually, it will seem like everyone is just
writing about (and to) each other. And that’s exactly
what they’re doing. Welcome to saturation.
2. It’s rarely just one conversation

When you want to understand how something works, it’s
not enough to listen to just one person, or even a few
people. If everyone’s saying the same thing, you’ve
reached saturation, right? Maybe . . . but maybe you’ve
only looked at it from one point of view, like talking
about politics to a bunch of people who think like
you do. In one of Ted’s pieces about customer expe-
rience, for instance, his literature review went through
several ways scholars characterized and measured how
customers interact with a brand (and whether or not
they’ll become loyal customers). Not only is there dis-
agreement about this, but often the various perspec-
tives seem like they don’t know each other at all. If he
had only looked at one, he never would have been able
to do his project well. In fact, while his paper was on
marketing, it was engineering that gave him his break-
through. It’s like the blind men and the elephant para-
ble: five blind men touch different parts of an elephant,
and each thinks they know what it is (a snake for the
trunk, a tree for the leg, etc.).
3. It’s all in how you organize it
Here’s the thing. Even if you reach saturation, you’re still
going to be overwhelmed. Why? Because sometimes
you have thousands of pages of material connected to
your topic, and you often only have five pages in which
to discuss it. So what do you do? You categorize. You
look for themes (groups of people at the party) and
the perspectives within those themes (the variety of
conversation). You order it in a way that makes sense
as a story that explains your topic and what you think
is going on. And you explain not only why it’s inter-
esting and important, but also how your project fits
in. In other words, you’re taking it to the next level.
So you’ll want to make sure you address not just the

general concepts of what other scholars have said,
but also in many cases their methods and results. You
don’t have to talk about everyone; one or two exam-
ples from each theme will usually suffice, and you can
always just cite additional examples without discussing
them in detail.
4. The lit review is the opening act. Your research is the
main stage
Imagine you’re at a concert (music or a stand-up come-
dian or whatever). Before you see the act you paid
to see, you’ll almost always see an “opening act” or
two. They get you warmed up and in the mood to
enjoy and appreciate the main act. After all, if you’re a
comedian, the last thing you want to do is make jokes
to people who’ve just gotten off work and are tired or
not in the mood to laugh. The literature review does
just that. It sets the stage for your own research, show-
ing why it’s important and what you’re doing that’s
interesting, new, and different. But if you don’t have
that opening act, it’s not going to make much sense.
The biggest challenge of doing a literature review isn’t

why, but how. Even once you have determined your catego-
ries, it often seems overwhelming, especially if you have a lot
of sources. There are many ways to approach the problem, and
we’re going to give you a couple here. Everyone has their own
process, but these have worked for us. The thing to remember is
that this process will lead you to making your argument. How?
Well, once you’ve established the literature, you’ve made a clear
argument about what’s been done, as well as what’s missing or
incomplete. So that allows you to bring in your method and
your choice of data as a way to help complete the puzzle, adding
yourself to the conversation.
Lacrosse stereotypes Instagram and humor Women’s athletic swimsuits Monogamy, polyamory, and the
I. Stereotypes of athletes I. User-generated social media I. Gender in advertising media
A. Gender theory (Instagram, video, II. Perception and portrayal of I. Monogamy constructions and
B. Sport memes, etc.) female athletes discourse, data on relationship
II. Portrayal of athletes in II. Humor and social media III. Perception and portrayal of satisfaction and longevity
the media III. Theories of humor female swimmers (maybe) II. Individual perceptions of
IV. Possible impact of advertising monogamy vs. polyamory
on women, including athletes III. Monogamy in the media
(maybe) IV. Polyamory in the media
Identity and The Last Jedi Re-thinking James Bond Identity in Orange is the New Politics and identity in Zootopia (2016)
(2017) I. White masculinity and the Black and Moana (2016)
I. Sci-fi, politics, identity media I. Portrayal of criminal justice in I. Race, authenticity, and politics in
II. Star Wars, politics, II. James bond and identity the media (vs. reality) animated film
identity III. Evolution of a franchise and II. Portrayal of women in the II. Disney and identity
III. Evolution of a franchise identity shifts CJS A. Princess vs. anti-princess films
and identity shifts III. Orange is the New Black III. Zootopia and Moana
The Office and Adaptations, power, and Images of plastic and pollution Fast fashion and environmentalism
mockumentary TV The Blind Side I. Activist photojournalism I. Advertising and fashion
I. Portrayals of work and I. Football films II. Activism and social media II. Fashion and unsustainability
office life in the media II. “White savior” films III. Environmental activism III. Fashion and environmentalism
II. Mockumentary genre The Blind Side
III. The Office series (US III. Adaptations of “true” stories
and UK)
Figure 3.1 From topic to literature review (samples from popular culture and society)
Gender and cyberpunk Social media addiction Representations of women in the Revenge and identity in Insatiable
2020 I. Addictive behaviors and Harry Potter series (2018)
I. Sci-fi, identity, and treatments I. Gender representations in I. Teen-targeted media, narrative,
gender II. Social media addiction children’s or teen literature and identity
II. Role-playing games (psychology focus on II. Harry Potter series II. Revenge in the media
III. Cyberpunk individuals) III. Gender in Harry Potter series III. Body image and the media
III. Social media use (trends and (obesity-focus)
addiction) (communication
focus on media)
Superman and religious Anti-Photoshop discourse and (Mis)representation of Africa in Female celebrity identity on Instagram
iconography legislation US media I. Celebrity branding and gender/
I. Portrayals of superman I. Advertising, gender, and I. Representations of Africa in sexuality
II. Superman and other body image the media II. Celebrity and social media
comic book icons II. Photoshop and/or image A. Positive over time III. Parasocial relationships and
as cultural/political alteration B. Negative over time aspirant identity
allegory III. Anti-alteration legislation II. Black Panther or other Africa- IV. Social media and self-portrayal
III. Superman and religion and “natural” body portrayal focused films from Black V. Social media and self-perception
movements African perspectives


It’s worth it. Trust me
At this point, you might be struggling to understand

the difference between an annotated bibliography and
a literature review, and I get it. It seems repetitive and
maybe even like busy work. It certainly can feel redun-
dant. What really sets a lit review apart from an anno-
tated bibliography, though, is that this is where you
begin your narrative. Remember, every piece of writ-
ing is persuasive, and this is when you go from read-
ing about others to framing your own argument (even
though it’s through other scholars at this point). Your
annotated bibliography simply summarizes your sources
thus far, keeps them organized, and explains why that
particular source relates to your topic. Your literature
review is where you make decisions about which topics
you’re going to keep, and begin exploring the themes
that exist between your sources. It will be in paragraph
form, because if it’s done correctly it won’t just save
you time writing your actual paper. This is actually the
beginning of your paper! You also don’t have to reinvent
the wheel – there’s usually tons of online guides that
your librarians have created, called LibGuides, which
can help with finding resources and how to complete
specific assignments. Check out our online resources for
some sample LibGuides.
Ted’s method
The first thing I do is download a bunch of articles, and save

the citations to Zotero. By “a bunch of” I usually mean at least
thirty, sometimes more. My favorite tool for that is Google
Scholar, because the links to full text (at least, when I’m on
campus) are right there next to the article and I can download
them quickly. The “cited by” and “related to” links are also
amazing.
Once I have a big set of articles, I rename the files like this:
AuthorLastName-Date-Topic.pdf, and although that might
sound like a pain in the butt (and it is), it’ll save you tons of
time later on. I put them in a file folder that is automatically
synced to Microsoft’s OneDrive so I don’t lose anything, and
then I start to read. But here’s the thing: I don’t read every-
thing right away. I read the abstracts for the articles and then
prioritize them. Usually that starts with sub-folders. In my
disability piece, I made a sub-folder for ageing in the media,
one for autism in the media, and one for theater and activ-
ism. Then I chose the articles I thought would be best out
of that list (I actually made a partial annotated bibliography
from the abstracts, and highlighted those articles to prioritize
in the document). I read those articles first, highlighting them
within Adobe Acrobat, and then put the highlighted articles
in a “read and highlighted” folder to keep myself organized.
I also looked up and downloaded any articles they cited that
looked useful or interesting; it wasn’t long before those sources
all started to refer to each other, which meant I was reaching
saturation.
This is the point where it often seems overwhelming for peo-
ple, and here’s how I dealt with that. I took all the highlights and
comments I typed into the PDFs and copied/pasted them in a
new document for each folder (ageing and media, for instance),
with an in-text citation (e.g., Gournelos 2009, 56) next to each
one so I could find it again. This created three very, very long
documents. Suddenly it wasn’t so much about what to write,
but what to cut; I had five to eight pages that I could use for this
at most, and I had fifty pages or so of quotations alone. So what
did I do? I highlighted them again, in different colors depend-
ing on what part of the literature review (what theme) they
fit. Then I organized them by color/theme and read through
them all together, putting quotations that were similar together
and reducing quotations by paraphrasing where it made sense.
I outlined the literature review with those sub-themes as head-
ings, and then started to write, cutting/citing/quoting from my
sources as I went. Fifty pages turned into ten, as the literature
seemed to just write itself, logically and in an order that made
sense. I was telling a story, in other words, drawing from the
stories of other people. Then I just went through and edited,
usually removing quotations or pieces that I thought were going
to be relevant but really weren’t, or needed to be one sentence
instead of an entire paragraph. It helps to remember that you’re
trying to answer a question, and sometimes there’s some great,
interesting research that gives you context but doesn’t help you
with your question. You can usually reduce or cut all that stuff.
That was it! End of literature review. I’ve used this for every-
thing from a proposal for a book to a journal article to a full,
three-hundred-page book. It works, and it keeps you sane.
Josh’s method
Once I knew what I wanted to study – best practices in health-

care communication − it was time to find out what everyone
else has already said about my topic. Unlike Ted, I start with the
library’s databases, usually LexisNexis or EbscoHost. Depend-
ing on how vague my search terms are, I can pull up anywhere
from twenty to twenty thousand articles; if necessary, I narrow
that with phrases that specify my context (e.g., Physician Com-

munication Patient Satisfaction). Once I have a manageable
number of articles to choose from, I begin by reading their
abstracts. When I confirm that the article’s goals and findings
align well with my topic, I save them to a folder in the database.
I’m too tactile to organize the literature review on my com-
puter screen like Ted does. Call me old-fashioned, but I print
out the twenty-plus articles that I think are the best. I take a pad
of sticky notes, and I write keyword descriptions on each and
stick them to the top of the front page of the article. One article
might have five post it notes stuck to it: (a) Trust, (b) Open-
Ended Questions, (c) Empathy, (d) Nonverbal Behavior, and
(e) Social Support – as variables that relate to Physician-Patient
Communication. The other nineteen articles may have a similar
number of sticky notes, some with recurring themes, others dis-
tinct. I then create piles by placing articles with similar themes
together. At the end, I may have three major sections in my lit-
erature review: (a) Physician Nonverbal Behaviors, (b) Physician
Verbal Behaviors, and (c) Physician Characteristics. Now that
the “conversation” is organized, I can start writing.
I always try to remember that my audience doesn’t need to
know everything. I ask myself, “What in these articles do they
need to know to help understand the conversation before I add
my voice?” Usually that means I include the basic (a) goal of
the study, (b) significant findings they mention (hypothesis sup-
ported? answers to research questions) and (c) the holes/missing
pieces to the puzzle that my study will address. I then go through
each article and highlight the study’s (a) goals, (b) major find-
ings, and (c) holes, then rephrase them in my own voice and cite
them. The first draft of my literature review is always choppy and
reads like a bulleted list of goals and findings from twenty odd
sources – because, well, it is. Next, I go through the literature
review again, adding preview statements at the beginning of each
theme, combining similar studies together, and transitioning
between themes to create a better flow. By the end of the liter-

ature review the reader should know exactly what you hope to
accomplish with your study. A great former professor once told
me: You’ll know you’ve written a great literature review when
the reader knows exactly what the hypothesis and/or research
questions are before they ever read them.
Maridath’s method
The research that I’m engaged in as an information science

scholar is a little different than Josh’s and Ted’s, but the litera-
ture review method is pretty similar. In a recent project on librar-
ians who teach information literacy/research methods courses,
I narrowed my focus after a quick look through the databases
to using “backwards course design” as a way to create effective
courses. It was through my early reading on my topic, when it
was still at its broadest stage, that I learned which part of my
topic I was most interested in. What I learned was that when
I redesigned the syllabus for my course, I did it beginning with
what I wanted the students to know by the end of the course
rather than starting with what I thought they should learn from
the beginning. I learned that this had a name: backwards course
design. It was through my initial reading that my topic gained a
focus, and I was able to narrow my research as I made my way
towards the annotated bibliography and literature review. This
is different from the kind of original research we talking about
later in the methods chapters of this book, except you might
consider it a kind of theory research (see Chapter 9).
My method falls somewhere between Ted and Josh’s –
I begin with searching a Library & Information Science-specific
database, because my field is so niche and specific. The one that
I prefer to use is called LISTA – Library, Information Science,
and Technology Abstracts with Full Text. You’ll note by the
name that some of the content I find in this database allows me

to get to the full-text PDF of the articles, but some of it only
provides the abstract, at which point I use Interlibrary Loan
(ILL). After I find roughly twenty full-text articles (usually each
open in different tabs), I use the citation exporter located within
the database to paste the citations in APA format into a Google
Doc (I fix them later). Then, I quickly read all of their abstracts
to see if I’m on the right track with my search terms. For any of
the articles that seem worth the time, I either request them via
ILL or read them in their entirety if I have full-text access. Usu-
ally, if I’m lucky, at least half seem good enough at this broad
stage, and so I then print each one. I like the feeling of actually
holding the articles in my hands, and truthfully I’m less likely to
be distracted by my computer and can focus better if I read them
this way. Unlike Josh, I prefer not to save the articles in their
respective databases, probably because I work so closely with
electronic resources and know that often these links sadly suffer
from code rot and can disappear. In other words, you’re better
off capturing citations at minimum or complete PDFs at best.
After reading each article, I write several keywords on each
that will help categorize and group them by themes. For this
project it was themes such as: “backwards design,” “informa-
tion literacy,” “course redesign,” “threshold concepts,” and
“metaliteracy.” I group these articles into piles based on their
themes, and for articles that strongly address more than one
theme I decide which seems to be the most important to the
author(s) and place them in that pile.
For each pile I then go through each article, fleshing out the
annotated bibliography in Google Docs, remembering to check
my citation format at this stage with a citation guide. I also use
a highlighter to pluck out quotations in the articles that I think
will be useful for the literature review. Next, it’s time to take
my Google Doc of annotations and summaries and put them
into order by theme, based on the piles of articles I’ve created.
As with Josh’s method, at this stage my annotated bibliography

really looks like a list. From here, I work to turn each sum-
mary into complete sentences, quoting excerpts from the arti-
cles where it makes sense. Each article is grouped by theme, and
these themes become sections of my literature review.
Choosing a
CHAPTER
4 method
This chapter will show you how to move from a litera-

ture review to the method you’ll use for your analysis (or
multiple methods if you want to get fancy). We’ll show
you how to draw from the literature to choose your
method logically, and what questions might be answered
with each. We break methods down into two main pieces:
“portrayal” (how things are constructed or shown) and
“perception/reception” (how things are interpreted and
viewed). Because ethics are a big deal with methods, we
have included “ethics alerts” throughout that will tell you
what to watch out for and why. At the end we’ll give you
an idea of how various methods could be used in the busi-
ness world.
4: Choosing a method 43
Once your literature review is done, you should have a clear idea
of what others before you have argued about your topic and
through what methods. At that point you have four choices for
your own method:
1. Precisely duplicate what’s been done before to support ear-

lier arguments if you think that’s necessary, or contradict it
if you think there are some serious shortcomings.
2. Duplicate what’s been done before, but with a new context
(a different TV show, a different country, a different demo-
graphic or psychographic of participant).
3. Duplicate what’s been done before, but with a new method.
4. Go out on a limb and argue in a different way using a differ-
ent method.
We don’t really recommend #1 or #4, because in most cases #1 will

be boring and won’t really expand the field significantly, and #4
is very risky. Usually you do #4 if little has been done about your
topic, and there’s not much that’s relevant in the broader literature
review. That’s very, very rare, though; in fact, it probably just means
you didn’t explore the literature expansively enough. Take a look
at that first; if you can stand on someone else’s shoulders, it’s a lot
easier than jumping up and down to see over a crowd.
No matter what, though, you’re going to have to choose a
method, and you’re going to want to do this in a way that can
be clearly explained in the paper and that can be easily repeated
if necessary. There are a lot of methods for research, and even
more varieties of each method. We’re going to walk you through
a few of them, but before we get to those chapters we wanted
to give you a feeling for what separates them (it’s all about orga-
nizing into themes, remember?). We’ll use a central example so
that you can clearly see the differences, which is a project Ted is
working on with a student regarding the Netflix series 13 Rea-
sons Why. Figure 4.1 also provides you with a set of methods
Topic 1: Cognitive disabilities in the Topic 2: Quality healthcare Topic 3: Fighting hunger in the USA
media
Focus: Provider/patient interaction Focus: Nonprofit organizations’ use of
Focus: Portrayal of autism and (interpretive paradigm) social media to communicate (critical/
dementia in the media (critical empirical paradigm)
paradigm)
Textual Close examination of two scenes Conversational analysis of ten Analysis of three highest “liked”
analysis from two plays about disability interactions between staff and Facebook posts from three highest
(one about autism and one about patients in the waiting room, budget hunger nonprofits in the USA,
dementia), dwelling on issues of dwelling on tone, information dwelling on use of imagery, tone,
narrative style, imagery, and tone, as provided, and two-way feasible call to action, and “shareability.”
well as place within the story. communication.
Content Analysis of trailers or advertisements Discussion of forty observed Review of all social media posts
analysis of twenty films and television shows patient/staff interactions in from ten anti-hunger nonprofits for
featuring a character with a disability, a waiting room, coding for the months of December 2018 and
particularly autism and dementia, best practices in information January 2019, coding for aesthetics
coding for issues of representation conveyance. choices, use of SMART guidelines, and
and focus/erasure of disability in type of persuasive appeal.
society.
Discourse Discussion of the concept “disability” Discussion of guides for patient/ Review of user comments on all social
analysis as it appears or is erased/concealed provider interactions in medical media posts from 2017 and 2018 on the
in critic reviews of and newspaper textbooks, hospital brochures, and Facebook and Instagram channels for
articles on the Netflix Original training videos, focusing on two five national and five local/regional anti-
Atypical and the CBS series The Big hospital chains and two medical hunger nonprofits, looking for elements
Bang Theory, as well as on specific schools in the state of Florida. of engagement with the issue or the
media relating to dementia in organization.
contemporary mainstream media.
In-depth Interviews with five spectators of the Interviews with ten sets of Interviews with marketing managers
interviews play The Curious Incident of the Dog doctors and patients after their from ten different hunger-focused
in the Nighttime and five spectators consultation to determine their nonprofit organizations discussing their
of the play Forget Me Not within a individual perceptions regarding strategies for social media posts.
week of seeing each play. Participants the interaction’s emotional content,
should include parents and children informational content, and next
who saw the film together. steps.
Focus groups Four groups of five, one group with Three groups of five, one of Four groups of five, one of social
people who know someone with doctors and nurses, one of staff media managers of local nonprofits
autism, and one group people who members, and one of patients. fighting hunger, one with their social
do not. Discussions about their Discussions about what makes media producers, and one with the
perceptions of autism after seeing for a good experience, what are target audiences for the organizations.
The Curious Incident of the Dog in challenges to that experience, and Discussions about intended and actual
the Nighttime and how it might alter how they think things could be uses and gratifications from the posts,
their understandings or behaviors improved. as well as perceptions of the posts and
towards autistic individuals. Identical their strategy overall.
framework for Forget Me Not.
Surveys Although surveys don’t necessarily Providing pre-consultation and A survey here could focus on potential
suit studying the plays themselves, post-consultation surveys to donors, asking them their perceptions
pre- and post-viewing surveys can be patients and doctors is a very of the organization, the issue as a
an effective way to understand how effective way to measure whether whole, their donation practices, and
theater impacts audiences. In this there is a disconnect in how each even specific questions about what they
case, asking audiences about their feels about the experience, and would like to see in a nonprofit website.
understanding of, experiences with, then those surveys could actually Ranking various possibilities rather than
or personal connections to autism be discussed with the doctors to listing individual ones as important can
and Alzheimer’s disease could reveal try to solve any issues that occur. be very useful here, since for nonprofits
whether the plays expand or change Asking questions about what there is the risk that respondents
Figure 4.1 Comparing methodologies

viewpoints. A survey done a month or information was conveyed, how will say that everything is important.
more later as a follow-up, moreover, easy it was to understand the Similarly, asking them to respond with
could reveal longer term impact. information, preconceptions about specific goals in mind helps to narrow
the doctor or hospital, etc. are all their focus (e.g., “Rank which of these
important possibilities. things would be most likely to attract
you to volunteer”).
Case studies In many ways the textual analyses are Looking at patient-provider The case studies used in this project
case studies. However, you could also interaction in a case study would would focus on specific efforts from one
study cases in which the plays were most likely study a specific organization, with information about the
performed in certain contexts (Forget hospital, private practice, or challenges, goals, and changes made in
Me Not) is performed in elder care hospital ward. The tighter the focus, the process by the managers and other
facilities, for instance. The key to a the better. This is telling the story of staff.
case study is specificity. that ward, rather than giving more
general data.
Archival Studying medical archives, media Doctors have changed quite a bit Looking at the archives of hunger-
research archives, or other sources could in the public imagination over focused nonprofit organizations
provide valuable information about time. Studying archives of medical could reveal how those organizations
how these medical issues have been materials, from advertising to (individually or collectively) view their
understood or portrayed over time. American Medical Association place in society, view their donors and
These could be notes, for instance, documents for doctors to clients, and structure programming.
from production, casting, and direction educational documents, could It can also reveal how they have
meetings for shows like The Big Bang provide valuable perspective into the approached communicating about that
Theory, but they could also be from construction of the profession, but programming or those stakeholders.
the stage adaptation of The Curious also how that profession has viewed
Incident of the Dog in the Night-time. patients differently over time.
Ethnography An ethnography of a new production In this case ethnography is a good Studying the lives of clients impacted
of the plays might be a way to fit and could be used to study by hunger is one way to approach the
approach this, but you could also residencies by accompanying new topic through ethnography (living with a
conduct an ethnography of autistic doctors throughout their workdays client of the organization, for instance),
children and their families (perhaps in to see how they treat and learn to but you could also work alongside the
a school) and compare it to another interact with patients and families. volunteers or organization staff for
of people with Alzheimer’s disease a set amount of time to understand
and their families and professional their culture, their struggles, and their
caregivers. various rewards/triumphs.
Experiments Experiments would be extremely As with the disability project, Experiments for this project could use
difficult to do here, because the studying how patients interact with multiple websites or social media posts
populations are so vulnerable. multiple forms of doctors could for various stakeholders (e.g., different
It would be easier to conduct reveal a lot about preconceptions categories of donors, volunteers,
an experiment with caregiver and differences connecting due to clients, etc.) to see their responses.
professionals in schools or nursing identity or attitude, and doing the Then those different media could be
homes, comparing how they treat same with doctors could reveal correlated with other data, like donation
patients who they have been told similar biases and inconsistencies activity or volunteer registrations, to
have autism or Alzheimer’s disease in approach. determine which types of posts work
with those they have been told have best with which stakeholder.
another condition, using actors to
simulate the experiences.

to answer different questions, connected to the three different

research projects in Figure 1.1 (theater and disability, quality
healthcare, and nonprofit digital communications).
Creation, portrayal, and representation
In this first approach, you’re studying something that has been

created to represent the world in some way. Usually this is media:
literature, art, music, television, Instagram posts, and so on. It
could also be speeches, conversations, and even nonverbal com-
munication like body language. The important thing here is to
recognize that something was created to convey a point of some
kind, even if it was never meant to be seen or read. It could be
a poem written by a musician to himself, or a notebook of jokes
written by a comic as part of her process. It could also just as
easily be the history of an event as told by a single spectator/
participant or a series of historians over time.
There are three major methods in this approach that we’re
going to outline for you here, and we’ll go further in depth
in later chapters. Each of these relies on what we call a “text,”
which isn’t necessarily words (in fact, in advertising, market-
ing, journalism, and public relations we call written words
“copy” to distinguish them from other texts). A “text” could
be a poem or novel, sure. It could also be a photograph,
a film, or even a person (e.g., an analysis of how a subcul-
ture portrays itself through clothing and accessories). Places
can also be texts; imagine how your classroom or office is
designed and laid out to create a certain perspective. Is there
a long boardroom table? If so, who sits at the head, and who
at the foot? What’s the table made of, and what’s on the walls?
Is your classroom arranged with seats in a circle? Is it a huge
lecture hall with a lectern at the bottom and a giant screen
for a PowerPoint presentation? How might that arrange the
way we think when we enter the room? On what social and

cultural backgrounds does that interpretation rely?
Close textual analysis (sometimes also called multimodal

semiotics)
• In-depth reading of one or a few texts, with emphasis on
their physical structure (e.g., shooting and editing for film,
layout for design, rhetoric for a speech).
• Example: While discussing the second season of 13
Reasons Why we thought it was important to look more
deeply at the introductory scene Netflix added to the
first season after controversy over the show reached a
certain level. It alerts viewers to potentially triggering
content (that is, there’s evidence to suggest seeing sui-
cide portrayed in the media can increase the likelihood
of people who are considering suicide acting on those
thoughts). We wanted to see how Netflix created that
scene, and what its details might show us about how it
could be read. We found that the language used, as well
as the dramatic tones and some of the visual choices,
actually make the problem worse. They almost chal-
lenge viewers, suggesting that to not be able to watch
the show would somehow be cowardice or evidence of
childishness. Note that we aren’t claiming how it will
be read, or what the intent was; we’re just talking about
what we see and hear.
• Limitations: Textual analysis is a great building block for
other methods, and often other methods will feature minia-
ture textual analyses throughout as a way to show examples
of elements that they are analyzing through another method.
However, at the same time it is limited by the number of texts,
and might inherently suffer from cherry-picking (choosing a
text because it supposedly proves your point). One way to
avoid this is to look for analyses in your literature review, or

texts that you can analyze, that contradict your initial findings.
Content analysis (qualitative and quantitative)

• Brief reading of a large number of the same sorts of texts,
usually with the aid of explicit “coding,” in which com-
monalities and differences are examined across all texts and
charted, usually in part for statistical significance.
• Example: In the first version of the 13 Reasons Why study,
the student began with a content analysis of the first season.
The codes she chose were drawn from reports on suicide pre-
vention in the media, what we call “best practices,” drawn
from multiple sources in medical and psychiatric journals.
She broke those practices into smaller parts, and then coded
for the number of times each episode did any of the posi-
tive (e.g., “showed someone asking for help”) or negative
(e.g., “glamorized death with post-suicide attention”) prac-
tices referred to in the literature. She added up the numbers
for each code, and was able to determine how much each
episode, as well as the entire season, did or did not adhere
to best practices for their audience’s safety. In reworking the
article, Ted performed the content analysis independently for
what we call “inter-rater reliability,” and then he and the stu-
dent chose certain scenes for mini-textual analyses that could
further explain the data with added depth and richness.
• Limitations: Content analysis is very useful, especially if
you include strong quantitative data combined with strong
qualitative elements. However, you can only analyze so
many texts, and so it is a challenge to avoid cherry-pick-
ing on the one hand, while providing adequate depth to
demonstrate nuance on the other.
Discourse analysis (often called critical discourse analysis)

• Brief reading of a large number of usually different kinds of
texts in order to discuss the ideological landscape surrounding
a central topic, such as journalism, popular media, educa-

tional systems, state institutions, or family systems.
• Example: To understand the controversy around 13 Rea-
sons Why, we considered a discourse analysis of reviews
and articles about the first season of the show, and perhaps
the second season if necessary. In that case we would have
compiled dozens if not hundreds of official (from estab-
lished television critic) reviews, newspaper articles like
opinion-editorials (op-eds), and even reviews from non-of-
ficial sources like blogs, posts on social media, and posts
on aggregate websites like Rotten Tomatoes. This would
have allowed us to understand how people were speaking
about 13 Reasons, and could have even expanded to dis-
course about teen suicide (or suicide) more generally. In
other words, we would have been able to look at the con-
text and debates that surrounded the show, rather than the
show itself.
• Limitations: Discourse analysis is an incredible tool, but
it often requires a lot of time and space. It’s very difficult
to do well either quickly or in a short paper. It is also more
qualitative than a content analysis, since you’re engaging
themes more than individual codes. However, if you limit
your sample size and look for something more specific, dis-
course analysis can be a very effective tool to study reviews
of media, sentiment about an event through social media
comments on that event (e.g., through a hashtag search),
or other important concepts and events.
Perception/reception
In this second approach, you’re not interested in what’s been

created. Instead, you’re interested in what people/societies/
eras/governments etc. thought and felt about what was created.
In other words, the content of the actual text isn’t as important
as the reactions to that text. It’s also important to recognize that
just because you talk to audiences doesn’t mean you’re writing

about those audiences. You might, for instance, interview busi-
ness leaders about their approaches to marketing an environ-
mentally friendly product, but your study is on the approaches
themselves. In that case, you’re simply doing what we call
“formative research” with interviews, not actually studying the
perception of the business leaders themselves. This is a crucial
distinction, because it has some pretty big ethical implications.
In formative research, you’re trying to get a feel for a topic (just
like when you’re doing a literature review) by exploring that
topic. That could be watching a TV show, visiting a country, or
talking to people. But that’s not necessarily going to be your
method.
IRL: ETHICS ARE IMPORTANT
When you’re making people (or animals, in some cases)

your primary focus of analysis, you might be subjecting
them to something that could be upsetting or dangerous,
or even simply exploiting them for the purposes of your
research without telling them you’re doing it. Imagine,
for instance, going on a bunch of dates with someone just
for the purposes of writing about dating (yes, that’s a How
to Lose a Guy in 10 Days reference). That might work if
you’re writing a column for a magazine, but it’s not ethical
scholarship unless you’ve told them what you’re doing,
how, and why. Most people probably wouldn’t go on that
date with you if they realized you were studying them,
right? It’s important to remember that your research isn’t
as important as the people (or animals) you’re studying.
For a study like this one, you’d need them to sign an

agreement acknowledging they consent to be studied,
and in most cases you would need to send the study
through what we call Institutional Review Board (IRB)
approval. The IRB (usually a collection of professors or
researchers) would then look at what you’re planning
to do and with whom, how you’re making sure they’re
safe and informed, how you’re compensating them, etc.
before allowing you to do the project (see Chapter 7’s
online resources for more on this process). This might
seem like common sense, but it isn’t. Many studies done
even in recent memory have inadvertently traumatized
their participants (e.g., the Zimbardo prison studies at
Stanford) and even led to their deaths (e.g., the Tus-
kegee syphilis study). In other cases, millions of dollars
have been made from a study and the participants or
their families received nothing in return (e.g., the use of
Henrietta Lacks’s cancer cells). Whether you’re working
in a classroom or a billion dollar company, make sure you
do all your research ethically. It’s not that hard, and it
actually makes your study stronger.
In-depth interviews
• Structured, semi-structured, or unstructured, interviews
often conducted with fewer participants but in great depth,
usually from 1–2 hours in length; in-depth interviews (IDIs)
try to get participants to tell stories and explore concepts in
detail, often with follow-up questions or examples.
• Example: So say we didn’t really care about what 13 Rea-
sons Why was doing, but instead we wanted to see how
people were reacting to it, or to issues around it. This

would be a good reason to do in-depth interviews. We
would choose a small sample of people (perhaps ten col-
lege students would be a good example in this case), and
if we wanted to get more detailed we could even ask stu-
dents to participate that have seen the show, hadn’t seen
the show, or who have even suffered from depression. We
would absolutely need IRB approval, especially for the
depression group, because it would then be a vulnerable
population that would also be potentially at risk through
this research process. In any case we would want to alert
the campus wellness center about our project, and have a
brochure or other materials, as well as contact information
for counselors, which we could give to all participants. We
should offer people that we interview compensation for
their time, either in money or some other form (we like
the idea of contributing to a related nonprofit organization
like a suicide prevention hotline). Then we would create
an interview guide of a series of open-ended questions that
would ask participants for stories regarding how the show
and other media have impacted them in the past, how they
dealt with it, what they would have liked to see or what
they were disturbed by seeing, and so forth. We’d audio
record these interviews (with permission and a signed
release, of course), and then transcribe them into a word
processor (and as you’ll see with Chapter 10, programs
like NVivo are very useful for this as well). You can use free
programs like Speechlogger or paid programs like Trint or
Descript for transcriptions to speed up the process. Then
we’d read the transcripts and pull out significant concepts,
like we did with our literature review: we’d group themes
across participants and write them out in detail. You’re
trying to bring out a sense of what they’re thinking and
feeling, and to do justice to their points of view.
• Limitations: Interviews are great, but the sample size is

small and you have to be careful to restrict your sample’s
range because of that. In some cases, like if you’re inter-
viewing professionals (CEOs, law enforcement, doctors,
etc.) this might be the best way to go because of difficulty
accessing them and the importance of individual perspec-
tives. But in general it means that while you will probably
find some interesting information, its applicability is limited.
Interviews are often a great way to both start and end a
project. You start with them in order to get a sense of what’s
at stake, but you often end with them once you have ques-
tions developed through other methods in order to expand
and add depth to your data.
Focus groups
• Small groups of individuals that are targeted to gain specific
information about a topic; usually random groups or groups
that are intended to get a range of opinions, focus groups
are widely used in industry studies of media, like whether a
new TV show will be popular, or in social sciences to gain
perspectives on things like body image, and so forth; focus
groups are great because you can use a semi-structured
interview guide but encourage conversation among individ-
uals, which you record and then analyze; however, they can
also result in “groupthink” or silenced voices.
• Example: Focus groups are often used in media analy-
sis, including by industry professionals who are testing a
new show/film/advertisement or deciding on whether to
renew/continue/change what they’re already doing. For
13 Reasons Why, we could do focus groups with counselors/
psychologists/psychiatrists about the benefits and problems
they can see with the show for their clients, or we could
do focus groups with students like in the interview exam-
ple (both again with IRB approval). In this case, we want
stories but we also want participants to communicate with

each other. Discussion is where the richness of this method
comes from, and you as the interviewer are actually just a
discussion moderator. Usually you’d do focus groups with
two or three sets of people that are either all similar (to see if
there are differences within them) or different (profession-
als vs. parents vs. students, for instance) to see if the audi-
ence differences translate to interpretation differences. In
this case, we might discuss Netflix/television use in general,
then dramas, then controversy in dramas. Then we could
watch part of or a full episode of 13 Reasons Why before
going back into discussion about responses. Participants are
paid as with interviews, and the discussion is again tran-
scribed and analyzed through the same process.
• Limitations: Focus groups are great for discussion, but
they can be bad for data about anything but discussion.
That’s because people will often respond with what we call a
“response bias,” perhaps regarding what they think the oth-
ers want to hear or don’t want to hear (often also called a
“social desirability bias”). They might make claims or censor
themselves in a way that tells us a lot about “groupthink,”
but not what they actually believe or even why they answered
without complete honesty. Even the person that is interview-
ing or mediating can produce this effect through what are
called “demand characteristics.” In the “Bradley Effect,” for
instance, researchers found that in the United States people
might claim to vote for a racially non-White candidate, but
they might still actually vote for the White candidate.
Surveys
• A questionnaire of some sort given to a sample of individ-
uals to gain their perspective on a topic; samples are usu-
ally large, but depth of reaction is often not as intricate as
focus groups or interviews; common scales are Likert scales,
and samples range from “convenience” to “snowball” to

“random.”
• Example: If we wanted to see how a large group of peo-
ple responded to or thought about 13 Reasons Why or a
related issue (e.g., portrayals of suicide, bullying, or sexual
assault on television), a survey might be a good method.
We would still have to choose a sample, and this could (like
focus groups) be a set of groups that have differences or
a larger number of participants in a more homogenous
group. We’d create questions that ask specific things about
the show, perhaps in response to a clip we’ve embedded
in the survey, and questions about whether someone has
watched and heard about the show, what their perception
of including such controversial topics is, and whether or
not media companies have an ethical responsibility to their
viewers. We would then gather the data and break it down
into usable conclusions. That might be percentages of the
group who felt a certain way, average responses for different
demographics, correlations between two different points, or
many other possible analyses.
• Limitations: Surveys can be a struggle because it’s some-
times difficult to ask questions that get at to what you
want to know, and even harder to do so in a short enough
survey that participants don’t get tired and just start fill-
ing items out without thinking. We’ll get into this more in
later chapters, but while surveys can tell us what the general
trends are, they aren’t always the best way of determining
the complexities and nuances of how people think. Check-
ing a box among a list of options can limit people’s true
thoughts and feelings. Sampling is also difficult, and attract-
ing enough participants to actually answer your questions
can be a real struggle. Even when you’re finished and the
data is analyzed, moreover, you may want to follow up with
other methods to gain more depth and perspective. After
all, maybe you’re only looking at incomplete data because

you’re only asking the questions you want to know, and not
allowing for respondents to give you contradictory infor-
mation (e.g., “loaded questions,” “leading questions,”
etc.). One other limitation is that people often only respond
to surveys if they’re (1) forced to do so, (2) incentivized,
or (3) feel passionate about the topic. When you analyze
responses from people who feel compelled to give them,
you may notice some false polarization (large amount of
positive/negative). This is why it’s so much more common
to see strong reviews for products than moderate reviews;
if something works the way it’s supposed to, or if we have
adequate service, we rarely feel that it’s anything special. If
it breaks or is bad (or changes your life for the better) you’ll
be more likely to write a review. So instead of looking for
5- or 1-star reviews on Amazon, we should probably focus
on what the 2-star and 3-star reviews say. Those will provide
the nuance or richness of information that will actually tell
you if the product is any good or not.

Think before you method
A word of caution on choosing methods in general: each

discipline in the social sciences and humanities has what
I like to call “pet methods.” For one thing, as I’m sure
you’ve noticed throughout this text, there are certain meth-

ods that just work better depending on the discipline, and
even depending on the topic you’re researching within that
discipline. There are also some major philosophical divides
even within disciplines, so even if it’s, say, your fifth class
in a particular major, it’s a good idea to work with your
professor or a colleague on selecting the most appropriate
method for your assignment. Surveys in general seem to
create the most amount of controversy. Why might that
be? Well, as we’ll discuss in Chapter 8, they can suffer from
the same issues as your literature review. Selecting the cor-
rect number of participants, the type of participants, and
even how you get your message out to potential partici-
pants can affect the outcome. Also, surveys are cursed by
notoriously low response rates, making matters even more
difficult. We’re not saying don’t use a survey, but definitely
understand the pitfalls of any method ahead of time. Your
librarian can help with these issues, including helping you
to design your study if you’re stuck.
Observational analysis
There are many ways to approach observation, and we won’t

go into most of these here because they’re often different from
discipline to discipline. Basically, all of these methods require
you to observe something in the real world and then draw con-
clusions from it. This happens in the methods we have already
discussed, but the emphasis here is on the observations and the
structure in which the text is observed (the time, place, situa-
tion, etc.). Some observational methods have particularly large
ethics requirements, so we’ve highlighted those below in ethics

alerts. We’ll look at some more later, but for now we’re just
going to highlight a few that you might encounter.
Case studies (often used in psychology and business)

• Study of a specific instance, often focusing on a business or
a person, that might explain or be paradigmatic of a broader
trend or issue.
• Example: With 13 Reasons Why, a case study approach in
this sense could follow one person’s response to the show
(e.g., if we were writing it from the perspective of a psychol-
ogist). It could also follow the business case for the show,
discussing it as a Netflix Original and how/why it became
so successful. Similarly, it could follow the public relations
challenges faced by the company during the controversy
over the show, and how Netflix did or did not follow best
practices for public relations in their responses.
• Limitations: Like textual analysis, case studies are limited
by number. The point here is to show why this particular
example is interesting and what we can learn from it, not
to make general points about, say, depression or media pur-
chasing or public relations. These are great for cautionary
tales or success stories, but we should always be wary of
applying one case study to another situation and assuming
it will follow the same process or pattern.
Archival (often used in history, political science, and

metasynthesis methods)
• Similar to a discourse analysis, archival research uses an insti-
tution’s records about an event, text, process, etc. to deter-
mine how it was created or discussed. This kind of research
is also used in metasynthesis, which brings together a bunch
of scholarly sources to examine their results all together,
almost like a qualitative content analysis of scholarship.
• Example: One of the most interesting things about any

television show is how it was created, from beginning to
end. Looking at an archive of materials around 13 Rea-
sons Why could reveal decisions from casting directors (and
maybe original casting videos), script changes (e.g., for the
third season actress Anne Winters suggested they include
abortion in some way), and reactions to controversy, popu-
larity, etc. This gives us insight into what the various people
that constructed the television show were seeing, listening
to, and talking about during the process of its planning and
creation, as well as, of course, the decisions to highlight it
on Netflix and to renew it for subsequent seasons.
• Limitations: Archival research depends on having an
archive, obviously, which doesn’t always happen. It also
depends on the researcher recognizing that any archive is
going to be incomplete, which will skew her understanding
of the topic.

Using archives
The word “archive” can be used to mean a variety of

things, unfortunately, but in this case we mean a physical
place that’s usually located in a library or museum with
a name like “Special Collections and Archives.” In this
repository, the archivist(s) collect and organize collections
of history as well as of (often) famous people like artists,
musicians, and celebrities. Occasionally, archives will also
collect and store items that center around a theme. These

items might be letters, maps, and other documents, but
they could also be artwork, clothing, or recordings. The
Archives of Social Change, which was created at the
W. E. B. Du Bois Library at the University of Massachusetts,
Amherst, is dedicated to preserving materials related to
social change in the United States. Archives generally have
abbreviated hours and often are by appointment only, so
you’ll want to make sure you allow yourself enough time
to get in and handle the materials well in advance of your
deadline. For more on archives and archival research, see
my note in Chapter 6.
Ethnography (often used in anthropology and business)

• Investigation of the discourses, events, rituals, and com-
munications within a community or a (set of) individu-
al(s) within that community; although auto-ethnography is
absolutely valid as a method, it must be chosen with great
care and attention to the method as appropriate for the sort
of impact and evidence necessary for the project.
• Example: An ethnography of 13 Reasons Why could focus
on the filmmaking set and discussions happening as the
show is being filmed, or even the editing room and the dis-
cussions/decisions made there. It could also be in a high
school where some students are watching the show, observ-
ing the ways it does or does not enter into their lives and
daily communication. In the first cases, we would speak
to movie executives to get approval, and then get consent
from everyone observed in the study. We would then spend
a significant amount of time (meaning, as much time as
they spend working) observing the creation of the show,

taking “field notes” (miniature textual analyses of discus-
sions, places, etc.), and extracting themes in the same way
we might in other methods.
• Limitations: Ethnography is a fascinating method, but it is
always limited at least in part by the perspective/perception
of the researcher, as well as social desirability bias. One of
the main issues is the rise of attribution bias or attribution
errors, which happens when we assign positive attributes to
people and situations to which we can relate to ourselves
and our experience, and negative attributes to those that we
can’t easily relate to or who are dissimilar. Worse, we tend to
assign those positive or negative attributes to their identities
or cultures, rather than to the individuals or the situation
in which something occurred. Prejudices like the negative
construction of “Asian drivers” or “women drivers” draw
from such biases and are not only incorrect but can also do
tremendous damage.
ETHICS ALERT!
In ethnography, you’re observing people in their “nat-

ural habitat” and drawing conclusions about them from
it. This has led to a lot of problems over the years, from
researchers observing simply to justify their preexisting
beliefs, to incomplete claims, to the exploitation of vul-
nerable groups. It’s essential in ethnography, even auto-
ethnography (the study of the self), to get consent from
all people involved, and inform them of what you’re doing
and why (and what they might get out of it, how they can
limit or control the data, etc.). Imagine you’re studying a

group and you end up revealing some information about,
say, infidelity, identity, or illness that could lead to social
stigma or other kinds of harm. This could destroy lives,
so you have to be very careful and always remember that
(1) you need IRB approval, and (2) you are in a posi-
tion of power as a researcher, and it’s your responsibil-
ity to be responsible and ethical. This is also for yourself,
and of course the people in your life that might inevitably
become a part of any self-study. The challenge is that when
you alert the subjects, they will act differently and thus
lose “authenticity,” so you have to find a way around that.
Experiments (often used in psychology, communication,

and business)
• Within a controlled environment, manipulating variables in
one group in order to determine changes.
ETHICS ALERT!
Experiments can be very revealing. They can show what we

see and don’t see, what we listen to, and what we just hear.
They can show us our fears, our biases, our hopes, and how
we respond to authority. They can also, however, result in
severe trauma. Children in the famous 1970 “blue eyes
brown eyes” study by Jane Elliott, for instance, in some
cases still feel that trauma fifty years later. Participants in
Philip Zimbardo’s 1971 study on prisons at Stanford Uni-

versity are similarly impacted. The list goes on, from Mil-
gram to Tuskegee. Experiments require careful planning,
careful consideration of the impact of the study, and care-
ful interactions with participants before, during, and after
the experiment. There are so many potential issues that we
can’t address them all here. Our example, however, will
show you one possible problem and outcome.
• Example: To end with a scary example, we could conduct

an experiment in which we took two groups of teenagers
that were at risk for suicide, separated them, and exposed
one of those groups to 13 Reasons Why. We would measure
how differently the control group (that wasn’t exposed) and
the experimental or “variable” group responded, perhaps
measured by pre-viewing and post-viewing surveys and per-
haps by conversations with a therapist. The risk here is of
course that being exposed to the show could aggravate their
tendencies to self-harm, and even result in someone’s sui-
cide. Experiments can be dangerous things, so tread lightly,
do your literature review exhaustively, and work with pro-
fessionals in all cases. And always, always, always work with
your IRB.
• Limitations: Experiments are fascinating but, like every
research method, they also have their limitations. Easily the
biggest critique of experiments is also one of its strengths:
the need to control. In other words, in order to see if what
you’re studying is causing a particular change in a group
of people, you need to focus on that variable and mini-
mize ever other possible influence on the group. So, if you
want to see how a film or TV show might affect people’s
emotions, you would need to make sure that you remove

as many other possible variables as well (e.g., other people,
room ambience, screen size, volume, etc.). The best way to
make sure that happens is by bringing them into a labora-
tory to control and standardize every other possible influ-
ence. However, in controlling and standardizing, you have
just now created an artificial environment that may or may
not reflect reality. Ironically, the very same laboratory room
you are using to control other factors actually is creating its
own effects on your participants. So, yes, experiments can
be a great way of isolating and focusing on how specific
variables may change the outcome of a group, but anytime
you remove human beings from their natural contexts (e.g.,
living rooms, classrooms, etc.) you run the risk of telling an
“artificial” story.
Request: Find out what the Request: How do our customers feel Request: Are our employees happy?
competition is doing and where about us, and are we doing well or What can we do to make them more
there’s opportunity for us to grow. badly? productive?
Textual Look at how your competitors are While it’s possible to look at customer Do a textual analysis of a break-room
analysis advertising their products and if blogs, videos, etc. if you have an issue or an office space, connected to best
there are missing demographics (e.g., with United Airlines), usually practices in the industry for productive
or situations (e.g., selling SUVs to you wouldn’t choose this method workers. For instance, Clif Bar &
“soccer moms” instead of to dads). because it’s about perception, not Company has an office filled with
portrayal. athletic equipment, and an in-house
massage therapist and gym.
Content Compare your competitors’ social Analyze all of the complaints and Although this often wouldn’t be the
analysis media presence to your own, coding other communication that comes best method for this question, you
for types of messages, media, through email and on social media, could analyze suggestions made by
demographics, even tone (in addition like comments on posts, tweets about employees through an anonymous
to likes, shares, etc.). Target at least your products or customer service, email survey, or responses made
the past year. and hashtag use. Code for tone, issue, during exit interviews when
and positivity/negativity, as well employees leave the company. Code
as whether it was resolved and/or for tone, similar concepts that might
responded to by your staff. be positives and negatives, and
forcefulness of feeling.
Discourse Because discourse analysis is about One of the most useful things you As sites like Glassdoor start to provide
analysis perception and social construction, can research is how people are extensive, anonymous reviews of
it doesn’t really help much when speaking about your company and its companies, it’s increasingly important
figuring out what your competition competitors. You could look at social to monitor them to determine what
is doing. However, it could certainly media accounts of “influencers,” or employees are saying when they work
reveal opportunities and where you just look at what people are saying for you and when they leave. This
and your competitors are succeeding when they hashtag you or comment can head off some serious problems,
or failing. For instance, when you on your own social media. You could showing you issues with ethical
Figure 4.2 IRL: How to do what your boss asks

learn that people are interested in also look at reviews on Amazon, or violations, bullying, inefficiency, and
socially good products, or companies newspaper coverage related to what other things that you should fix.
with progressive politics, you you do.
can tailor your message and your
products or services to match.
In-depth If you want the most rich, detailed By selecting a few representative Similarly to the customer interviews,
interviews information on why people are customers (for example by location, you might select employees from
choosing your competition, in-depth revenue amount, or company size), different departments or levels in the
interviews can really become and conducting IDIs with a set list of organization, and administer your
immersed into the perspective of questions. Be careful to avoid leading IDI. While you can expect richness
the customer. Rich descriptions and questions of course, and if possible of data with this method, you most
unique stories can help paint a clear get a third-party to administer the IDI. likely will spend time looking for
picture of why customers/clients are common themes that emerge from the
choosing your competition. qualitative data.
Focus To help flush out some of the major Focus groups can be extremely A focus group to determine employee
groups stories why people are choosing your helpful in determining how customers satisfaction might be useful, however,
competitor, you can conduct focus feel about your company. This method in order to protect anonymity and
groups to hear from several people can also be helpful when attempting avoid groupthink, you’d probably
at once. Focus group research can to make claims about which particular want it administered by an outside
help remind and confirm the attitudes groups like or do not like your facilitator. For this particular question,
associated with competitor allegiance. company, based upon the makeup a focus group might not be the best
of the focus groups you build. Focus method. Surveys and especially IDIs
groups can sometimes suffer from might be better in terms of providing
“group think,” so be careful of this rich feedback in a setting that avoids
pitfall. bias and makes your participants feel
as though they can speak candidly.
Surveys You want to find out what are the Surveys can be useful in getting a Surveys provide employees with the
most common reasons people wide array of responses fairly quickly, anonymity they need to speak freely
are flocking to your competitors, but you might limited on how much about the company, but you won’t
conducting surveys can help you information you get. It’s nice to have a get the same amount of qualitative
find trends and patterns. It can also summary of overall attitudes, but you responses as from an IDI or focus
identify how high/low customer’s may not get the whole story. group.
attitudes rank when they purchase
your competitors products
Ethnography An ethnography of a group that you Embedding yourself into campus It’s important sometimes to get a
want to sell your product to is a great tours, dealership test drives, and feeling for what work is like on the
way to get good information. You can other situations can help you gain ground. Are people overworked? Do
easily figure out what people buy, a close look at how people interact they feel undervalued? What do they
when, and how much they spend, in those strange situations. What worry about, and what might make
but ethnography helps you truly do families talk about when they’re their lives better and easier? Those
understand the many reasons why. buying a car or choosing a school, and answers are rarely as simple as “more
what are the things you didn’t expect? money,” because you still need to
know why.
Experiments Experiments are a great way to The challenge and benefit of Again, this would be far more involved
confirm whether specific variables experiment is that the variables you than some of the other methods, but
are causing a specific change in a are measuring need to be exceedingly you could develop interventions to
group of people. In examining a clear. When assessing attitude/affinity determine employee productivity and
competitor’s business strategies, you often experiment design researchers satisfaction and compare the overall
could determine if a specific strategy will assess hormone levels and/or scores across the conditions.
is influencing customer acquisition or brain activity to determine how much
sales. someone likes/dislikes a company//
person.

PART
Methodology
overviews II
Close textual
CHAPTER
5 and thematic
analysis
Chapter 5 is the first of our methods chapters, and we’ll

take you through how to analyze individual and small sets
of texts. This can range from media, to people’s clothing,
to places, to situations (e.g., how different cultures tradi-
tionally handle it when people meet for the first time). We
will discuss how “multimodal semiotics” helps you frame
this approach, but we’ll also address how textual and the-
matic analyses can be used to determine relationships,
power issues, and social norms.
Textual analysis has many names, probably because it comes

from different places and perspectives. This method was in many
ways developed by historians and scholars in the visual arts,
music, and literature. It looks at the elements of a text to deter-
mine what is occurring within it, without in most cases making
claims to audience interpretation or “author.” Before going any
5: Close textual and thematic analysis 73
further, it’s worth looking at those in a little more depth. While

scholars often refer to the “author” (e.g., a painter, poet, or
director of a film) as the creator, it is also usually implicit that
what is actually being read is the text, not the intent behind
the text even if we use phrases like “Shakespeare is creating a
tragic flaw in Hamlet.” After all, who knows what the author
was actually doing, no matter what they say they were doing?
In reality, it doesn’t matter and to make this kind of claim is
what we call an “authorial fallacy.” Similarly, while we often say
things like “which makes us identify and empathize with the
character,” that’s not really making a claim to an audience inter-
pretation, but instead what you as a scholar are suggesting is a
likely audience interpretation. Otherwise, you’re committing an
“audience fallacy,” since you’re analyzing one thing but claim-
ing things about another.
IRL: DON’T ASSUME. THINK AND ASK
In one famous marketing failure from 2017, Pepsi released

an advertisement featuring Kendall Jenner in which she
joins a crowd for what looks like a street protest full of
beautiful, multi-ethnic people. They are confronted by
a line of police officers, and Jenner prances to the front
of the protest so that she can give an officer a Pepsi. He
drinks it and is happy, and so then everyone plays music
and they’re all friends. This was a not-so-subtle reference
to a Black Lives Matter demonstration against police bru-
tality, in which a young activist named Ieshia Evans calmly
stood her ground against charging police in militarized
riot gear. It was also a reference to the famous “Flower
74 Part II: Methodology overviews
Power” photograph from 1967 taken by Bernie Boston,

in which protesters against the Vietnam War placed flow-
ers in the muzzles of rifles pointed at them by a squad
of soldiers. In either case, the advertisement was not only
in general bad taste, but was also extremely offensive to
much of the demographic Pepsi sought to target (cultur-
ally open, diversity-focused millennials). Anyone on staff
should have been able to immediately see the problems
with this ad, and should have spoken up. Even if they
didn’t see it, they should have asked others for their inter-
pretations. This is a great example of how a failure to use
even basic methods of critical research can lead to a public
relations crisis, in this case a crisis that cost the company
millions of dollars.
Textual analysis is often called “close textual analysis,” and

some scholars call it “multimodal semiotics,” which means the
study of signs in a text from a variety of perspectives. Under-
standing what “signs” are in this case helps explain textual analy-
sis as a whole. Think about it this way. Let’s say someone draws a
hexagon for you. OK, great. You have a shape. Congratulations.
However, when Ted does this in his classes, invariably most of
the class interprets the hexagon as a stop sign. If it was colored
red, obviously the connection would be even stronger. How-
ever, nothing about a hexagon says “stop,” which means that
our perceptions and beliefs are so embedded within our culture
that a simple shape immediately means (or “signifies”) some-
thing profound. And what’s so profound about a stop sign?
Well, it means “stop,” of course. But really only if you’re in a
car, right? And only if you’re driving that car, right? And does
it only mean “stop?” Who’s telling you that, and why? What are
the consequences if you don’t stop? A stop sign is actually say-
ing “stop . . . or else.” That last bit is important, because while
a stop sign is signifying “stop,” it is also signifying a system of
ethics in which cars should be careful to avoid people, other
cars, and other objects, as well as signifying a system of power
in which if you don’t stop, you might get a ticket from a police
officer. Either way, it’s going to be a bad day. So that’s textual
analysis. You look at as many details as you can to get an under-
standing of all the different levels that build the sign, or what
we call “signification.” We’ll take you through some examples
throughout this chapter that will make this clearer.
Details in a textual analysis come from many different places.
In literature they might be the way words work together
(meter, rhyme, assonance, etc.), in film they might be light-
ing, placement in a scene (mise en scène), and color, and both
might examine setting, narrative, and dozens of other elements.
It’s helpful to categorize textual analysis into some general
concepts: formal/aesthetic (how it sounds, looks, etc.), allusions
(references to other texts), narrative (the story), setting (place
and mise en scène), etc. All of this depends on what exactly
you’re analyzing. So let’s start with a couple of possibilities.
1. Art: One of the most famous examples for textual analysis is

a 1929 painting about textual analysis, René Magritte’s The
Treachery of Images. It’s a painting of a pipe floating with no
context on a tan background. Under the pipe in French it
says “Ceci n’est pas une pipe,” or “This is not a pipe.” So . . .
ok. What’s going on here? There are only two signifiers,
really: the pipe, which is painted fairly realistically, and the
words (sometimes called “the copy”). Of course, this is a
pipe. But it’s also a painting of a pipe, and when it’s shown
in class it’s a computer projection of that painting of that
pipe. But the pipe is also a representation of other things,
right? Perhaps masculinity. Perhaps home. Perhaps upper

class “distinction.” Perhaps hobbits. But what Magritte is
saying here (or rather, what the painting or text is saying) is
that we need to think deeply about any signifier, because it
rarely only means one thing.
2. People: Maridath, Josh, and Ted are all shameless people-
watchers. There’s nothing quite like trying to figure out what
someone is thinking, what their background is, and who they
really are just by examining them from a distance. But what do
we look at? Nonverbal communication like gestures, clothing
(“artifacts”), eye contact, hair, personal hygiene, the volume
of their voice? Verbals, like the words they use, the language
they’re speaking? Setting, like the place and whether they
“fit” in that place? Context, like the social situation we see
them in and the people/events/objects around them? The
answer is all of those things and more. A crazy bachelorette
party would look completely normal at a club in Las Vegas.
Take the same party and put it in a breakfast truck stop diner
in a small rural town at 7 am and it looks insane. Deep textual
analysis looks at all of these things, because the richer your
data the less likely you are to make mistakes due to bias or
your own situation and mindset.
3. Place: Places are fascinating to study, even boring ones like
a classroom. How do students know to sit in their regular
seats, rather than at the front of the room? Are the chairs in
tiers like a lecture hall, or small tables where students look
at each other, or in one big circle where everyone sits down?
What color are the walls? Are there interesting paintings on
the walls, or are they blank or dull? Is the furniture made of
old, expensive hardwood, or mass-produced industrial plas-
tic? What does all of this say about a university’s perspective
on education? On the students? Do they trust students to
pay attention? Do they trust students to have respect for
the faculty member? Do they value student voices as much
as the professor’s voice? A textual analysis doesn’t just look

at what’s visible; it also looks at what is invisible, whether
that is something that is erased (like women and non-White
authors from many literary canons) or something that is
simply not present but could be (like a round table in a
boardroom or dining room instead of a long rectangle).
4. Television: One of Ted’s favorite examples of textual anal-
ysis is the opening credit sequence to a show that he grew
up with: The Cosby Show. It’s an incredibly important tele-
vision show for the history of US and global media, since
it was unheard of for a middle-class Black family to appear
as the center of a situational comedy (sitcom) full of mainly
positive, affirming humor. It was a wild success both in the
USA and internationally, with audiences of all backgrounds.
We can see how that happened in part through the opening
credits. It is a simple set of still images with copy placed in
front of them that tells the audience who the actors are,
starting of course with Bill Cosby (whose character, despite
the name of the show, is actually named Cliff Huxtable). The
narrative takes place through a series of successive images, in
which a family gets out of a van and plays baseball, basket-
ball, and other games in the park. This is the most “Ameri-
can” of American images: a park, a happy family, a van, and
baseball. At the end we discover that these images are all
memorialized in photographs placed in a photo album. Sig-
nifiers of race and class other than “middle class” (sweaters,
jeans, and other simple clothing) are removed entirely, as
is any conflict. In other words, it is the perfect image of a
post-racial society in which race, class, and other categories
no longer seem to matter.
5. Advertising: Flaws with textual analysis often start with lim-
ited analyses, letting the ideology of the research guide the
data rather than allowing the data to speak for itself. In one
infamous ad, the fashion brand Dolce & Gabbana released a
controversial image that many claimed depicted gang rape.

Indeed it could be read that way. One man with no shirt on
is holding a woman down on the ground, while three other
men stand watching (two of which are at least partially shirt-
less). That sounds bad. But there are other signifiers as well.
They are standing in a place that doesn’t really exist, with a
sky full of clouds that reaches to the horizon, water that is
somehow in the scene as if they’re floating on an ocean, and
huge slabs of steel or concrete that make up the landscape.
Even the men aren’t all on the same plane; one of them
actually seems to have his legs end impossibly halfway in the
ground. So this is a fantasy. But whose fantasy? The men are
mostly clad in tight denim (except for one in a frilly purple
dress shirt), and their skin is shaved hairless and covered in
shiny oil. They are all both thin and extremely muscular. The
woman is on the ground, but her pelvis thrusts upwards, her
face does not look at all unhappy, she is wearing what seems
to be a satin version of a one-piece swimsuit with strappy
four-inch high heels, and her makeup and hair are perfect.
So is this the fantasy of men who are coded as gay (and gay
men are certainly one of the target demographics for this
company)? Is it the fantasy of a woman coded as wealthy
(the other target demographic)? Or is this the fantasy of one
of these men? If it’s his, is the man interested in the woman,
or the other men? Is it of abuse, and if so why are her hands
not clenched, her arms not straining, and her pelvis lifted up
to meet him? Is she “going along with it,” or is she an active
subject rather than an object? As with any analysis, incom-
plete analysis is bad research and leads to bad conclusions.
Whatever the interpretation is, make sure it deals with all
the signifiers available.
6. Situations: Textual analysis can be very useful as a way
to understand and compare various interpersonal inter-
actions. For instance, you could look at how staff at the
DMV treat people that come to the counter (like the

sloth scene in Zootopia), or how people communicate
when they match on Tinder. One of our favorite exam-
ples is introductions. If you compare how people intro-
duce themselves to each other in different cultures, you
can tell a lot about what that culture values and how.
For instance, in Japanese you say your place of employ-
ment first, then your family name, and finally your per-
sonal name. In the United States, however, people will
usually just give their first name, only in certain business
situations also adding your last name and your company.
Unlike in Japan, you as an individual are the priority, not
your family or the company.
IRL: WHEN WOULD YOU USE THIS?
Textual analysis is one of the most common tools in busi-

ness across all industries, whether that is understanding
what another team is saying in negotiation through body
language versus verbal signifiers, analyzing a sales bro-
chure to make sure it accurately and effectively portrays
a product or brand, or analyzing a movie trailer to make
sure it does the film justice and doesn’t needlessly harm
or offend an audience. Similarly, listing a house on the
real estate market requires knowing how to portray that
house so that the intended market will be able to imag-
ine themselves living there, and asking for a higher salary
depends on correct interpretations of your boss’ state of
mind, goals, and feelings about your performance. Each
of these things draws in large part from textual analysis.
Ethnography uses many of the same techniques as textual

analysis, although it includes other elements like being fully
embedded in a community and getting consent from that com-
munity. This method has grown to emphasize what the anthro-
pologist Clifford Geertz called “thick description” of an event,
society, people, and so on. There are many fantastic examples of
powerful ethnography, some of communities like street gangs
and some of scholars themselves (e.g., experiences with their
own alcoholism, in what we call “auto-ethnography”). One of
the most famous examples is a 2001 book from Barbara Ehren-
reich called Nickel and Dimed, which describes how she lived as
a minimum wage worker to examine issues of systemic inequality
and access. We will discuss the method a bit more in Chapter 7,
but it is important to understand that while some of the tech-
niques are similar, ethnography is centered on people and social
interaction, whereas textual analysis emphasizes the text itself.
For further reading
Geertz, C. (1973). The interpretation of cultures: Selected essays.

New York, NY: Basic Books.
Ehrenreich, B. (2001). Nickel and dimed: On (not) getting by in

America (1st ed.). New York, NY: Metropolitan Books.
Content and
CHAPTER
6 discourse
analysis
This chapter builds on Chapter 5 to show you what hap-

pens when you want to analyze a much larger number of
texts. Studying a large number of texts would take you
forever in textual analysis, but content analysis gives you
a systematic, repeatable, and more easily correlated set of
data to draw from. It’s one thing to watch one episode
of a TV show or one “State of the Union” speech. It’s
another to watch twenty of the same type of text and see
how they’re similar or different from one another. Con-
tent analysis is great because it is both a qualitative (type)
and quantitative (number) method, and you can choose
how far you want it to be on either side depending on
your goals. Here we’ll give you some suggestions on how
to choose a sample and then “code” your data, making it
easier on yourself.
OK, let’s take a step back. Textual analysis is awesome, but it’s
also limited. When you look at that many signifiers, how do you
know what’s important? How do you know that what you’re
reading into the text has broader significance beyond the text
(e.g., to other advertisements by the same company) or con-
nects to audiences the same way (e.g., what is it from the viral
video that made it viral, and was that a good thing or a bad
thing)? Those kinds of questions really need a higher sample size
(N), and while this will sacrifice some of the depth, mini-textual
analyses of specific examples can often be a useful way to incor-
porate depth into other, broader methods like content analysis
or discourse analysis.
Content analysis
The first of these methods, content analysis, tends to lie some-

where in between a qualitative (studying the qualities of texts)
and quantitative (counting something in or across texts) method.
In this case, you look at a series of texts that are related in some
way. You want them to be as similar as possible, so you might
look at a series of advertisements in the June issues of five fash-
ion magazines, or all speeches from a certain president, or a large
set of poems about love, or the Corporate Social Responsibility
reports from leaders of a certain industry. Similarity is important
because you’re going to ask a series of questions about each text
that you can answer quickly and easily. We call those questions
“codes,” and we’ll use that term a lot. Codes could be “how
many people appear in this advertisement?” or “Is the person
in the image making eye contact with the viewer?” or “What
percentage of the report is on technology?” When you’re cod-
ing, it helps to have a spreadsheet where you can put the codes
as columns and the texts as rows, and then quickly go through
each text. Hopefully you can have most codes answered as, for
6: Content and discourse analysis 83
instance, yes/no, a number, or a percentage. You’re trying to

reduce ambiguity. However, in some cases you are looking more
for qualitative data, in which case you might code for “empha-
sis” rather than presence or absence. In that case, you might rate
things on a scale from 0 (not present) to 3 (highly emphasized)
or in categories like ones used to describe social media posts
for a nonprofit organization (event promotion, story of com-
munity impact, request for donations, announcement of volun-
teer opportunity, etc.). As we’ll mention later in Chapter 7, it’s
sometimes helpful to look for elements that occur multiple times
within a text, that occur strongly, that occur across texts, or that
seem surprising in comparison to other texts.
This might seem overwhelming, but it’s actually a very effi-
cient way to go about analyzing a lot of data. It’s systematic
so it can be reproduced, you can easily add to your sample of
texts, and you can find similarities and differences (as well as
connected or correlated data) quickly. One thing to remem-
ber is that you’re most likely going to want to add mini-textual
analyses to your data. While numbers are all well and good, it’s
important in most cases to give some examples that show your
readers what you’re talking about. For instance, doing a con-
tent analysis of all speeches given by Barack Obama before he
was elected and then while he was president is going to give
you some cool numbers (e.g., the percentage of words and/
or time that he devotes to reaching out to people that disagree,
the percentage that he devotes to reaching out to those that
agree, and the percentage that he devotes to bringing the two
sides together). However, it won’t give you the richness that a
couple of examples might offer, especially from famous speeches
that might extend the analysis (e.g., “Yes We Can” or “A More
Perfect Union”).
The important thing here is to draw a series of codes in the
same way that you might make categories for a textual anal-
ysis. For a nonprofit’s social media, for instance, you might
have categories like “aesthetics,” “storytelling,” and “call to

action” in addition to numerical data like reach (“likes”) and
engagement (“shares” and comments). Those categories would
then be broken down into many subcategories. Storytelling for
instance could have a code for voice (Who seems to be speak-
ing? The organization, a volunteer, or a staff member?), a code
for content (What’s the post about?), the intended audience
(donors, volunteers, clients), the call to action (donations, an
event announcement, a thank you to a donor or volunteer), and
other similar concepts. These are all answered either as yes/no
questions with a code for each, or as a list of questions where
you can answer from a set of predefined answers that you build
by doing a quick run-through of your sample. Similarly, aesthet-
ics could focus on whether there is media present and what kind
(video, live video, still image), how much copy is present (words
and/or characters), how professional the media is (1 for not
professional, 2 for not bad, 3 for very professional), and whether
the colors suit the brand’s standards. The list is potentially end-
less. However, you need to start with your research questions.
Once you know what you want to find out, you can determine
what codes make the most sense for you to look for. You’ll prob-
ably end up with a fairly long list (probably closer to thirty than
ten!), but the longer the list, the easier the process will be later
on. We suggest doing this in a spreadsheet program like Micro-
soft Excel where you can easily compile or average data, add or
remove rows, highlight cells, and make tables and graphs so you
can visualize your results.
One of the best ways to approach a content analysis is in
groups, where you and at least one other person can come up
with the list of codes (the “coding scheme”) together, and then
you can do the analysis (“rate” the texts) independently. Before
you’ve done the ratings you can also code a few texts together in
what we call a “norming session” so you’re sure everyone agrees
on what each code means and how to interpret the categories.
After you’ve rated them, you can then calculate what we call
“inter-rater reliability,” which basically means you’re figuring
out how close you were in the analysis. Using Excel or SPSS, you
simply tabulate each of the coders scores in adjacent columns
and run a Cohen’s Kappa (two raters) or Fleiss’ Kappa (3+) anal-
ysis to determine the consensus (or lack thereof) between the
coders. If you score 82%–100% reliability (k =.90) your coders
are probably sharing a brain. Strong reliability is within 64%–
83% (k =.80 –.90). Much less than that, and you will want to
consider having another norming session. This type of analysis
allows you to see how subjective your material is and whether or
not the best way to describe it is through quantitative data, or if
it only allows you to see general trends that you’ll then have to
spend more time explaining in deeper, qualitative descriptions.
See our online resources for some step-by-step guides on how
to do this in Excel.
IRL: BREAKING DOWN

THE COMPETITION
One of the best uses for a content analysis is to compare

your organization or client to similar examples (including
competitors). You could even do this for yourself when
you’re looking for a job, or on cars before you buy one.
This is an incredibly valuable skill, and it is a quality that
distinguishes good managers from bad ones. In his work
as a marketing consultant, for instance, Ted often com-
pares what a company offers to what competitors offer
in terms of specific customers. He draws codes from best
practices in marketing, but he also draws codes from what
those customers want (to see, hear, etc.) drawn from other
studies and from interviews or focus groups he has con-
ducted. As a health management consultant, Josh com-
pares the various interactions patients have with doctors,
nurses, and other staff across not just the one hospital he’s
working with, but also other hospitals. He codes for best
practices to increase satisfaction and outcomes found in
the literature on patient-provider interaction. Maridath
works with multiple library databases to determine what
they offer for their prices, coding for not just what they
have and how they work, but also what the university fac-
ulty and students need to access, and how much they over-
lap with the existing content the library offers. These are
all content analyses, even if they aren’t always done in the
same systematic way that we have advocated here.
Examples of content analysis codes always depend on your

research questions, but a few examples can illustrate how it
might work.
1. Advertising: Say you want to compare how gender appears

in multiple forms of advertisements, not just the Dolce &
Gabbana example we discussed above. In that case you
could choose every ad released from Dolce & Gabbana
(and perhaps a competitor or two) in the past year or several
years, or perhaps every ad in certain magazines or websites
within a specific timeframe. The first approach would show
you how Dolce & Gabbana and/or its competitors con-
struct their advertisements through the use of gendered or
sexed individuals. The second approach would allow you to
compare how different outlets construct gender and sexu-

ality, and connect them to their “media kits” or “press kits”
that tell you who they’re targeting and how (or simply infer
audience from their articles). Codes would be drawn from
a few textual analyses showing us what elements might be
important, like various ways in which men and/or women
might appear. If you’re asking a different question, those
codes might be less relevant, but in this case, they’re essen-
tial. For instance, if you’re studying the wedding indus-
try, you might want to look at how De Beers constructs
diamonds through their famous campaigns starting with
“A Diamond is Forever,” and look at how the situation
constructs gender through wealth and social class. Different
questions, but same method.
2. Television: If you are analyzing The Cosby Show, you might
want to look at how sitcoms as a whole construct identity
over time. In the first, you’d use the textual analysis we did
above to form the relevant codes, expanding as you go. You
could choose every sitcom’s opening credit sequence over a
set period of time, coding for things like the role images of
the city might play in the narrative (something you’ll find a
lot in credit sequences), the impact of music (e.g., essential
throughout, for instance in Diff’rent Strokes, The Jeffersons,
or The Fresh Prince of Bel-Air), and the visual appearance
of each character. You could also do a more quantitative
analysis looking at numbers of people that appear or are
coded as different races, and see how often and when we
see integrated or segregated content. In gender studies, for
instance, we have what’s called the Bechdel Test, named
for comic artist Alison Bechdel. With the Bechdel Test,
scholars code films or television shows for one simple cri-
teria: whether two women ever speak together about any-
thing other than a man. This is a search for gender portrayal
and equality in the media, and many subsequent scholars
have expanded this to other criteria to show further levels

of depth (e.g., whether the women are named or are sex-
ualized). Scholars have also often used content analysis to
look at how children’s programming constructs, silences, or
erases history and identity, particularly of girls, non-White
ethnicities, and the economically disadvantaged.
3. Social media: Social media is usually produced in large
quantities over time, so it’s an ideal candidate for a content
analysis. To stick with our fashion theme, one way to use
social media in a content analysis is to look at how fash-
ion “influencers” (people that are followed for their style
or perspectives on fashion) portray themselves on Insta-
gram. You would then choose perhaps ten individuals, per-
haps five celebrities and five non-celebrities, and code their
past twenty-five Instagram posts. This is important because
then you’re coding a variety of people, but also crucially
looking at all posts, not just ones that are about fashion.
That would allow you to determine whether something else
makes them an influencer, like sexuality, images of food and
working out at the gym, or even “authenticity” like pho-
tos of pets or casual days at home with their family. Taylor
Swift, for instance, made a name for herself in large part on
the perception of her authenticity on social media, and that
authenticity itself is something to be studied since it encour-
aged what we call “parasocial relationships” with her fans
(that is, the imagined sense that they “know” her).
4. People: One of the ways in which content analysis can be
used to study people is through conversation analysis (some-
times called “discourse analysis,” although we use that term
in a different way). In this case, you might study how peo-
ple speak about fashion in a clothing store, and code for
differences depending on the people involved (e.g., eth-
nicity, male and/or female, customer and/or salesperson,
and every combination thereof). Do people mention body
weight? Sexiness? An event, a romantic partner, a desired

“feel?” While this might sound a bit like ethnography, and
it is in fact similar, the difference here is how you’re going
about looking for information and then processing that
information.
SAMPLE CONTENT ANALYSIS

CODING SCHEME
Focus: Food bank Facebook posts

Target of Analysis: Twenty most recent posts
Food bank name
Date of post
Time of post
Hashtags (#)
Hashtags (list)
Length of post (words)
Link? (presence/absence)
To what? (category: blog post, external news, donate
page, volunteer signup page, other)
Media? (presence/absence)
Image? (presence/absence)
Of what? (category: facility, volunteer packing food,
volunteer serving food, employee, client, client eat-
ing food, client working, child, child eating food,
child working, food, boxes of food, other)
Tone? (category: funny, cute, sad, inspirational,
informational, other)
Aesthetic quality (emphasis: 1 = low, 2 = medium,
3 = high)
Graphic? (presence/absence)
Of what? (category: map, infographic, drawing/ani-
mation, event marketing, other)
Tone? (category: informational, funny/cute, inspira-
tional, other)
3 = high)
Video? (presence/absence)
Of what? (category: donation process, sourcing pro-
cess, client profile, inspiration, program informa-
tion, impact information, volunteer profile, donor
profile, employee profile, other)
Tone? (category: sad, happy, inspirational/hopeful,
angry, detached, other)
3 = high)
Length? (minutes)
Clear call to action? (presence/absence)
To do what? (volunteer, donate, attend an event, other)
Critical discourse analysis
While content analysis is a useful tool, it is not as useful for under-

standing the broader context surrounding an issue or event. In
other words, content analysis is very useful for studying the same
type of text with a limited timeframe or sample, but it’s not as
good at seeing how the issues themselves are socially determined
(e.g., what “fashion” means, what “woman” or “girl” means,
what “Black” means, etc.). Usually when we think of those ques-
tions we turn to historians, who often search through archives of
material compiled by individuals, universities, museums, librar-

ies, and other sources. Historians are concerned not just with
what happened, but what people said about what happened and
how different people portray or construct the event. Similarly,
political scientists, environmental policy scholars, law scholars,
and other researchers will often look at specific archives of mate-
rials to determine the history of a law, policy, or tradition, and
how it has been applied or constructed over time. Often there are
huge archives of materials, including letters, drafts of documents,
and other related items that can provide crucial insights into the
history of a time or place. Archives of city planning maps and
development approvals can tell you about the changes in a town
and its connections to racial or ethnic identity (e.g., gentrifica-
tion). Presidential libraries can explain how a certain decision was
made, and what advice was given during debates around pivotal
events (e.g., the Cuban Missile Crisis). Art libraries can provide
sketches and images of studies that reveal the origins and possible
other end results for a film, a photograph, or a painting.
However, when scholars want to discuss something perhaps
more current than can be found in an archive, they often move
to “historiography,” or writing about the writing about history.
The reason for this is that if you’re looking for something that
might have been silenced, it makes sense to feel around the issue
rather than just look at the issue itself. If you go back to the
“blind men and the elephant” story from Chapter 3, in which
five blind men each touch a different part of an elephant and
make claims as to what it is (the trunk is a snake, the leg is a
tree, etc.), historiographers often look at how we can get a more
complete picture of the puzzle, perhaps by talking to each of
the men rather than trying to describe “reality” itself. Imagine
trying to explain something invisible; you might describe what
it does or has done, or how others react to it. That’s what we’re
calling “post-structural discourse analysis,” or “critical discourse
analysis.” Basically that just means that we’re looking at how a
social, cultural conversation builds and maintains a concept with-

out necessarily talking about it directly. For instance, changing
“global warming” to “climate change” or “friendly fire” to “col-
lateral damage” doesn’t alter what you’re talking about, but at
the same time it does alter your perceptions. Changing the word
“truck” to the phrase “sport-utility vehicle” to “SUV” also
changes things, as does “station wagon” to “crossover” or “used
car” to “pre-owned car.” In other words, we’re not talking about
reality, we’re talking about how we imagine and construct what
we think of as reality. The most famous discourse analysis scholar
is the philosopher, historian, and psychologist Michel Foucault.
He studied not madness itself, but madness as constructed by civ-
ilization, and did the same with prisons and punishment, sexual-
ity, and even history itself. After all, what we define as “crazy” or
“criminal” or “a genius” has changed radically over time, and the
story of who is crazy (or gets to be crazy) is tied to other issues of
class, gender, race, and privilege of all sorts.

Using the archive and others’ archives
There comes a point where you might need to consult some

primary sources for your research. What do we mean by
primary sources? For starters, this definition can get quite
slippery, and like a lot of other definitions in this book it’s
going to vary by discipline. Here, what we mean is orig-
inal physical (or digital) documents created by members
connected to the topic or event you are researching. These
archives usually live in your college or university library, and

many also live online. For the physical archives, you might
need to make an appointment with the librarian (usually
called an archivist) to view the collections, usually during
specific hours of the day. Because the materials are often
rare and fragile there are often rules that must be followed
to gain access. So what might you find in the archives? For
example, if you were researching a famous painter or musi-
cian or other figure, their archive might include items such
as correspondence (letters), journals, and items archivists
called “ephemera” – which is literally “stuff.” These can
include the sketches that then became the paintings, drafts
of speeches, even letters to mistresses! Like your secondary
research, and your selection of methods, it’s a good idea to
consult several archives’ materials on one person or topic
to avoid the dreaded cherry-picking. Your librarian can
help put you in touch with your college’s archivist, and can
even recommend archives at other museums or universities
that you might consult.
Discourse analysis derives from a mixture of fields like

anthropology, sociology, political science, and history. The
focus is not only that language constructs our reality, but also
that the construction is an active and constantly evolving pro-
cess, a social practice of sorts. That means that it often looks at
conflict, and tries to understand what’s going on from the dif-
ferences that might exist (or the silences that exist) within var-
ious forms of social discourse. It also often draws its analysis of
these competing forms of reality or history from a large num-
ber of texts that come from different types of media, different
sources, etc. You will most likely code for themes just like you
would in a content analysis, but probably not at the same level
of detail because you are looking at longer and/or a larger
number of texts. The key here is that you’re not studying the
texts themselves, you’re studying what they’re saying about
something else. You’re still looking for patterns, though, or
what discourse analysts often call “interpretive repertoires.”
There are a few things you can focus on: outliers, or what
have been called “deviant” cases; consistency and coherence
with previous work on the topic (e.g., your literature review);
understandings of an event or issue (e.g., letters to the editor,
movie reviews, advice columns). You can also simply present
your data and ask readers to come to their own conclusions.
Here are some examples of discourse that might help you get
a feel for it:
1. Business: One of the most interesting ways to use a dis-

course analysis is to look at how the media, employees, ana-
lysts, and internet commentators might construct an issue
or company. One of the most famous examples in more
recent memory is the electric car pioneer Tesla, whose stock
is known for being unpredictable and volatile. It is guided
by comments from CEO Elon Musk, earnings reports, con-
jecture from analysts about suppliers and customers, com-
mentary from amateur investors on sites like Seeking Alpha,
media coverage of the company or CEO (even his love life
or use of recreational drugs), and public conversations with
investors. Each source has its own agenda and background,
and the concept of “Tesla” is completely different from the
company itself. In fact, this is true of almost everything in
business, since business is built on future opportunity as
much or more than past success.
2. History: Discourse analysis is used quite a bit in criti-
cal history studies. For instance, an analysis of textbooks
across the nation (or globe) could reveal quite a bit about
how various regions treat (or conceal, erase, and misrep-
resent) the Civil War, slavery, and indigenous populations.
Connecting that to records of school board meetings in
which boards, parents, and teachers meet, as well as the
presence or absence of discussions about this issue on the
news and social media, can provide a snapshot of how
the culture is actively constructing and silencing certain
histories.
3. Arts: Art isn’t just made of the artifacts themselves; in fact,
that is only a small part of what we know of as “art.” The
well-known, award winning, controversial Broadway musi-
cal Hamilton, for instance, could be studied in terms of
how news sources, professional and amateur reviewers, and
actors spoke about the play. This was particularly interest-
ing as actors came out actively to contest popular under-
standings of history, noting that their versions in the play
weren’t accurate, but that neither were the versions in his-
tory books. In other words, Hamilton and the discourse
around it was in many ways a struggle over who gets to
write history and how.
4. Brand perception: Airlines are an oligopoly of compa-
nies that everyone hates. No one likes to fly, but many
of us have to put ourselves through the misery of air-
ports, security lines, immigration, and tiny seats just to
save some time. They’re all around the same price, with
mostly the same service. But perceptions of the airline
can make the difference, as United Airlines has found
out repeatedly. Internet sensations such as the “United
Breaks Guitars” song from Dave Carroll or the infamous
video of security personnel beating a passenger as they
dragged him off a plane in order to give his seat to an
airline staff member, have caused incredible damage to
the brand. To fix it, any analyst would need to look
closely at the news coverage around the events, but also

social media responses to the various videos, memes that
are being circulated, and ways in which other airlines
like Southwest capitalized on the brand’s failure to deal
with the crisis effectively.
Coding a discourse analysis is similar to the process for cod-

ing a content analysis, but it’s rare to get yes/no categories
here, and you’re usually not comparing “apples to apples,”
but rather different media, times, or channels of communi-
cation. For instance, analyzing political speech could look
at speeches, YouTube comments, news coverage, and opin-
ion-editorials around a single event. What you’re usually
coding for is particular categories of discourse and, perhaps,
content like tone and rhetoric. In this example you might
look for categories like “approval/disapproval” or “positive/
negative,” but you could also look for references to other
events or political figures, inclusion of data, references to
other news sources or studies, or rhetorical devices like fear
appeals, emphasis on hope, or discussion of collaboration.
Numbers here could portray the percentage of comments
online that say one thing versus another, but also percent-
ages of speeches and news articles devoted to various issues
and perspectives.
Discourse analysis is a challenge because it introduces much
of the complexity and depth of a textual analysis with the large
number of texts in a content analysis (even more, in many cases).
However, the key here is that you aren’t actually doing thick
description, you’re simply looking at how prevalent or intense
certain kinds of discourse are in various media. In other words,
you aren’t looking at the true depth of a text or set of texts,
you’re looking at how those texts frame an issue. The issue is
what is central here, as well as ways of looking at/speaking about
or erasing/concealing that issue. That makes it an incredibly
powerful research tool.
SAMPLE DISCOURSE ANALYSIS

CODING SCHEME
Focus: Reviews of Will and Grace 1998 original and

2017 reboot
Target of Analysis: Twenty newspaper articles for origi-
nal and twenty for reboot
Source
Author
Date
URL
Length (words)
General response
Positive (yes/no)
Negative (yes/no)
Neutral/mixed (yes/no)
Funny? (yes/no)
Predictable? (yes/no)
Nostalgic? (yes/no)
Controversial? (yes/no)
Discussion of positive portrayal of homosexuality (yes/no)
Emphasis on positive (words)
Emphasis on political/social change (yes/no)
Discussion of negative portrayal of homosexuality (yes/no)
Emphasis on negative (words)
Discussion of homosexual content as deviant, weird, etc.
(yes/no)
Emphasis of “weirdness” etc. (words)
Discussion of impact on viewers? (yes/no)
Relevant to American culture? (yes/no)
Reboot different from original? (yes/no)
Ethnography,
CHAPTER
7 interviews, and
focus groups
Unlike text-based analysis, which is centered on what a

text itself is doing, the methods in this chapter are all con-
cerned primarily with human interaction, perception, and
reception. Ethnography is a kind of “participant observa-
tion” that embeds you within a social group; interviews
are deep discussions with a strategic sample of individu-
als, and focus groups are broader conversations with small
groups. In each case, while the data you collect is also
analyzed and coded, what you’re looking for is how peo-
ple interpret their reality, how they tell stories about that
reality, and how they create knowledge through actions,
rituals, and social systems.
7: Ethnography, interviews, focus groups 99
Let’s refresh. The reason we do research is to capture and report

knowledge, or the “truth,” about something. For some, the truth
is a single thing that is out there to be discovered, regardless of
how an individual perceives it. For example, if a doctor is trying
to locate the source of your headache, they could make a decision
based on fMRI scans rather than your description. You probably
aren’t a doctor, after all, and you might thus mislead them (as
Dr. House would say, “people lie”). The interpretivist researcher,
however, sees truth as subjective depending on who is observing
and who is being observed. “Truth” then can be best understood
as someone’s perception, often conveyed through conversation.
So, after visiting the medical doctor you may also visit a psychol-
ogist, who diagnoses that the source of the headache as stress at
work. You get some therapy to avoid brain surgery. Both health-
care providers were able to locate a “truth” about your headache,
but in the second case that truth comes from the messiness and
complexity of life. Interpretive research comes from the belief
that really interesting “truths” are as much in our heads as in the
world outside, and the research goals focus on how we perceive
and receive reality. Three of the most common ways of doing this
are through ethnography, IDIs, and focus groups.
Ethnography
One of the most interesting, but also most problematic, meth-

ods of research you can conduct is ethnography – from the
Greek ethnos (people) and grapho (I write), and meaning “writ-
ing about people.” In this method, you usually embed your-
self within a group and take notes (what we call “field work”)
about the people in that group. This is a kind of textual analysis,
but in this case you’re studying the details of people and their
interactions from within the social system itself. The process is
fairly simple. You choose a target group to observe, you join
that group in some way, and you observe them, taking “field
notes” along the way that include detailed information about
events, verbal and nonverbal communication, and other ele-
ments. These notes can then be coded in a similar way to textual
analysis or discourse analysis, which we discuss in depth in Chap-
ter 10. It’s important to remember that what you’re coding for
and how you’re coding for it can change, depending on your
questions, background and discipline, and what your literature
review has taught you. There are also three primary ways to
conduct ethnography: undercover ethnography, non-participant
observation, and participant observation.
Undercover ethnography
This method was often used to study groups from within as
a completely embedded participant. That means that research-
ers would claim to be a part of that group (like an undercover
police officer) in order to observe participants as organically as
possible. There are many reasons to do this. For instance, what if
you want to study prison culture? Gangs? Human trafficking or
organ trafficking? All of those have been the subjects of under-
cover ethnography. However, at the same time we have to be
very careful here. Because the participants can’t give consent if
you’re undercover, you can end up exploiting and misrepresent-
ing them. That means that undercover ethnography is rarely if
ever ethical as research. Leave it to the police or journalists, who
have a different job to do.
Undercover ethnography is also misleading, because it’s very
difficult to be completely organic as a part of a group if you’re not
actually a part of that group. That means that if you are separated
a bit through the act of observation and covert note-taking, you
might introduce elements of bias and expectation that can lead to
changes in the behavior of the people around you. For instance, if
you’re studying gang violence and are undercover, should you be
violent? And if you feel you need to be violent for “authenticity,”
does your act of violence influence others to be more or differ-

ently violent? The answer is usually yes, it does. In other words,
authenticity here is actually impossible, which means that being
undercover is often self-defeating. It can never be truly organic
and might skew your results even more towards what you think
a group should be doing, rather than what they would do if you
weren’t there.
Non-participant observation
In this method, you’ve received IRB approval and informed
consent from all participants (see the online resources for some
guides on how to do that successfully). However, you are
attempting as much as possible to not participate in the group’s
interactions. This could happen through video recording, but
there are challenges in terms of the depth and detail that can be
captured. Of course, being there and taking notes might also
reduce depth, since you won’t catch everything and might be
immersed in some detail or emotional moment (e.g., a riveting
story from a recovering alcoholic in an Alcoholics Anonymous
meeting) and miss something important (e.g., that half of the
room isn’t paying attention to that meeting, or that someone in
the back is shaking from alcohol withdrawal). The biggest chal-
lenge here, however, is that simply the act of observing some-
thing can change that thing (according to particle physics, it
always does). Certainly having an observer present during an AA
meeting could change the dynamic of that meeting, especially if
people are communicating trauma. Video recording the inter-
actions might even make the problem worse. Since researchers
are also given a kind of cultural power and authority, this impact
could be massive. In other words, it’s almost impossible to have
non-participant observation; the best you can get in most cases
is a limited form of participant observation. Imagine yourself
as a child playing with friends. Even if your parents aren’t in
the room, you know they’re there and often act accordingly. If
they’re in the room it might be worse, and certainly if you’re

being recorded it might be worse still. This is the same principle.
Participant observation
Participant observation basically means that you’ve gotten IRB
approval and informed consent, and you consider yourself an active
participant in the group you are observing. That means that you’ll
have to observe yourself as well, and of course in some cases like
auto-ethnography you’re focusing primarily on yourself. However,
in most cases you’re simply acknowledging the biases, changes, and
limitations to observing interactions in any organic or “authentic”
way. That means that your work will need to be deep enough to
acknowledge what is going on in what you’re observing, but also
engage how you might be impacting the events and interactions.
It’s a great way to try to understand group interactions, but chal-
lenging to do both ethically and in a way that provides truly useful
data. Combining ethnography with other methods can often be
a useful way to add elements that might be missed, especially if
it happens before and after your observations. Surveys are great
here, but so are the other two methods we’ll discuss in this chap-
ter: in-depth interviews and focus groups.
In-depth interviews
In-depth interviewing is an acquired skill. Many people think

they’re going to be good at it but often aren’t. Your first time
conducting an interview may not go as planned. Perhaps your
participant isn’t talking as much as you thought they were going
to. Maybe you’re going through your questions too fast. Maybe
you’re not pausing enough, or not long enough. Then suddenly
the interview is over, and you’re not sure you have anything
significant. Uh oh. Relax. Good interviewer skills take time and
practice to develop. Let’s discuss some important principles.
Bias
You have bias. Admit it, then try your best to avoid it. While
you can conduct an experiment behind a one-way mirror and
make every effort to remove your presence from the study,
as long as you’re studying it, the bias is still there. Qualita-
tive interviewers make no apologies about their involvement
and participation in co-creating and reporting on the truth
of people’s experiences. Remember, you have just developed
an extensive literature review, a research design, and a set of
research questions about your project. At this point, you are
knowledgeable and have a good idea about what people think/
feel about your topic. That is a good thing, but it can also be
a problem, since you might end up accidentally guiding the
research towards what you’re expecting to find. There are some
ways to reduce this effect though, and creating a good inter-
view guide is a good start.
IRL: REMOVING BIAS FROM

A JOB INTERVIEW
The literature shows us that people might not get jobs

even if they’re qualified, due to a variety of reasons ranging
from the color of their skin to the assumed ethnicity of their
name to assumptions about their socioeconomic class. Bias
could also be a result of the club or group they were a part
of in university, or the fact that they did or didn’t play a
certain sport. With this many variables, it’s challenging to
remove bias. Removing names from resumes is one possi-
bility, but how do you do that when applicants have to list
other information that includes their name, or when part
of what you’re evaluating is the creativity and design of

their resume? One of the best ways to eliminate bias is to
locate it and discuss it openly as an interviewer before look-
ing at any materials. Think about what might cause you to
be biased, why, and whether it actually should disqualify
someone for the job. That alone can make a huge impact.
In one fascinating example, orchestra administrators were
confused as to why there were so few women that made it
to higher positions. Some people said that men were just
better, others that they were more passionate musicians,
but most were skeptical of both explanations. So they did
an experiment in which they would start concealing the
musicians performing for juries (the groups of judges that
choose whether you get in or not) behind a screen. What
happened? A lot more women and people of color made it
through. Then they realized that juries could still assume a
gender by what they heard when the person came out on
stage, since high heels made a distinctive sound. They had
everyone wear slippers, and again more women made it
through. These juries didn’t think of themselves as sexist;
they just had preconceptions in their heads about what a
“genius” looked like, even though what they were sup-
posed to be measuring was how they sounded. This exper-
iment was an attempt to create a “double blind” interview,
where neither the participant (the musician) nor the judge
(interviewer) knew who the other was, what they looked
like, and so forth. This is one reason the peer-review pro-
cess for scholars is so important, as well. It shouldn’t mat-
ter who you are; it should only matter how good your
ideas, methods, data, and analyses are.
Interview guides
One of the first actions you can take to help curb your own bias
is to construct an interview guide, which is a prepared script with
a list of questions that you’ll use as you do the interview. There’s
a balancing act here between very scripted and very unscripted,
though. Do I write out every question and deliver it verbatim?
Is it okay to ask questions that aren’t on my script? Do I need
to stick with the same order each time? The best interviews will
often feel like a natural conversation, and at the same time, it’s
important to understand that structure is necessary to compare
responses and prevent leading questions.
In a “non-structured interview,” there is no interview guide
at all. Often corporate executives who have been conducting
employment interviews for several years will claim that they
need no prepared list of questions. They allow the interview to
flow like an organic conversation and make decisions based on
intuition. There are a few downsides to this interview approach:
1. It becomes difficult to compare responses across your popula-

tion if you ask wildly different questions between participants.
2. Interviewers who don’t use guides will subconsciously ask
more “leading questions,” which means that they uninten-
tionally guide participants’ responses, often because they
mean to establish “common ground.”
There are distinct differences between an unscripted conversa-

tion and a guided research interview. Conversations build rela-
tionships, and both individuals are seeking common ground
and areas of agreement. Non-structured or low-structured
interview guides can actually build a false sense of agreement,
however. The interviewer may unconsciously hint at their per-
sonal attitudes about a topic, which often results in influencing
(and perhaps changing) a participant’s response. Imagine you’re
interviewing for an internship, and your future boss says, “So,
it says here on your resume that you lived in Boston for a while.
I’m a huge Red Sox fan – are you?” Well, if you weren’t before
the question was asked, our bet is that you are now (and will go
home and burn your Yankees jersey).
On the other side of the continuum is the “highly-structured”
interview guide. In this case, interview questions are written and
delivered verbatim. We call this format “Stick to the Script,” and
it’s not unusual when there are many interviewers speaking with
many candidates (say, for entrance into college). You state them
exactly as they are written, in the exact order, with no deviation.
Comparing multiple responses is much easier to do in this case,
but there are some problems:
1. It stifles any opportunity to make the interview feel like a

natural conversation, which can increase the sense of ner-
vousness and even the interviewee’s desire to answer in a
way they feel is “right” rather than “true.”
2. It doesn’t allow for improvisation or for following unex-
pected pathways, which can remove important elements
that could impact your study or decision.
3. It doesn’t allow for a great deal of depth, which comes from
more natural interaction and storytelling.
We suggest that you aim for the middle ground, a “semi-

structured” interview guide. This approach is both natural and
structured. Participants should feel as though the conversation
you are having is unique. However, as a researcher you also
want to make sure you are rigorous and neutral in collecting
information, and so you have a pre-scripted list of questions and
follow-up questions that cover each topic. The conversation can
flow naturally and jump from question to question, and if some
responses seem both pertinent and interesting, you can ask par-
ticipants to “explain a bit more” or to “give an example.” Yes, it
takes a bit more skill to navigate a semi-structured than a highly
structured interview guide. Anyone can simply “Stick to the

Script.” But with time and practice, you become more comfort-
able with the balance of a natural yet structured conversation.
Questions
Like interview guides in general, the questions you ask also range
from open to closed. On the one hand, highly open questions give
the participant complete freedom to respond in any direction they
would like. Questions such as, “Can you tell me about yourself?”
and “What does it all mean?” can be useful sometimes, especially
if the participant starts going in a direction that you didn’t antic-
ipate; however, questions that are too open can often lead your
participant to go off on a tangent and take up way too much of the
time you’ve set for the interview. Remember that at some point
you’ve got to get to the questions you need answered!
On the other hand, highly closed questions are great for short
interviews, and like highly structured interview guides they make
responses easier to compare. Unfortunately this also means that
participants will often just try to answer your question, which
reduces the chances of unexpected directions and responses, or
even the richness of detail that interviews can provide. They can
also make the interview feel cold and distant, reducing the likeli-
hood that your participants will tell the truth or disclose informa-
tion that might be more controversial, personal, or uncomfortable.
For instance, a closed question might ask something like “did that
make you happy?” or “did you have a close relationship with your
parents?” but it might not encourage the participant to explain
why they felt a certain way, how they felt if they didn’t feel a certain
way, or examples of situations that shaped their feelings.
We suggest that the key for a useful in-depth interview is to
focus on having enough structure to facilitate getting to the
point, but that you also focus on having your participants tell
stories and go deeper (see Figure 7.1). The deeper you get into
their examples, the more data they show you rather than tell
you, the better. Good examples of moderately open questions

might be “tell me about any challenges you experienced during
a recent trip you took to the doctor,” or “explain how viewing
this advertisement makes you think and feel.” In other words,
if you’re conducting a qualitative in-depth interview, we advise
you stick to the semi-structured guide. Ground each of your
questions in a specific topic/issue to give your participant some
direction, and save the more closed, limited response questions
for surveys (which we’ll get to in the next chapter).
Highly open Moderately open Highly closed

questions questions questions
Strengths Your happy place Strengths
*Gives participant *Participant has *Saves time
completely both direction * Easier to
freedom to and freedom compare
respond in any in their response participant
way. *Allows for responses
participants’
Weaknesses Weaknesses
response
*Takes more time *Limits the
comparison
*Prone to leading freedom
*Scripted
questions and of the
questions
response bias participants’
prevent
response
interviewer bias
and prone to
selection bias
Example Example Example
What is health? What was your Were you
experience like satisfied or
at your last visit dissatisfied
to the doctor? with your last
physician visit?
Figure 7.1 Types of questions

Primary and secondary questions

Think of primary questions as the main points of an interview
guide outline. They introduce each distinct topic you hope to cover
in your interview. The secondary questions are usually follow-up
questions that will bring out more in-depth information. A good
semi-structured interview will have both primary and secondary
questions laid out. While the number varies depending on your
research questions and the time you have for the interview, we
suggest you prepare around ten to fifteen primary questions, with
one to two secondary questions each. This isn’t a random num-
ber; more than fifteen questions will most likely lead to “interview
fatigue,” which means that a tired participant will often give you
short responses because they are tired and want the interview to
end (or worse, they’ll answer in the way they think you want them
to answer to avoid follow-up questions).
In many ways, the interview questions are a conversational
guide aimed at addressing your research questions. So, you
might start by rewriting your research questions in ordinary lan-
guage your participants can relate to. For example, you could
change “What behaviors lead to high levels of communication
competence in healthcare?” to “Tell me about your doctor’s
communication style.” You can also draw from your literature
review, especially the theories that are guiding your research.
For instance, in the literature on disability, scholars often discuss
the ability for media to create empathy with viewers. Therefore
you could change a research question like “How do audiences
experience empathy with disabled characters” to “How did you
feel about Christopher? What made you relate to him, and what
was difficult to relate to?” This may take a few drafts, but make
a long list of primary questions that covers each topic of your
literature review and your research questions, then narrow it
down to ten to fifteen questions. Keep these questions moder-
ately open, and then develop and write out one to two second-
ary questions to gain more detailed information.
Primary question: Secondary questions:
What communication 1. If you can, please describe

challenges, if any, have a story that best illustrates
you experienced with your this. (Ask for story.)
physician? 2. What about this
experience made it so
challenging? (Probe for
more info.)
3. Is there anything else
about the challenges
that you can think of
that you have not yet
told me? (Clearinghouse
probe.)
Figure 7.2 Primary and secondary questions
How do you “follow up” on information you don’t have yet?

This actually isn’t that hard, especially if you have a good lit-
erature review. Ultimately, secondary questions clarify and/or
gain more information. If you have an introspective, talkative
participant, you may rarely need to use them, but always have
them on your interview guide and ready to go just in case you
get someone that is shy, somewhat hostile, or just doesn’t know
what to say because they’re uncomfortable with the situation.
Regardless of how your participant responds, you can always
ask for a “story or experience” that grounds responses in a real-
life event. You can also probe for more information by asking
what specifically made it so challenging, or you can always throw
in a “clearinghouse probe” at any point in the interview (espe-
cially at the end) which simply gives the participant a second
to think about “anything else” they might want to say. Keep
in mind that some of these issues might not be at the forefront
of their minds. When a participant spends time thinking and

talking about life experiences, new examples and stories might
surface from prior questions. The clearinghouse probe is a great
way of collecting those accounts.
There are a few problems you can run into when designing
your questions, and we want to run you through those so you
can avoid them.
1. The “awkward pause”: This is when you ask your partic-

ipant a moderately open question then, quickly, follow up
with a bipolar question. For example, you might ask your
participant to explain how they felt about their physician’s
communication at their last visit. After you ask the question,
though, your participant pauses and you can sense that they
are struggling to find the right words. It’s ok to let them
think about this for a while. Don’t be afraid of seemingly
long pauses (and yes, the pauses seem much longer to you
than they actually are). This means you have asked a good
question that requires some thought.
2. The “bipolar trap”: This is when you jump into a pause (or
don’t even wait for one) before attempting to “save” your
participant by interjecting: “So, would you say you were
pleased with the communication with your doctor?” Your
participant looks relieved and responds “Yeah . . . pretty
much.” What could have been an in-depth response about
the nuances and complexities of your participant’s commu-
nication experience has now been reduced to a simple yes/
no response.
3. Leading questions: Are all of your questions neutral? Have
you removed all subtle hints about how you may want par-
ticipants to respond (and have you thought enough about
what you want or are biased to want)? Remember that an
interviewer has a great deal of power over the interviewee;
often without realizing it, social desirability bias can make
an interviewee give the answer they think you want instead

of how they truly feel. The fact is that people who are being
questioned often want to please the interviewer and give
the “correct” answer. As you read through your questions,
ask yourself: Do any of my questions seem leading in one
direction or the other? In asking the question this way, does
it seem like I want them to respond in favor or in direct
opposition? An example might be something like “It must
be difficult communicating with a physician who focuses on
their computer during your consultation, right?” You can
rephrase that to remove the leading elements: “What chal-
lenges, if any at all, have you experienced in communicating
with your physician?” is much better.
It’s not just problems you should watch out for. You should also
remember that there are a few things that can make things a lot
easier on you, saving you time or even saving an interview from
being unusable. Here are a few:
1. Rehearse: Once you have written your interview guide, be

certain that you test it by reading it out loud to a friend
that doesn’t know anything about the topic. The questions
may sound great and make perfect sense to you (because
you wrote them), but the phrasing, word choice, and intent
may be confusing to other people, especially those who hav-
en’t done the research and so might not know what certain
terms mean. Always practice the entire interview before you
begin interviewing research participants. If there is any con-
fusion, rephrase the questions. Make it as simple and clear
as you can.
2. Recruiting: We’ll spend some more time discussing sam-
pling methods in Chapter 8, but here we want to talk
briefly about how you might recruit participants for your
in-depth interviews. In most cases you’re probably looking
for a specific population with specific criteria. For example,

if I want to find out how families of those who suffer from
alcoholism cope with their situation, I would probably look
for an Al-Anon group. Selecting people like this, from a spe-
cific population, is called “non-probability sampling,” since
it’s not a random sample of everyone. If you’re looking at
an Al-Anon that is local, or at students from your university,
you further restrict your sample to “convenience sampling”
(a sample that’s easy to get). “Snowball sampling” is use-
ful when it’s difficult to find established groups. With this
method, you find one or two participants and ask them to
connect you with others in their community that might also
be good participants, and then you ask those to connect
you, and so forth.
3. Willingness: In order to get permission to interview peo-
ple in the group, you will need to draft both a recruitment
announcement and a consent form. You’ll need to run these
through your school’s IRB, and then send the announce-
ment to a director within the group who will either post or
send it to members. Remember, you need and want will-
ing participants. Once you have a list of willing participants,
find a neutral setting that won’t alter responses and get their
written consent before starting.
4. Recording the interview: Don’t assume someone is will-
ing to be recorded without asking them first, because it’s
unethical and might also be illegal. You need, in writing, a
signed document from your participant that they are abso-
lutely okay with you asking personal questions about the
issues related to your research, recording those responses
(via notes, audio, and/or video), and perhaps using their
accounts as excerpts in your project. When they responded
to your announcement, after all, they may not have realized
what the process entails, and the might be uncomfortable
with it (especially audio and video recording). Before you
begin the interview, you need to let them know all of this
through a prepared written statement about your project.
Let them know that:
1. You will be recording the interview, transcribing it, and
removing all personal names and places, even though
you may use excerpts in your project.
2. Some of the questions may be personal in nature – and
at any time they can choose not to respond.
3. At any time, you will stop the recording if they ask you to.
4. At any time, they can leave with no hard feelings.
5. If the interview stirs up any unanticipated emotions, you
have contact information for counselors if necessary.
Put all of these statements in your written consent form,
read it aloud to them, and then ask them to sign it if they
are still willing to participate. If they say consent, then you
can begin recording. Small digital recorders are fairly easy
to find, and your phone can do this too (just make sure you
turn it to “silent”). Please record through two devices, and
test them first to make sure the audio is usable. Trust us. Let
them know when you begin recording, place the recorder(s)
on the table, and begin your questions.
5. Take field notes: Focus on the questions, creating a natural
dialogue, and record any notes you think might be neces-
sary. We like to write down the setting, time of day, length
of the interview, and any nonverbal elements that you think
might be significant. For instance, if there was an emotional
response to one of the questions or a noticeable pause, write
it down. You might find that over half of your participants
became emotional over one of the questions, and if so it’s a
good thing you noted it from the start. In the end, you’re
not sure what will and won’t have a significant impact on
your findings. To be safe, write down more than you think
you might need.
6. Transcribe: The audio recorder frees you from the bur-

den of writing down every response. Later you’ll transcribe
the interview verbatim (yep, every single word) so there’s
no need to worry about that now. Transcribing in-depth
interviews might be the most painful task of any research
method. You can outsource this job to one of the transcrip-
tion services we’ve included in the online resources to this
chapter, which range from professional human transcrib-
ers to programs for your desktop that let you slow down
the speed of the file to digital text-to-speech programs.
Depending on how fast you type, transcribing an interview
takes around two to three times as long as the actual time of
the interview (that means twenty one-hour interviews could
take you easily between forty and sixty hours!). Transcribing
your interviews will make you more familiar with their data
and nuances, though, so we suggest you transcribe at least
one personally.
Since you have a semi-structured list of questions, it’s important

that you maintain a natural conversation feel with your body
language. Your posture should be engaged and inviting. You
want your participant to feel comfortable opening up to you,
while at the same time, not showing bias by rewarding “good
responses” with overly affirmative reactions. After the interview
is over, thank your participant and let them know that their
story is valuable. Remind your participant that the interview has
added great value to your research as it will add necessary insight
about the topics you are studying.
In-depth interviews are a great way to get to know the details
of your project and your participants, and it can reveal a ton of
information you never expected or even thought to look for.
What they don’t do, however, is show you how a group of people
might interact with your project. Interviews are therefore a lit-
tle artificial and more focused on past, remembered experiences
rather than experiences as they happen or are negotiated within

a social setting. If you want to get more social information,
you’ll need to conduct focus groups.
Focus groups
The process for creating focus groups is fairly similar to that for
creating interviews, but sometimes it gets a little more challeng-
ing because you’re trying to work around many people’s sched-
ules. We like to use online signups like WeJoinIn or Doodle for
these, depending on who we’re including in the focus group.
Both services allow you to make meetings easily at a time that’s
convenient for everyone.
In many cases, you’ll estimate around four to five people per
focus group, although many use much larger numbers. The rea-
son you want to keep it small is to reduce issues of groupthink
(when people start to fall in line with each other in order to
preserve a sense of community) and power (when one person
dominates a conversation). On the other hand, sometimes those
interactions give you valuable insight as to how discourse plays
out, and you can get a feeling for people’s responses in post-focus
group surveys or interviews if you’re worried about people that
didn’t speak as much. The goal, however, is to have a conversa-
tion. Interviews are often about deep individual storytelling. In
focus groups you’re trying to learn more about how participants
relate to one another, and how they can learn from and bounce
off of each other, than you are about individual responses.
Focus groups use interview guides in a similar way to inter-
views, but you’re more a moderator than an interviewer in this
case. Often researchers will bring in media such as magazines
for participants to flip through or show short videos and ask
for responses/discussion. This allows you to see how the vari-
ous groups relate to specific content, and how they figure out
differences in interpretation. Because of the challenges with

equal time or groupthink, if you have specific questions you
need answered across the board, pre-focus group and post-focus
group surveys or short interviews are often very useful.
Final thoughts
In each of the methods in this chapter, remember you’re per-

forming a kind of “participant observation.” Basically, in this
case it’s important because you should recognize that your pres-
ence itself will modify what is said and how. If you’re speaking
about racial tensions and you’re considered a different race or
ethnicity than your participants, will that change things? Proba-
bly. If you’re discussing body image and you look like a model,
or if you dress completely out of fashion, will that impact the
results? Probably. If you are interviewing people who are of a
lower social or economic class than you, or are younger or of
another disenfranchised group, will your interpretations of them
be biased? Again, probably yes.

Navigating the IRB
The IRB is set up at institutions to protect the rights

of human subjects (participants in your surveys, focus
groups, and IDIs) and to promote standards of profes-
sional research on campus. When you’ve selected methods
and you know you’re going to involve human participants,
it’s a good idea to get IRB approval. Not only are your
librarians familiar with this part of the process, but they are
often even members of their institution’s IRB. They can
also help with finding quality human participants for your
research. Because they are engaged in their own research
and have an idea of other research that is happening on
campus, they might even be able to put you in touch with
groups on campus or in your wider community that could
serve as resources or participants.
The key is to make sure that you’re attempting to remove

bias as much as possible, as well as to get both institutional
consent and approval through an IRB and individual consent
and approval from your participants (or their legal guardians
if they’re minors or otherwise unable to consent due, say, to
cognitive disability). It’s also important to remember that you
aren’t looking for certainty, or an objective way of knowing,
here. You’re looking for the depth and complexity of percep-
tion and reception (that is, interaction and belief). Researchers
more interested in objective certainty, who are looking for data
that can be more easily quantified, correlated, and/or predicted
across large samples often turn, instead, to surveys.
Surveys
CHAPTER
This chapter discusses one of the most broadly used

research methods: surveys. In it we will discuss how to do
a survey “right,” meaning by thinking intentionally about
your research question, your sample and population, the
questions you need to ask, and the scales participants will
use to respond. We will also give you an overview of some
mistakes that are often made in survey design, and how
to fix them. This chapter should help you avoid common
pitfalls and produce something that is valid, useful, and
doesn’t drive you crazy when you try to analyze the data.
Surveys are one of the most widely used (and misused) methods
that you see “in the real world.” In some ways that’s because
they’re so common, and seemingly so easy. We’ve all responded
to a survey at some point, and we’ve all given one. After all,
“How many of you like pizza?” is a survey. Simply put, surveys
are just a tool for collecting consistent data from a large sample.
Unfortunately, like anything that’s overused, many surveys are
poorly designed and don’t gather meaningful data. Part of the
reason is that there are just so many types of surveys and so
many types of questions you can ask. They can range from small
questionnaires about your customer experience to more open-
ended inquiries about why you quit your job. Before developing
a survey, you should ask yourself: What do I want to know? Who
do I want to ask? How do I want to send the survey out, and will
that impact responses? In the end, you’ll need to consider your
audience, the format, the questions, and the overall layout to get
the most out of your results.
Sampling: who do I want to ask?
If you’re conducting surveys, your goal is to be able to identify

patterns across a large group of people. You’re trying to figure
out what the majority of your participants think about some-
thing. Advertisers, for instance, love to rattle off how much of
the population favors one thing over another. But “four out of
five dentists agree our product is the best” doesn’t tell you much
in itself. Did they survey every dentist in the world? Only five
dentists? How many dentists would you need to survey in order
to be reasonably sure about your conclusion? This is where
“sampling” comes in (see Figure 8.1). It would be impossible
and unnecessary to survey every single person in your “popula-
tion” (e.g., all dentists). There are two main ways to cut down
your population: probability and non-probability sampling.
Probability sampling
Probability sampling is a method for randomly selecting partic-
ipants, usually using computer programs (e.g., Microsoft Excel
or SPSS) that number and randomize an entire population.
8: Surveys 121
Figure 8.1 Populations vs. samples
Random sampling is when you try to get a representative sam-

ple of the population (like in the national census or some big
public opinion polls). It’s intense and a lot of work, especially
if you use systematic sampling, which selects one out of every
three people (or every fifty, or any number depending on how
many you need). This method is a great way to curb selection
bias, and it will help ensure that your sample is an accurate rep-
resentation of your population. But is it always possible? Let’s
say you wanted to study the perceptions of college students in
the United States. Say you had a computer-generated list of
every college student in the country. It would be impossible
to reach out to each of them to take your survey. Probability
sampling works great in theory, but it isn’t how most people
conduct surveys.
Non-probability sampling
If you have a specific population in mind, you might need
to turn to another method that doesn’t rely on the general
population. There are a few ways you can do this. In con-
venience sampling, you select people close by or who you
know to take your survey (e.g., team members, classmates,
friends and family, or your social media friends). Purposive

sampling happens when the researcher strategically decides
which people will give meaningful responses. If you want to
know how people feel about your company or product, you
might want to sample only current customers, or people who
patronized your store in the past year, for instance. Lastly,
snowball sampling is beneficial when you are dealing with a
population that is difficult to locate. For example, if you’re
studying how alcoholism affects families, you might only be
able to locate a few willing participants. After they finish the
survey, you might ask them to help connect you to others
who fit the population criteria and are willing to participate,
and then ask those people for similar connections until your
sample grows big enough.
IRL: MAKING YOUR SURVEY HAPPEN
Getting your sample is one of the most daunting parts of

doing a survey. There are some automated services that
can help with this, though, and our online resources guide
you through a few. Mechanical Turk, an Amazon com-
pany, can get you thousands of participants very cheaply.
A little more expensive but usually better quality service
comes from the company Prolific Academia, which also
recruits and filters participants. If you’ve got lots of money,
you can also use a consulting firm like Gallup, Deloitte,
JD Power and Associates, and other organizations that
can often help you speak to customers and other specific
groups. When you’re ready to actually conduct the survey,
8: Surveys 123
two of the biggest online services in both academic and

business settings are Qualtrics and SurveyMonkey. They
provide a user friendly online format to create, send out,
and collect survey data quickly. Many advantages of online
surveys include cost savings, data collection efficiency, and
geographical dispersion, as well as maintaining complete
anonymity to help reduce social desirability bias.
Survey design and question types
Unfortunately, just because you know who you want to talk to

and how doesn’t mean you actually know what you need to ask.
Before you write your questions, it helps to look back at your
research questions and/or hypotheses. For instance, universities
want to know how effective faculty are at teaching (RQ1), and
how “good” the courses are (RQ2). One of the ways colleges
find this out is by surveying students about their perceptions of
their experiences with the faculty member and the class in the
Course & Instructor Evaluation that you take for every class
at the end of the semester. The reason colleges want to know
this information is because they want to measure faculty perfor-
mance and be sure that students believe they’re learning. But
does perception actually mean objective performance, especially
compared across disciplines, classes, and faculty? Probably not.
Still, student perception is a useful way to understand part of
the story, and it helps to ask several types of questions. You can
use numbered scales (e.g., a Likert-type scale), a set of possi-
ble responses (e.g., a categorical scale), yes/no questions, and
open-ended questions that ask for examples of your experiences.
A good survey will have an assortment of all different kinds
of questions aimed at trying to see what a group of people’s
attitudes are about something. Here are a few examples, each

with its own strengths and limitations.
1. Demographic questions
Even though each individual response should be anonymous,

your audience will probably want to know some demo-
graphic specifics about your sample. Common demo-
graphic questions include age, race/ethnicity, and gender,
but you may also want to know a few other things. Studies
on businesses might want to know the type of career each
participant has, family research may ask for marital status
and number of children, and political surveys may want to
know your address. Would socioeconomic status or house-
hold income be useful? With college class evaluations, how
about how many years you’ve been in university? In the end,
we advise you to include only demographic questions that
are relevant to your research questions and/or hypotheses.
2. Closed questions
In the last chapter we talked about the difference between open

and closed questions. Open questions give participants free-
dom to explain rich details about their attitudes and expe-
riences, whereas closed-ended questions limit responses to
a categorical question (e.g., What are the main reasons you
haven’t left your current job?) or a scaled question like a
Likert-type question (e.g., How much stress do you expe-
rience at work?). In both cases you are hoping to tie their
experience or attitudes to common preexisting attitudes,
opinions, labels, or categories. Your literature review should
help you choose what makes the most sense for your survey.
The goal is to identify and confirm variables so that you can
8: Surveys 125
Categorical question
1. What are the main reasons you haven’t left your
current job? (check all that apply)
 Strong coworker relationships
 High job autonomy
 High salary
 Ethical management
Figure 8.2 Categorical question
compare, contrast, and predict patterns across large sam-

ples. For example, you could show that employees in certain
professions don’t quit because they like their salaries, but
others stay based on their ability to be creative.
Scaled questions are meant to identify degrees of attitudes,
for example, how much or how little you like something. One
of the most commonly used scaled questions are known as
Likert-type scales (see Figure 8.3), which are usually used to
test hypotheses. Remember, a good hypothesis will have at
least two variables that are related in some way. These vari-
ables are “co-related,” or correlated positively (which means
that when one variable moves, the other variable moves in
the same direction) or negatively (which is when the vari-
ables move in the opposite direction). Say you have the fol-
lowing hypothesis:
H1: Interns who report strong communication sat-
isfaction with their coworkers will also report high job
satisfaction.
What are the two variables? Communication and job
satisfaction. What is the relationship between the two
variables? Positive, since we think that increasing commu-
nication satisfaction increases job satisfaction. The best
Likert-type scale questions

For the following questions select the number that best
describes you.
Strongly Agree Agree Uncertain Disagree Strongly Disagree
5 4 3 2 1
1. I am very satisfied with the conversations I have with
my coworkers.
2. I like talking with my coworkers at the office.
3. The conversations I have with my coworkers are very
rewarding.
4. I do not enjoy talking with the people I work with.
5. I hope to have more conversations like the ones I have
with my coworkers.
Figure 8.3 Likert-type scale questions
way to measure this in a survey is to find scaled questions

that represent both communication satisfaction and job
satisfaction. In Figure 8.3 we’re using an existing scale on
“communication satisfaction” to figure out their degree
of satisfaction on a 1–5 scale. Similarly, you would pres-
ent a Likert-type scale for job satisfaction, and then in
your data analysis see how they are (or are not) related to
each other (see Chapter 11 for how to figure out those
relationships).
Another useful form of scaled question is called the
semantic differential question. Instead of a 1–5 scale from
strongly agree to strongly disagree, the question has two polar
opposite terms on each side of the continuum. Participants
are asked to select how they feel within that scale. This type
of question can be clearer than a Likert-type question and
allows participants to respond more accurately in the middle
(see Figure 8.4).
8: Surveys 127
Semantic differential question

Your romantic partner is
Safe ____ ____ ___ ___ ____ Dangerous
Candid ____ ____ ___ ___ ____ Deceptive
Straightforward ____ ____ ____ ___ ___ Tricky
Respectful ____ ____ ____ ___ ___ Disrespectful
Considerate ____ ____ ____ ___ ___ Inconsiderate
Honest ____ ____ ___ ___ ____ Dishonest
Figure 8.4 Semantic differential question
Validity and reliability
If you’re writing questions, particularly if you’re going to test

a few hypotheses, you might ask yourself: “why do I need to
find existing scales? Why can’t I just write my own questions to
measure the variables in my hypotheses? Can’t I just write a few
questions to get at how satisfied someone might be in a specific
context?” Well, yeah, maybe. You can write out five questions on
a Likert scale. And those five questions might look like they’re
addressing satisfaction, but you might be wrong and have wasted
the survey. People also often ask why we don’t just ask one ques-
tion instead of five. Well, it’s not that easy. As perfectly as you
may have worded the question, people may misinterpret it, rush
through it too quickly to fully understand its intent, or simply
click the wrong box by mistake. It’s always better to include a few
questions, worded similarly, and then take the average score to
get the most accurate response. Think about your college classes.
What if the entire grade depended on one quiz score? Would
that be an accurate depiction of how you are doing in the class?
Probably not.
Figure 8.5 Validity and reliability
The only way to know if you have good, useful questions

is by testing them for validity and reliability. Reliability refers
to how consistent each question within your scale is with the
other questions. Examine the communication satisfaction scale
again. Notice how similar each item is? A great scale will have a
few questions that ask about the variable you want to measure,
no more, no less (see Chapter 11 for how you can test this).
Validity determines whether or not the questions you are asking
actually represent the things they should. Do they represent the
variable you intend?
8: Surveys 129
If you think about your questions as a game of darts, valid-

ity would be the darts that surround the bullseye (closely
reflect the variable you’re testing). Reliability would be how
consistently you throw those darts. Take dart board #1 in
Figure 8.5. Each item within that scale is very similar with
one another. They are all assessing the same variable, but it
might not be what you intended to measure. The darts on
dart boards #2 and #3 are all over the board, which means the
questions are so different that it’s almost impossible to deter-
mine if they mean what you intend. Dart board #4 is what you
really want. The questions are measuring what I want, and
responses are all consistent.

How might I track down existing,

valid, and reliable scales?
While you can go it alone and make your own questions,

it’s usually better to use ones that are already valid. Many
disciplines have established and extensive scale source
books that you can use, and we’ve linked to some of these
in the online resources. You can also look at articles that
are doing something similar to your project, and draw from
their survey scales. Just make sure they’ve chosen valid,
reliable scales first! This process can be a bit lengthy, so
you might decide just to search for these scales themselves,
rather than searching for articles that contain them. To find
useful scales you can do a couple of things: you can go to
the professional associations for the field in which you’re

researching, such as The American Psychological Associ-
ation (APA), who includes these scales in their database
PSYCHtests, the American Sociological Association for
Sociology, Rubin, Rubin, Graham, Perse, and Seibold’s
(2009) sourcebook or the National Communication Asso-
ciation for Communication, and Bearden, Netemeyer,
and Haws (2011) for marketing. You’ll also notice that a
quick Google search will help you find some freely avail-
able scales, as well as bring up some titles of some phys-
ical and ebook titles in your respective discipline that you
can then search on your library’s catalog. Some LibGuides
are specifically devoted to scales (which should come as
no surprise at this point!). Lastly, if you find a particularly
decent-looking scale sourcebook and your library doesn’t
currently provide access, you should request it via ILL.
Open questions
Not all questions can easily be evaluated in terms of reliability

and validity. If your research project has been framed around
Research Questions instead of Hypotheses, you are probably
interested in exploring people’s experiences in more depth or
richness than a closed-ended survey could describe. In this
case, you should use open-ended questions that hope to cap-
ture thoughts, feelings, and/or perceptions. You aren’t trying
to confirm how two or more variables are related, in this case, so
much as you are hoping to find connections between influential
themes. While you can do this with in-depth interviews, what if
your topic is very sensitive? What if you’re asking about instances
8: Surveys 131
of abuse, or the stigmas against poverty or being bullied? In

these cases, while an open-ended survey might not be able to
probe for additional information, because social desirability bias
will be reduced dramatically participants may be more inclined
to give rich and honest replies.
Common survey question mistakes
If you’re writing your own open-ended questions to gain more

information from your participant, be careful not use any of the
three most common survey question mistakes:
Bad question Why are these bad? Corrected

example
Double-barreled By asking two 1. What do you
questions: “What distinct questions value most
do you value at once, you run about your
most about your the risk of losing job?
job and why you valuable responses. 2. Why keeps
do you remain in Instead, ask them you at your
this line of work?” separately. current job?
Leading Leading questions 1. What does
question: direct the participant it mean to
“Would you to answer in a certain be a “team
agree that you way. Participants player” at
are a team tend to respond to your work?
player at your what they think the 2. In what way,
organization?” researcher wants to if any at all,
hear and not always have you
honestly. Instead, exemplified
remain neutral. this?
Figure 8.6 Bad questions

Loaded question: Loaded questions 1. Have

“How have trap your you ever
you resolved participant. It witnessed
unethical assumes action has unethical
dilemmas you’ve already happened behavior at
encountered?” and are guilty your work?
regardless of how 2. If so, what
they respond. actions did
Instead, use a you take to
prompt. resolve it?
Survey flow and design
Now that you’ve written your demographic questions, posed a few

categorical questions to determine differences, implemented a few
existing Likert-type scales to assess relationships between variables,
and developed a few open-ended questions to explore nuances,
you need to step back and take a look at your survey’s layout and
design. It should be extremely clear and easy to follow both in flow
(is it logical? Is it overwhelming?) and in question wording (do they
know what I’m asking? Will this give me useful data?). Keep in mind
that your participants in many cases are either being incentivized or
required to take your survey. So, a good survey should flow well,
be written in plain language, and have a simple format. There’s
a delicate balancing act here, though, between a strong, reliable
survey and an exhausting survey. “Survey fatigue” is real, and you
don’t want your participants to get so tired of taking it that they
stop halfway through or just give random answers to finish more
quickly. Some course evaluations are so long that students simply
don’t have the energy to write in the more open-ended questions
that most professors actually value and read, for instance. So then
the survey isn’t accomplishing what it set out to do.
8: Surveys 133
Survey types
Although it may seem like a simple issue, you’ll need to decide

on the best way to deliver the survey. Face to face? Phone? Mail?
Online? The decision isn’t as simple as it seems, because sam-
pling methods and survey types go hand in hand. For instance,
if you’re using a convenience sample of people in your store, you
might go old-school with paper, pencil, and a clipboard, or you
might add a brief survey at the end of their receipt or transaction
at the register’s iPad. Face-to-face surveys like this are useful
because the data is gathered immediately and you do not have to
worry as much about “user error.” However, as with interviews
they can be time consuming and can be heavily impacted by
both interviewer and social desirability bias. Phone surveys are
still used by large corporations who are committed to probabil-
ity sampling and hoping to report on public opinions. You may
have received a call from Gallup about which political candidate
you are voting for, or connected a Nielsen Media box to your
cable that monitors which television shows you’re watching.
Mail surveys suffer from low participation because it’s easier to
throw away junk mail than it is to hang up on someone, but the
cost is low and you can incentivize participation with coupons,
gift cards, and even cash in envelopes. Today, though, most
surveys are online surveys because they can easily be created,
changed, shared, and modified, and some allow for including
multimedia content or even conversations between participants.
IRL: COLLECTING DATA AT WORK
Most smart professionals are constantly looking for skills

that they can put on their resume and mention in future job
interviews. “Data analytics” is one of the most sought-after

skills in the current job market, and a research methods
course teaches you how to do it strategically and effec-
tively. Say your boss is looking for ways to increase revenue,
decrease employee turnover, and boost overall company
morale. You hear your supervisors talk about this in nearly
every meeting, but no one seems to have any answers. You
might mention to your boss, “You know, I’m not exactly
sure what the solution is, but I can certainly put together
a survey with some common reasons why people might
stay and leave their place of work.” Then, you could use
SurveyMonkey or Qualtrics to ask employees questions on
(a) employee engagement, (b) job satisfaction, (c) burnout
potential, (d) and coworker satisfaction and assess where
everyone’s attitude is on these scales so you can see where
the problems are. You could reward them with a free lunch
and a raffled grand prize for participating. After you get
the results, you could hold workshops or even hire a con-
sultant to fix the issues they identify. Congratulations, you
just paved the way to your next promotion.
For further reading
Bearden, W. O., Netemeyer, R. G., & Haws, K. L. (Eds.).

(2011). Handbook of marketing scales: Multi-item measures
for marketing and consumer behavior research (3rd ed.).
Thousand Oaks, CA: Sage.
Rubin, R. B., Rubin, A. M., Graham, E., Perse, E. M., & Sei-

bold, D. (2009). Communication research measures II:
A sourcebook. New York, NY: Routledge.
Observational
CHAPTER
9 methods in
practice
(participatory-action research,
experimental research, theory,
and metasynthesis)
This chapter is the last of our methods chapters and gives a

brief overview of a few different methods that are each based
primarily in observation. Participatory-action research (PAR)
focuses on working directly within communities to improve
their conditions, and is often a mix of other methods to pre-
serve a scholarly approach while not separating the scholar
from the community. Experimental scholarship contrasts
“control” and modified groups within highly controlled envi-
ronments to see how changing a variable might impact a sys-
tem, process, or interaction. Theory, broadly defined, is a way
to observe your surroundings and then draw from existing
knowledge, logic, and intuition to come up with explanations
for how things work and what they mean. Metasynthesis is a
sort of highly augmented literature review, where you bring
together a variety of sources on a topic and show how looking
at them all at once provides some interesting insights.
There are many ways to do research, as we’ve seen. Different

researchers simply think about their work differently. Is there
anything wrong with someone saying they want to know how
people think about their sex lives, and want to understand it in
as much specificity and “certainty” as possible? This can certainly
reveal many unanticipated results, as Alfred Kinsey found out in
the 1940s. But that data isn’t necessarily any more “accurate,” or
more “useful,” than someone sitting in their living room think-
ing about their personal experiences and imagining the structures
and concepts that gave rise to the various myths and preconcep-
tions that create our understanding of what “sex” and “gender”
even are. Similarly, does conducting twenty focus groups tell you
much more than working intensively with an organization that
helps counsel teenagers struggling with their sexual identity?
While we can’t go into every possible research method in this
book, since that’s not really the point, we do think it’s important
to give you a brief overview of a few other kinds of methods that
are grounded in observation.
Participatory-action research
The first of these methods, Participatory-action research (PAR),

is similar in some ways to ethnography, which we addressed in
earlier chapters. Both are a form of “participant observation,”
which means that the researcher is both observing and a part of
what they are observing. It doesn’t matter whether you want
to be a part of it or not; simply the act of observing a group
changes that group if they’re aware of it. However, while in eth-
nography the researcher is present but usually doesn’t attempt
to interfere or change the relationships or situation that they are
observing, PAR researchers argue that because simply observing
changes things, it is perhaps more ethical to actively intervene
to make the world a better place. PAR is primarily involved in
work with community groups, therefore, and researchers see
9: PAR, experiments, theory, and metasynthesis 137
themselves as active collaborators. Through interaction and dis-

cussion they form action and activism plans to positively impact
that community. While this always runs the risk of the researcher
pushing their beliefs onto the community if their experiences,
attitudes, or beliefs differ too widely from the participants’, the
intent is to embody the change that critical scholars often claim
is central to their work (rather than to only write about change).
Central to PAR is collaboration, not only in action but also in
making meaning and defining problems (and solutions) them-
selves. For instance, a scholar working on poverty would often
discuss the conditions with the community to decide as a group
what can and should be done to create change. This might
be fundraising, but it might also be organizing labor groups,
fighting predatory businesses, creating community gardens, or
any number of other solutions. In other words, PAR doesn’t
assume it knows the solutions. Like other forms of research, in
its ideal form it is guided by the data, not by the preconceptions
and desires of the researcher. Moreover, it attempts actively to
empower the subjects of the research by making the process of
research a collaboration that is grounded in the needs, desires,
and agency of those subjects. The line between activist, advo-
cate, volunteer, consultant, ethnographer, and researcher more
broadly defined are blurry here, but intentionally so.
Outcomes from PAR could be a policy or plan to present to the
local Board of Education, an infographic flyer to teach people how
and why to use condoms, or a cookbook or cooking classes for sin-
gle parents to create inexpensive, nutritious meals in a short amount
of time. The key here is to remember that this comes from a con-
versation with the community about priorities and desires. What do
they want, what do they need, and how? How can they be involved
and empowered in the process, and how can you help them as a
servant leader rather than as a voice of power and authority?
Example: PAR is often not a method that we think of con-
nected to the job market or professional development. However,
it can be a very powerful tool and useful skill. If you’re working on

corporate social responsibility, for instance, you’re trying to figure
out how to make your company do something to benefit both
itself and the community. You might start by looking at how the
various stakeholders within your organization think of their lives
and their relationship with the company, and the struggles that
they encounter with both. For instance, in one of Ted’s classes,
students were assigned to create a public relations campaign to
make the community better in some way. One student was very
involved with our Sexuality, Women’s, and Gender Studies pro-
gram, and she spoke to her community about the challenges they
face. Rather than emphasize pay equality, maternity leave, or other
large social issues women face, however, she chose a different
approach. What she heard again and again from women related to
menstruation, particularly that they worried about not having easy
access to pads or tampons. She found that this led to frustration,
shame, and anxiety. She also found through her literature review
that while women often feel disconnected from their employers,
a perception that the employer was thinking of them and their
needs increased job satisfaction and retention. Therefore, a stra-
tegic intervention would be to place baskets of free tampons and
pads in every women’s bathroom. The best part was that this was
incredibly inexpensive, and yet had a high return on women’s lives
and the bottom line of the company.
Experimental research
Experiments seem like they’re easy to do, since the idea isn’t
that complex. Take some of one thing and don’t do anything
to it. Then take some of the same thing and do something to
it. Compare. Then go eat ice cream and celebrate a successful
experiment. In reality, experiments can be one of the more dif-
ficult research methods to design and pull off. We’ve already
discussed some of the ethical issues with experiments, as well

as the necessity for using an IRB, in Chapter 4. What we hav-
en’t discussed in any detail is how to actually do that. The main
challenge with experiments is striking the balance between con-
trol and reality. With experiments in the social sciences you’re
designing a procedure to study humans that must be controlled
and standardized for each person. Therefore, potential outside
influences are minimized. At the same time, you’re trying to
create a scenario that mimics the real world so that you can draw
real-world conclusions.
In theory, in an experiment you want to see if doing some-
thing to a group of people will result in a change. For example,
perhaps you’ve developed a potential cure for an illness, and
now you want to see if this pill will cure it. In order to find out,
you’re going to need two groups of people, assigned randomly
(we can’t risk picking these groups ourselves and potentially
tainting the data). One group will serve as the experimental
group (i.e., the group that will experience the pill, interven-
tion, or stimulus to determine change). The other group will be
the control group (i.e., the group that will participate in your
experiment but will not be subjected to the pill, intervention, or
stimulus). This seems simple enough, but here’s the challenge:
both groups need to be tested in the same, controlled, environ-
ment. Since people’s experiences in real life are anything but
controlled and standardized, you have to create an environment
that is the same for every person in your experiment – otherwise,
you can’t be sure that what you introduced to them is the thing
that is causing the change. Critics of experimental design call
this an “artificial setting,” and they have a strong point. Bring-
ing people into a controlled environment (often a laboratory or
an office staged to appear like “real life”) isn’t real life. It is arti-
ficial. So while you are doing your best to make sure that every
person in your sample is experiencing the same environmental
conditions, you are also creating an artificial world that may not
reflect real life findings. Nonetheless, experiments are a great

way of isolating whether or not a specific stimulus is causing a
change to your sample.
Example: We couldn’t resist using a famous study that Josh
uses in his class. This study examines the connection between
kissing and allergic reaction reduction (because science, right?).
In this experiment the researcher wanted to know whether the
act of physical intimacy, specifically measured via kissing, could
reduce allergic reactions in the body. So how would you design
that experiment? And how would you accomplish this in a way
that could confirm that it’s the kissing that is reducing the allergic
reactions? Dr. Kimata began his experiment by randomly assign-
ing skin allergy sufferers into two groups: an experiment group
and a control group. The allergy sufferers from both groups
were instructed to pair up with their romantic partners for this
experiment. He used “kissing” as his manipulated variable and
then measured the degree of change in participants’ IgE levels
(levels of an antibody produced by the immune system during
allergic reactions). Because the outcome of the experiment
hinged on the degree of change in patients’ IgE levels, he had
to take samples before and after the couples “made out.” At this
point, you probably have a few more questions. Where are these
people kissing? Is he near them? And how do we know that it’s
the act of kissing that’s changing the IgE levels, and not merely
physical touch or even close proximity to their partners. Great
questions. One of the biggest strengths of true experiments is
that each one has a control group. Dr. Kimata had all of the par-
ticipants undergo the exact same conditions. Both groups came
into a low-lit room with soft music. The experiment group was
instructed to kiss for thirty minutes, while the control group was
told to sit next to each other, under the same conditions, and
embrace each other for the duration of the experiment. Sub-
jecting a control group to the very same condition, except the
manipulated variable (kissing), can help ensure that there aren’t
other factors affecting the outcome. As for the artificial setting,

there’s no great way around this. Researchers using experimen-
tal design believe that the pros outweigh the cons here. Sure, it
would be interesting (and absolutely creepy) to see if the results
would change if IgE levels were taken before and after an organic
kissing session within the couple’s natural environment. But this
would be very difficult to compare across your sample. What
if some couples had music and others did not? What if some
had short kissing sessions, while others had longer ones? What if
some didn’t stop at merely kissing? We think you get the point.
Standardized, artificial settings are the only way to try to silence
the effects of other possible variables contributing to the story.
In the end, Kimata’s results were confirmed. The kissing group’s
IgE levels were significantly lower than the control group levels,
and so the act of kissing was shown to reduce allergic reactions
in his sample. Interesting, right?
Metasynthesis
This method is, like PAR, a relatively new approach to scholar-

ship that attempts to address shortcomings in the range of meth-
ods researchers use. In this case, it isn’t involving the researcher
in the detailed activities of a community. In fact, while it is often
qualitative, it is a broad rather than narrow method. In some
ways, you could say that a metasynthesis is a discourse or con-
tent analysis of the scholarship about an issue. However, while
a metasynthesis often looks a lot like a literature review, it’s very
different. Instead of just surveying previous studies and orga-
nizing them, it looks deeply at the data used by those studies
and brings them together into a new, larger whole. This can be
very useful, since individual data sources and samples might pro-
duce very conflicting results, as can different methods. Bringing
together disparate approaches to the same topic can reveal new
connections, inconsistencies, and trends that wouldn’t be visible

through a small set of studies (if you think back to the “blind
men and the elephant” metaphor we mentioned in Chapter 3,
you’ll see what we mean).
Example: Because research on businesses doing good for
the community is so inconsistent, it can be a big challenge for
employees or even CEOs to advocate for it. Shareholders (or
bosses) often want to know how much money it will return to
the company, or if it is a “waste” of money. A metasynthesis
can really help with this. By gathering all the various forms of
data from the hundreds of studies done over the past couple of
decades on corporate social responsibility (CSR), a researcher can
more accurately point to why the results from previous research
are so disparate. They could then suggest that the company
duplicate the more successful methods with their own products
and their own intended customers or employees to determine
impact. After all, even retaining a few employees could save the
company hundreds of thousands of dollars, so you don’t nec-
essarily need to show a huge impact to demonstrate that CSR
is worth doing. The important thing is that you can show that
you’ve done your study using the best practices for researching
CSR, so that your boss or your company’s shareholders have
confidence that you’re right about the impact.
Theory
Some of the most impactful research ever conducted was done

by people who sat in their homes, offices, or even in parks and
libraries. It didn’t necessarily involve data the way we normally
think about data. It didn’t necessarily involve “participants” or
“respondents,” and wasn’t “repeatable” the way we often think
research should be. Instead, it is the product of deep and consid-
erable thought, discussion, evaluation, editing, and a mixture of
ethics, logic, and emotion. “Theory,” as we often call this form

of research, doesn’t need to be proven to be powerful or influen-
tial. In fact, it might not be possible to “prove” theory within the
current limits of society, science, or technology. But without it,
we wouldn’t know that the Earth is round, the stars are massive
balls of energy and matter, that gravity and time are related, or
that light is both a particle and a wave. We wouldn’t have com-
puters, and we wouldn’t have philosophy. We wouldn’t be able to
conceive of a world without war, hunger, or abuse. We wouldn’t
have dreamed of democracy, or a free press, or electricity.
In some ways, you could say that theory is philosophy. It
takes observations and conversations about the world and
processes them through the human mind and the social imag-
ination, creating ideas about how things are created, orga-
nized, changed, and disrupted. Like most philosophy, it takes
years of study to be able to produce theoretical work that is
grounded in true understanding; at the same time, like phi-
losophy it can often seem like common sense, especially after
the fact. But like a painting by Jackson Pollock, just because
something can be done doesn’t mean it has been done, and
certainly not with the same inspiration or intention. Theory,
like all research, adds to a conversation and takes it to the next
level. It is not a method we recommend to researchers in the
beginning of a project, but at the same time it is something
you shouldn’t ever avoid. Even the questions you have, and
the choice of your method, is in many ways already creating
theory. Once you have those ideas, however, at least in early
stages, we recommend that you let data guide you, and not
just your ideas in and of themselves.
Example: The early days of CSR were very different than
many approaches taken today. A car company might have
donated to the local ballet, or hosted a 5k race to promote breast
cancer research, or bought expensive paintings for a museum.
Each is considered to be CSR, but these approaches were often
not connected at all to what the company did or stood for, who
its employees were, or who their customers were. Some theorists
then suggested that CSR wasn’t inconsistent in impact because
it wasn’t strategic. In other words, the problem wasn’t CSR
itself. Theorists like Coombs and Holladay argued for “strate-
gic corporate social responsibility,” and others like Porter and
Kramer argued that we think of CSR as “creating shared value”
between a company and a community. These arguments weren’t
focused on data; they were focused on ideas, and in many ways
they reframed both what was studied and how.
For further reading
Kimata, H. (2003). Kissing reduces allergic skin wheal responses

and plasma neurotrophin levels. Physiology and Behavior,
80(2–3), 395–398.
Kimata, H. (2006). Kissing selectively decreases allergen-specific

IgE production in atopic patients. Journal of Psychosomatic
Research, 60, 545–547.
PART
Data analysis
and the final III
push
Analyzing
CHAPTER
10 qualitative data
and making
sense of detail
This chapter goes into more detail about how you ana-
lyze qualitative data once you’ve gathered it. Focusing on
textual analysis and interviews/focus groups, we discuss
how you determine what to look for, and how to choose
between a tightly controlled and structured approach and
a loose, open one. It’s probably not a shock that we sug-
gest something in between, a “managed” approach that
draws from existing research and your research questions
rather than hoping that the data will inherently guide you
(or that it will do so without bias). In each case we sug-
gest that you look at what might be “silent” or “erased”
material rather than only what you see. Finally, we give
you some hints about how to make this process easier for
yourself as well in the way you go about it and by getting
some help from a computer program.
10: Analyzing qualitative data 147
While we’ve given you an overview of many different methods

throughout the book, we haven’t discussed in detail how to
turn your raw data into anything useful. In other words, how
do you break it all down? In some ways we’ve prepared you for
this even as far back as the “choosing a method” and “writing
a literature review” chapters. What you were doing there was
actually a thematic analysis of your project, figuring out what
was important, what wasn’t important, and how you should
structure your argument as a conversation with other research.
This is also how you analyze almost all qualitative data. You look
for themes, structure the data into those themes, and then talk
about how the data fits into or interacts with the themes. This
can be a very organic process by which you allow the data to
take more control, or you can “manage” it by placing it into
existing structures, often called “typologies.” Both have their
benefits, and both have their problems.
(Managed) textual data (often used for

textual, content, and discourse analysis)
In the project on 13 Reasons Why, it was possible to watch the

show and see what popped into our heads. Those could have been
aesthetic features, scripted dialogue, settings, plot, and other ele-
ments. Some of the best textual data is organized in that organic
way, because it allows you to easily discover elements that you
might not see otherwise. For instance, in one student’s project
on the show Big Little Lies, she wanted to study politically pro-
gressive images of women in the show. However, when doing
an organic content analysis of the first season she realized that
the images weren’t quite as progressive as they seemed. While
women often appeared to dominate the storylines, imagery,
and narrative, that was with several qualifications. In some cases
their strength was also connected to psychological instability,
148 Part III: Data analysis and the final push
modified by abusive relationships, or contained within partic-

ular settings. In fact, when women were seen to have any sort
of power, it was usually away from the home and in a natural
setting like the beach. In other cases women were primarily seen
as mothers taking care of their children; their “empowerment,”
in other words, was almost always qualified by other elements.
However, it never would have occurred to us to code for natural
vs. human-made settings until seeing it come back as a trend
during the process of the content analysis. Organic textual readings
allow the data to modify the typology itself.
There’s nothing wrong with having a clear typology, either.
When doing the 13 Reasons Why project, it made sense to look
for elements drawn from best practices regarding effective treat-
ment and conversation with people at risk for suicide, because
that’s what we were looking for. In another project where we
looked for successful web development for nonprofits, it made
sense to look at what the scholarly research and the discussions
with industry professionals about best practices had to say about
success or failure. This adherence to a typology sometimes means
that you limit yourself to that data exclusively. However, it’s also
possible to remain open to other possibilities, and you can always
code for additional data. The key is to remain critical of your
own work, and aware that neither you nor previous scholars have
finished the project; otherwise, why would you be doing it?
There are many ways to actually do this, but the way we often
recommend is to make a spreadsheet in Excel. Yeah. We know.
But trust us on this one, it’s going to make your life a lot eas-
ier. Imagine you’re doing an analysis of social media posts. What
will you code for? Some of this will come from your literature
review, and some is just logical depending on your research ques-
tions (see our suggestions in Chapter 8). For instance, you’ll want
general information like the platform, the date and time it was
posted, how many views, shares, and likes it received, and how
many comments there were. Great. Each of those is a column in
the spreadsheet (we’re up to seven now). How about if there’s

an image, a video, or just text? That’s either one more column or
three more (we suggest three, so now it’s ten total). How many
words are written? How many hashtags? Is there a clear “call to
action” like a challenge video or a request to donate? Is there a link
for further reading? Is there some engagement like a poll, quiz, or
game (we’re at fifteen minimum now)? What quality is the image
(low, medium, high), and does it look like it’s taken by them,
or is a stock photo (seventeen)? Are there people, and if so how
many? What tone do they use (happy, sad, playful, angry)? Is there
an ethical appeal, a logical appeal, an emotional appeal (twen-
ty-two minimum)? What race or ethnicity do the people seem like
they are, if there are any, and what gender (twenty-four)? How
many of each type (twenty-six)? Do they tag or reference anyone
else, including maybe another organization or influencer (twen-
ty-eight)? This gets long, but it becomes an incredibly efficient
and rapid way to analyze and structure your analysis.
When you make each code, figure out what kind of code it
is. Is it for the presence or absence of something (1 or 0)? Is
it how strong/forceful/emphatic it is (scale of 0–3, maybe, 0
being “not present” and 3 being “strongly present”)? Or is it a
category like “young/old,” “product/event/news?” (You can
even make Excel give you a drop-down menu so you don’t have
to type each choice every time. Check out our online resources
for a guide.) Is it a number like “120 words?” Once you have
the spreadsheet with your codes, you can do your analysis. What
you’re looking for are correlations. You do this by searching for
anything that looks different, or anything that looks like it hap-
pens a lot (one way to do this is to use “conditional formatting,”
to automatically highlight things in different colors, and the
online resources show you how to do that). But it doesn’t have
to be fancy, either. Excel will automatically generate charts if you
highlight a couple of pieces of data (see our online resources
again), but if you use a couple of simple tools you can get a sense
of how different sets of texts compare. For instance, if you want

to know how many people on average appear in two organiza-
tions’ Facebook posts, you can use an AVERAGE(FirstCell:Sec-
ondCell) formula for each organization. Similarly, that would
measure how negative (-5 to 0) or positive (0 to 5) a politician’s
rhetoric is, or how many sentences out of the entire speech were
positive or negative (you’re coding for a number there). You can
also add up how many posts have a call to action, or how many
use an emotional appeal. Once you do that, of course, the cor-
relations become much easier to see (you will see how to analyze
this in further statistical depth in Chapter 11).
While content analysis and discourse analysis do sometimes
rely on correlation and numbers, it’s also usually important
to discuss some examples of each of the trends you’re seeing.
These often take the form of miniature textual analyses, just so
your reader understands what exactly is happening with various
trends. It doesn’t take many; one or two examples (what we
call “paradigmatic examples”) is usually enough to give readers
enough of an idea that they understand your point.
Remember that sometimes you might be looking for low
numbers as much as high, or you might look for things that are
“outliers” (posts that are very different). For instance, wouldn’t
it be interesting if the only posts that show someone working
in manual labor show people of color, or men (so White or
Women would be coded as “0”)? Wouldn’t it be interesting
to see why two out of thirty posts went viral and got over one
million views, in comparison to the other posts? What changed?
This is really where you get to the “managed” concept. While a
textual, content, or discourse analysis might seem to be totally
organic, usually we’re choosing to focus on certain data over
other data, and connect one category or code to another. The
reality is that all research is managed at some point or other,
but you should try to keep in mind that the amount of manage-
ment, and how you manage, is really important. After all, your
preconceptions or biases might send you in a bad direction. For

instance, one student working on images of the police during
a time of intense protests against police brutality performed a
content analysis of news segments but didn’t notice that a great
deal of her data didn’t actually fit her desired outcome (to show
that the police were unjustifiably portrayed negatively), or that
she wasn’t coding for things she needed to (that the victims
were portrayed as thugs or shown with negative images, and
that protesters were often characterized as “rioters”). Once she
realized that was happening, the entire project, and her conclu-
sions, changed completely.
(Managed) grounded theory (often for

interviews, focus groups, and surveys)
This is even more important, and yet sometimes harder, for

research on participants. As we discussed in Chapter 7, bias is
inherent and inescapable, especially when you’re working with
people. There are so many variables that it’s far more difficult
to address them all effectively. We mentioned that video and/
or audio recording participant research like interviews and focus
groups is a helpful way to approach it, especially if you have
more than one researcher analyzing the data. Transcriptions are
also extremely useful, since you can remove issues like accents,
skin color, and paralinguistics (e.g., that women in the United
States speaking English often end sentences with a nonverbal
“question” sound that might reduce their forcefulness or credi-
bility with some raters, or that Chinese women often have a dis-
tinct voice from men in the way tones are used for the words).
But what do you do when you have the transcriptions? How do
you actually analyze what people say?
In reality it’s not that much different from doing a literature
review or the types of textual analysis we discuss above. The data
can guide you here, as can any hints for themes that you might
find in the literature to help you out. For instance, as we dis-
cussed above, in Owen’s (1984) article on thematic analysis he
suggests looking for repetition (a participant that says the same
thing multiple times), forcefulness (the strength of a partici-
pant’s statement, which might or might not appear with other
participants), and recurrence (when several participants say the
same or similar things). This framework is really useful, because
it allows you to eliminate even more issues of bias. In many ways
you’re looking for both similarities and dissimilarities, and that
gives you a lot of power as a researcher to make sure you’re
doing the data (and your participants) justice.
The process of pulling out themes from the data, rather than
choosing themes and then laying them on top of the data or
forcing the data into them, is called “grounded theory.” A more
structured approach, like we saw above, is called “managed
grounded theory,” in which you have an idea of themes before
you start but you remain open to them evolving or even being
discarded as you look at your data. Rather than a spreadsheet,
however, we recommend that you take all the transcriptions
(and we can’t emphasize enough how important it is to tran-
scribe! See our online resources for some suggestions on how to
make that faster) and highlight them according to theme. This
looks a lot like the way we talked about doing literature reviews
in Chapter 3, right? That’s no accident. It really is the same
process. Find themes as you go. Anything that stands out, that
you see repeated within or across transcriptions, or moments
when you ask yourself “wait, but what about . . .” are things
to code for, even though in the last bit you’re just pointing out
the absence of something (an emotion, a logical connection,
etc.). Then you label each highlighted section with the partici-
pant number or pseudonym and place all the highlights of each
color together. Those groups of quotes then become the basis
for your analysis.
The fun thing about this kind of data analysis is that you can
allow the participants to speak, just like your paradigmatic exam-
ples “speak” about your data in text-based analyses. This kind
of research tends to use a lot of paraphrases and a lot of quotes,
sometimes even extremely long quotes, to make their points.
You’re not necessarily trying to say what these mean, you’re sim-
ply showing in a structured way how participants make sense
of their reality, how they take in and then express various stim-
uli like a TV show, conversation with a boss, or relationship
counseling appointment. That doesn’t mean you aren’t being
persuasive and making an argument; you are. It’s just that the
data speak for themselves, and if you’re being ethical, they speak
accurately. Remember, these have a relatively small N; you’re
not interviewing thousands of people in most cases, so you have
very little ability to generalize. But you can certainly say that
these people felt these ways about the topic, which can be just as
powerful and accurate as a survey with thousands of responses.
WORK SMART, NOT HARD: USING

TOOLS LIKE NVIVO FOR ANALYSIS
Remember, with qualitative analysis, you’ve signed up to

grapple with rich, detailed, complex data. Whether it’s
over twenty hour-long interview transcripts or hundreds of
newspaper articles, you probably have a ton of information
to sort through. Luckily, there are a few software programs
that can make your life a bit easier. NVivo, a qualitative
data analysis software package, is one of the most widely
used tools in helping researchers create coding schemes and
themes for large amounts of data. Not unlike the “search
toolbar” on most websites, NVivo allows you to create your
own coding categories and search conditions for those cate-

gories. Let’s say you are analyzing the portrayal of emotions
via character dialogue across several screenplays. Yes, a sim-
ple solution would be to create searches for (a) “anger,” (b)
“sadness,” (c) “jealousy,” and so forth. But think about how
many different words and phrases could be used to portray
the same emotion? If you only searched using the specific
word for each of the seven primary emotions, you would
miss synonyms and similar phrases that could be included
within your coding categories. Nvivo allows you to build
onto your categories by including phrases, terms, slang, and
other words that represent your specific categories. Once
the Nvivo program tags each of the phrases and aligns them
within your coding scheme, you can further subdivide your
categories as you see fit. This can even be done with images
and video, and you can share it among collaborators for a
smooth process.
Silences
We think it’s worth emphasizing that one of the most import-

ant things to be aware of as a researcher is the silences in your
data. What are your participants not telling you, and why? If
you know that statistically 90% of a group of college students is
having sex, and yet no one wants to talk about it, why is that? If
you know that most of them have at least some sexual interest
in their same sex, why do so many of them claim to be “totally
straight?” Why would the one or two students of color in a class
not speak up about issues that impact them directly, and the
experiences they’ve had that prove it? We’re kind of stacking the
deck here, of course. The answer to almost all of those questions

is social power, and the stigma that is often attached to speaking
out against power. As researchers, we have to remember that
this isn’t just about interviews or other forms of human subject
research. It’s also in our textual data.
How? That seems like it’s counterintuitive, right? The texts
and the data from the texts will guide you. But what if there’s
something there that you’re not looking for? For instance, “third
wave feminism” strongly critiqued other members of the femi-
nist movement for speaking about “women” as if they were this
monolithic group, rather than acknowledging that individual
women are in fact very different from each other. The road to
empowerment and the road to equality often diverge. Women
of color experience femininity and oppression very differently
than those of the dominant group, and in fact the third wave
pointed out that many of the arguments made by feminists at
the time simply weren’t as relevant to them (e.g., does having
more women in leading roles in the movies matter when you feel
unable to report sexual assault to the police?). This is a type of
silence that we can find by expanding our view from what we’re
looking for to what’s there, but also trying to look for voices and
experiences that might be silenced or erased from the texts. The
history of society may be written by the dominant, but it’s our
job as researchers to challenge what we think we know. As we
mentioned in Chapter 7, bias is something we all have, and it’s
our job to try to remove it as much as possible. Common sense
may seem logical, after all, but different family, school, economic,
and social experiences make for very different versions of it.
For further reading
Owen, W. F. (1984). Interpretive themes in relational commu-

nication. Quarterly Journal of Speech, 70(3), 274–287.
Analyzing
CHAPTER
11 quantitative
data and
making sense
of numbers
Building on the last chapter’s approaches to qualitative

results, this chapter shows you how to analyze number-fo-
cused data. We mainly focus on surveys and how you can
use the numbers you get from those data. There are a lot
of terms in this chapter, but we place them in tables and
charts rather than go over them in too much detail. We
want you to know above all what they are and why you’d
use them. We’ll also go over using graphs and tables, and
talk about how you could use tools like SPSS and Excel to
make this process faster.
11: Analyzing quantitative data 157
Loving and hating statistics
It’s important to point out that all research attempts to report

what is “true.” In the last chapter, we talked about analyzing
and reporting on people’s qualitative accounts, their written
and/or oral communication about particular events. Since we
will never fully know exactly what’s happening inside some-
one’s head, social scientists have to rely on people’s accounts.
Qualitative scholars like words and descriptions. Quantitative
researchers condense participant accounts to a number. While
this removes some great thick description, there are some
big advantages as well. If you’ve chosen to collect numeric
responses to questions, you’re most interested in finding
trends and patterns across a sample. Simply put, numbers are
easier to collect in large amounts and much easier to com-
pare than people’s words. That also allows you to turn the
data into infographics (visual representations of data, like pie
charts, so people can understand them quickly and easily).
So if you want to know how every student feels about their
college experience, having them report a number on a scale
will give you data that can be presented in a statistic. Other
questions on your survey could find out if there’s a differ-
ence between freshman and senior attitudes. And even more
questions could get to the bottom of whether there’s a rela-
tionship between satisfaction and graduation rate. In the end,
everything you collect with numbers serves these purposes:
(1) it describes your sample, (2) it tests for differences, and/
or (3) it tests for relationships. This is all useful, but most
people have a love/hate relationship with statistics. There’s
something comforting about wrapping our minds around a
number that seems to represent a specific, definitive truth.
On the other hand, the idea of taking a course on statistics
sounds excruciating, and we know that any c omplex system
is extremely difficult to fully describe by only using num-

bers. After all, 42% of all stats are made up on the spot. Or
is it 76%? We forget, but you should trust us anyway because
we’re wearing suits.
Descriptive statistics: who are your people?

What are your variables?
The most common way people use statistics is to describe (1)

their sample (e.g., demographics) and (2) what the people
within their sample have said and/or feel. The two most com-
mon descriptive statistics you might use are percentages and fre-
quencies. It might be important to know how many people in
your sample are first-year students vs. graduating students, for
instance. It also might be important to know what percentage
of people are employed and unemployed when trying to figure
out what to do about homelessness, or what percentage feel that
everyone is entitled to healthcare, shelter, and food. We’re cer-
tainly pointing out the obvious here, but when reporting your
sample’s demographics, relaying the percentages of who they
are and any other pertinent identifiers, roles, categories, and
any checkbox questions help tell the story of your sample to
your readers. Frequency is how often each of those percentages
occur, which is useful when you have more than two options
(year in school, for instance, rather than graduate vs. undergrad-
uate students).
Describing variables
Let’s say you want to tell your audience how satisfied a group
of people feel about a product. When you report this, you are
relaying your sample’s “bell curve” for that particular variable,
otherwise known as a “normal distribution” (see Figure 11.1).
A bell curve is often represented by the following figures:
Figure 11.1 A “normal distribution”
• Mean (M or µ): the average score of every single response

(e.g., average grade in class is a B).
• Median (Mdn): the middle score if they were lined up from
lowest to highest value (e.g., of twenty-one students, what’s
the eleventh highest/lowest grade?).
• Mode: the value that is most often reported (e.g., most peo-
ple got Bs)
• Number of cases (N): the number of units (e.g., number of
students) analyzed.
• Range: the field of values from the lowest to the highest
(e.g., grades range from D+ to A-).
• Standard deviation (SD or σ): how all of the scores are dis-
tributed across your sample in comparison to the mean
(e.g., were most grades in the B range, or were they mostly
As and Ds?).
These numbers need to be seen together, since each tells part

of the story, but not all of it. I’m sure you’ve combed through
a few amazon customer reviews before purchasing something,

right? If the average (M) score is 4.5/5, this tells you that, on
average, people were generally very satisfied. But what if only
three people (N) reviewed the product? Would you still trust
that number? What if out of a thousand reviews the average is a
3, but all reviews are either 5 stars or 1 star? That might make
you look a little closer to see if there are stories about it breaking
in the first month, or about bad customer service, right?
This becomes especially important when your range is wide.
In many ways, the mean alone only tells half the story. A large
standard deviation communicates that the sample’s scores were
scattered greatly from the mean, while a smaller standard devi-
ation tells you that the scores were closer together. Reporting
age is a great example of why SD is important. You might list the
mean age of your sample as twenty-five, with a range between
eighteen and seventy-five. But it’s your SD score that will tell
you how spread out (or close) the ages of your sample are. For
example, Josh worked on a research project about long-distance
romantic relationships. His research team noticed that the mean
age of the sample was twenty-seven years old, with a range
between eighteen and sixty-five. Further examination showed
that there was fairly large standard deviation, which told them
that there was a wide range of ages, instead of a large cluster of
twenty-seven-year-olds in his sample.
While we’re discussing the best way to report your data,
remember that a “normal” bell curve assumes that your score
distribution is just that: normal (i.e., the curve is completely
symmetrical, which means that the number of scores below the
mean are equal to those above the mean). Sometimes, though,
you might have a skewed distribution, either negative (where
most data is lower than the average, as you can see in Fig-
ure 11.2) or positive (where most data is higher than the aver-
age, as you can see in Figure 11.3).
Figure 11.2 Negatively skewed distribution curve
Figure 11.3 Positively skewed distribution curve
If your curve is in fact a normal bell curve, you should be fine

to just report the mean score. Because of the potential in skewed
distribution, though, it’s a good idea to relay the three central
tendency measures (mean, median, and mode) to give your
reader a more complete picture of what’s happening. Reporting

the median and/or mode can be useful if a few of the values
in your data set are extremely high or low and, therefore, will
throw off your average. Let’s say you wanted to report how far
people travel to and from work every day. You’ve got five hun-
dred employees, but two of them reported thousands of miles
of average commute time due to international travel, whereas
everyone else lives in the neighborhood. So what would have
been an average of 15.5 miles now shows up as 210.7 miles. In
these skewed cases, you should include the median to help clar-
ify the inflated average distance. In another example, the mean
temperature of Orlando, FL, is 73℉. That’s a great number for
Disney to tell tourists. If you live there, however, you know that
summers can easily get to 93℉, which is really hot. In this case, a
mode of 82℉ is more representative of an average day. In other
words, in Orlando more days of the year have a high of 82℉
than any other temperature.
While descriptive statistics are the most common and sim-
plest way to report number data, it doesn’t tell a very complex
story. Yes, there’s great value in knowing that 40% of your staff
is female, or that 62% of the employees at a company have been
there fewer than three years, or that 30% of your sample are
reporting workplace burnout, but these numbers don’t tell us
anything about the differences between genders, between new
hires and experienced employees, or those who burn out and
those who don’t.
Finding differences through inferential

statistics
Statistics that reveal these kinds of group differences and possi-

ble influential factors at play are known as inferential statistics.
They can help you infer, draw a few possible conclusions, and
for the most part tell a more complete story. Do college seniors
know more than freshmen? About what? Are there significant
differences in how we pay men and women, and is that all men
and women or just those who have children? Does socioeco-
nomic status influence certain outcomes? Does race play a role?
Most statistical analyses focus on finding significant differences
between groups. We want to briefly walk you through three
common tests for identifying differences between groups and
variables. But before we do, let’s review a few definitions.
• Independent variable: the variable selected by the research

that does not change and is not affected by the other
variable(s).
• Dependent variable: the variable that changes based on
movement from the independent variable.
• Continuous variable: a variable with a numeric value any-
where between a specific minimum and maximum value
(e.g., 1–7).
• Categorical variable (nominal): a variable with a specific
amount of optional responses, not necessarily in logical order.
t-tests and p-tests

The t-test was developed in 1908 by William Gossett as a quick
way to monitor the quality of Guinness beer. Yep, a hired bio-
chemist used science and statistics to track the quality and flavor
of one of the world’s most popular beers. (And you thought
stats had no real-world value.) In short, the t-test (t) is the sim-
plest, clearest way to determine differences between two groups.
Let’s say you want to determine whether seniors have better
communication skills than freshman. A t-test compares the com-
munication skill level averages (dependent variable) among the
two groups: freshman and senior (independent variable). If the
mean value of the senior communication skill level (4.8/5) is
significantly higher than the mean value of the freshman level
(4.3/5), you could conclude there is a difference between the

two groups. Remember, the independent variable can stand
alone – it doesn’t change – and it always fits into a group or
category (nominal) when doing a t-test (year in school). The
dependent variable (continuous) is the variable that changes
depending on which group you’re considering (in this case, the
communication skill score). For a quick tutorial on how to cal-
culate a t-test, see our online resources.
But how do you know if this difference is not just random?
Is there a strong number pattern to show that these two groups
are in fact different? A quick significance level (p) test on a com-
puter program like SPSS or Excel will calculate your numbers
to see if what you found could have been just dumb luck (com-
pletely random) and what percentage is showing a distinct pat-
tern (see our online resources for how to do this). Most scholars
agree that if over 95% of your numbers (i.e., p <.05) show a valid
pattern then it’s safe to say that you found something significant
and can declare that the two groups are in fact different. We
wish we could tell you that there’s a great logical explanation for
why the stats community has decided that 95% probability is the
magic number, but there’s not. It’s tradition, and for now it’s
the standard. It’s not uncommon, though, to see significance
levels reported at p <.05, p <.01, or p <.001 (i.e., 95%, 99%, or
99.9% significance, respectively).
Now keep in mind that the t-test can only be used if you have
only two groups (independent variable) and one continuous (scaled
or interval) dependent variable. Anything more complicated than
this, like adding in a third group or measuring more than just com-
munication skills, and you’ll need an upgrade your test.
ANOVA
Similar to a t-test, the ANalysis Of VAriance Test, or as it’s more
commonly known the ANOVA (F) test, is useful when you
want to compare differences between groups. Unlike the t-test,
Dependent First independent Second Third independent
variable variable – year in independent variable – race
school variable –
gender
t-Test Communication (only two groups) X X
competence Freshman vs. Seniors
One-way ANOVA Communication (two or more groups) X X
competence Freshman vs.
Sophomores vs.
Juniors vs. Seniors
Two-way ANOVA Communication (two or more Male vs. X
competence groups) Freshman Female
vs. Sophomores vs.
Juniors vs. Seniors
Factorial ANOVA Communication (two or more Male vs. White (not Hispanic)
competence groups) Freshman Female vs. White vs. African
vs. Sophomores vs. American vs. Asian
Juniors vs. Seniors American vs. Other
Figure 11.4 Tests and variables

the ANOVA, is great for comparing more than two groups.

Depending on how many groups you want to compare, there
are a few different ANOVA options. As we just demonstrated in
our example above, a one-way ANOVA is used when you have
one independent variable (with three or more categories) and
one dependent continuous variable (e.g., year in school and com-
munication skill). A two-way ANOVA is used if you have more
than one independent variable that you’d like to test (e.g., male
freshmen communication vs. female sophomore communica-
tion). Want to explore what happens with three, four, or even
more categories of people? Then you’ll want a factorial ANOVA
to really flesh out the complexities of different cross-groups to
determine how they would differ (see our online resources for
how to do these tests). Testing differences across groups and
co-cultures of groups can really yield some rich details and
showcase the complexities of group memberships on particular
outcomes. And yes, just like the t-test, you’d have to check the
significance level (p-value) to make sure your p < .05 to declare
any significant findings across the groups. Take a look at the
differences between these tests in Figure 11.4.
Finding relationships
Remember, in some ways all of science exists to disprove the

idea that everything is random. We call this the null hypothesis:
“nothing is patterned.” When you conduct scientific research,
your entire goal is often to disprove this hypothesis. You are
showing your audience that there is a pattern, an observable
relationship between concepts. The natural sciences have discov-
ered there is a connection between sunlight, rainfall, and plant
growth, and that everything on this planet with mass submits
to the force gravity. The connection between these natural phe-
nomena is not random; it is measurable and even predictable.
And while many of us may get a bit squeamish thinking about

human behavior and social forces as patterned and predictable,
you have to admit that if you step back and think about it, there
have certainly been a few recurring (and maybe even predict-
able) patterns in your life, right? Any time you’ve ever uttered or
heard the phrase “see, I told you so,” it’s only because someone
noticed a pattern, a connection between behaviors and events
that led to a prediction about what would likely happen. Finding
these connections and patterns in your own research can be one
of the most valuable stories your data can tell. In this section, we
want to highlight two of the most used tests to infer both con-
nections between variables (correlation) as well as predicting
what future outcomes might be (regression).
Correlation
The simplest inferential statistics test is the correlation. This type
of preliminary test is used if you want to quickly rule out or
confirm that two variables are related in some way. Say you’re
conducting research on nonprofit organizations. Is there a rela-
tionship between socioeconomic status and donor activity? Or
is there a connection between people who share the mission of
the nonprofit and those who volunteer? Relationships between
variables can happen in two ways: positive or negative. A posi-
tive correlation means that the two variables move in the same
direction. So, if you find that individuals who align closely to a
nonprofit’s mission also volunteer more often than those who
do not, then you have a positive correlation between “shared
mission” and “volunteerism.” In the same way, you could also
conclude that individuals reporting a low degree of affiliation
volunteer less. Correlations are reported using the (r) symbol.
After you’ve checked to makes sure the p-value is p < .05, you
can report how correlated the two variables are. Keep in mind
that a 1:1 correlation or (r) = 1.0 means that your two variables
are identical. Oops. So, as a rule of thumb, “strong” correla-

tions will be between .60 and .79 and “very strong” correlations
are between .80 and .99.
A negative (sometimes called inverse) correlation means that
as one variable increases, the other variable decreases in value.
So, if you wanted to know why volunteerism is low for a particu-
lar organization, you might test this alongside an organization’s
“schedule inflexibility,” “distance from home,” or other vari-
ables that you think might help explain the story. If you found
that any two variables were significantly negatively correlated,
you could conclude that they would result in someone being less
likely to volunteer.
Again, correlations are fantastic for quickly determin-
ing some kind of relationship between two variables. But we
can’t stress enough that correlation does not necessarily equal
causation. It just means that when one variable moves, so does
the other variable. It shows a pattern, but it can be dangerous
to say that one variable is causing the other variable to move.
For instance, let’s say you want to know why the population of
earthworms in your neighborhood is decreasing. After rigor-
ous observation you’ve concluded that there is a strong positive
correlation between squished worms on sidewalks and opened
umbrellas nearby, and therefore to save the worms we should
ban umbrellas. That would be a ridiculous conclusion, but it
reminds us that correlations rarely tell the whole story. Obvi-
ously, in this case, the rain is causing both of these to happen,
but one doesn’t cause the other. Not only that, but neither
would cause a large decrease in the worm population. It’s
important to realize that you can mislead readers into drawing
false conclusions from spurious correlations when you state that
one variable caused another variable to occur without recogniz-
ing that a third (or possibly more) variable is absent from the
equation. Check out our online resources for a set of hilarious
(to us), spurious correlations.
Regression
The holy grail of inferential statistics is predicting future change,
and you often do this with what we call “regression analysis.”
While statisticians and theorists love to debate whether or not
you can ever truly show causation, regression analysis gets us
closer than the other tests. Linear regression examines how one
predictor, or independent variable influences one outcome, or
dependent variable. Similarly, multiple regression allows you
to include several different predictor variables (independent) to
determine the outcome of another variable (dependent). Keep
in mind that good statistical analysis is about telling the story
behind the numbers, and regression allows for that kind of com-
plex storytelling. It has four major benefits:
1. It allows for several predictor variables to be included in the

mix, which tells a more complete story.
2. It uses “naturally occurring” variables (i.e., actual and cur-
rent participant perceptions) as opposed to artificially pro-
voked laboratory variables, so you can make predictions
without having to do an experiment.
3. Each variable is individually accounted for to see how strongly
it affects the outcome variable (i.e., beta weight, ẞ).
4. It explains how much (percentage, or R2) of the outcome
variable is being predicted by the other variables.
As an example, let’s go back to our study on nonprofit volun-

teers. After much research, you’ve landed on the following four
predictor (independent) variables that would lead to volunteer
participation:
1. Identification with the organization’s mission.

2. Previous volunteer activity.
3. Peer group influence.
4. Interest in meeting new people.
Running a multiple regression (more than one independent

variable) analysis will tell you several important things. First, you
will be able to determine which of these variables (if any) actu-
ally predict the likelihood of volunteer participation. Referred
to as the beta (ẞ) coefficient (or beta weight), this value will
tell you how strongly correlated each variable is to the depen-
dent variable (volunteer participation). (We give you a guide on
how to calculate the beta coefficient in our online resources.)
As always, check the p-value to make sure p <.05, and then you
can report the beta weights to determine the strength of each
predictor variable. Just as with correlation tests, the closer to
1.00, the stronger the predictor. Lastly, the big advantage of
multiple regression is the ability to see what percentage of the
outcome variable is represented by the other variables. In other
words, you can check the (R2) value in your output to see what
percentage of volunteer participation is explained by your four
variables. This is important because it lets you know how close
you’ve come to telling the whole story. If your model explains
and predicts a large percentage of volunteer participation, you
can celebrate. If it’s a smaller percentage than you anticipated,
then it’s back to the drawing board because you have some more
investigating to do.
Okay, we admit it. We dropped a lot of terms on you in this
chapter. You may not need to know all of these terms or conduct
all of these tests, but we wanted to give you several options for
how you might analyze your numbers. Below we’ve included
a reference chart (see Figure 11.5) that lists most of the major
concepts and tests we discussed. Use it as a quick reference when
you’re stuck trying to remember the difference between median
and mode or t-test and p-value.
In the end, no matter what your method, your research
should tell a story. For qualitative analyses, you have the benefit
of using the words of your participants to tell that story. But
for quantitative analyses, you must realize that there are stories
behind the numbers. Those stories can be simple or complex,

but if you understand how to use the numbers and what they
mean, they can be extremely powerful.
Descriptive
statistics
Mean (M) The average numeric value of all of the
data within a variable.
Median The numeric value in the middle when
the data is ordered from lowest to
highest value.
Mode The numeric value that appears the
most frequently.
Number of The number of participants or data
cases (N) points analyzed.
Range The lowest and highest number values
in the data set.
Standard The incremental distance each data
deviation (SD) point is from the mean and the other
data points.
Skewness When there is an uneven amount of
scores either higher or lower than the
average in a normal curve.
Independent The variable that is selected by the
variable research; it stands alone, and does not
depend on the value of the dependent
variable.
Dependent The variable that changes based on the
variable movement of the independent variable.
Figure 11.5 A quick guide to statistics terms

Inferential
statistics
t-test (t) A test to determine if there’s a difference
in the mean scores between two
groups. An independent samples t-test
compares two separate groups, while
a paired samples t-test compares the
same group over time (pre/post test).
p-test (p) A test telling you how significant your
differences/correlations are. This is usually
measured at the 95% level (p <.05).
ANOVA (F) An “ANalysis Of VAriance” test
determines significant differences in the
dependent variable across more than
two groups.
Correlation (R) Any mutual association between two or
more variables.
Linear A test to predict how one variable
regression (independent) predicts the value of
another variable (dependent).
Multiple A test to predict how more than one
regression variable (independent) predicts the
value of another variable (dependent).

Persuasion,
CHAPTER
12 presentation,
and publication
Before you’re done with your research, you have to actually

write it. While your literature review (Chapter 3), meth-
ods (Chapters 4–9), and data analysis (Chapters 10 and
11) are big chunks of that process, it’s important to see
how to tie it all together. We’ll talk about some best prac-
tices for writing a good research paper here, some things
to avoid, and various ways you could end up explaining
this project to other people (in a poster, in a presentation,
and in a paper). More than anything, though, we want to
show you how to make a smart, interesting project that
will get people thinking.
We hate to say it, but no one really cares about your research. It
might be brilliant, but no one is excited to read a paper (often
not even if they know it’s good). It’s your job to make them
(ok, help them) care. There are a few ways to do this, but it all
starts with the beginning of the paper. Your title is really import-
ant, because it foregrounds what’s going to happen. You can try
to be cute about it, but at least try to be specific about what
you’re doing with a hint as to why. And keep it short. Sometimes
we start with a title and go from there; with this book, in fact,
we started with a title and that guided our approach. Then we
changed the title, because it was pretty bad. But it still inspired
us to structure everything in a specific way and for a specific set
of audiences. Hopefully you liked it.
After the title comes the abstract, which you again could write
before or after the paper is completed. However, you’ll always
want to change the abstract when you finish, since if you’re doing
your job right the data that you’ve analyzed should have guided
you to conclusions you didn’t necessarily expect when you wrote
the abstract. Basically an abstract is a very brief (fifty to one hun-
dred fifty words, usually) introduction to your paper that people
can read instead of (or hopefully as a prelude to) your full paper.
It should include your topic, your focus, your method, and, in a
very brief form, your conclusions. You might also talk a bit in the
beginning of your paper (in your abstract and your introduction)
about why the research matters. If this sounds a lot like an anno-
tated bibliography entry, that’s because it is. It’s exactly like that.
After the abstract comes the body of the paper, although you
still need an introduction. This is where you can either really get
someone interested in your project or make them stop reading.
You can’t just jump into the literature review; you might have
hated writing it, but they’ll hate reading it just as much. And
you definitely don’t want to start with general, vague statements.
There’s not much worse than reading someone writing about
“society” or “from the beginning of time” or “the beginning of
humankind” or some other such ridiculous statement. Yes, we
know that almost everyone has been trained to write that way.
Stop it. Please, for everyone’s sake, cut everything vague and
undefined like that out of your paper.
12: Persuasion, presentation, publication 175
It’s much better to start a paper with something interesting

that has recently happened. Did a politician get into a scandal
that inspired your research or should inspire us to read it? Did
a new TV show come out that got a lot of attention, and did
they change that show in response? Was there an event that
sparked a lot of cultural debate (e.g., White people calling the
police on people of color for no reason other than because . . .
well, because racism)? Tell us about that conversation or event
or scandal, and suck us into why your project is important not
just to you, but to us as reasonably intelligent, interested indi-
viduals. That sets us up for you to describe your project a little
bit, including a very brief overview of your issue’s history, the
literature about the topic, your method, and your primary argu-
ments. Then you bring it together in a thesis. For instance, here
are some examples from one of our classes:
• Through the content analysis of twenty trailer remixes, my

project looks at the tools remix trailers use to mimic and
critique the film industry and teach media literacy.
• This paper examines hypermasculinity and homophobia
within Black communities, and how they are portrayed in
media through the film Moonlight.
• Perceptions and use of viral challenge videos on social
media will be analyzed through surveys and focus groups to
understand why people across generations might behave in
unusual ways to gain fame, benefit charity, and fit in.
• This argues that the Mad Men characters Joan Holloway and
Peggy Olson are given a partial empowerment in the work-
place, but that it serves to familiarize the other women with
patriarchy and male hegemony and ultimately strengthen
oppression.
• This study finds that as societal views of homosexuality lib-
eralized, the frequency of positive reviews of Will & Grace
remained largely unchanged, but their relative comfort with
and praise of the show’s gay premise increased. This praise,

however, is in nostalgic hindsight of what the show once
was, not of its current state. Thus, to keep its series reputa-
tion as a landmark, the Will & Grace reboot must engage in
more counterhegemonic content, such as expanded repre-
sentation and cultural critique.
These are specific to a particular class, obviously (Popular Cul-

ture and Society), but you can hopefully see how this same
approach can be applied to history, business, sociology, litera-
ture, political science, and many other fields.
Once you’ve hooked us, you can carry us through your liter-
ature review, methods, and data. Remember that your lit review
doesn’t stop at that section of the paper, though. You can and
should integrate your sources throughout the other sections,
because you’re in conversation with them. After you’ve com-
pleted those parts, it’ll feel like you’re done but also that some-
thing’s missing. We like to call that the “so what” question. In
other words, “okay, so now I’ve read all this. So what?” This is
where we get to what’s called the “discussion” section of the
paper, which has two goals: (1) to show why the project was
important, how it added to the literature, and how it could be
applied, and (2) to show that the project wasn’t perfect, but
had some very real limitations due to sample, method, or other
issues that could be solved in further research (by you or some-
one else). This takes the place of what most people are taught
is the “conclusion” of a paper, in which teachers often tell you
something like “go back to your introduction and say that stuff
again.” Why anyone gives that advice is beyond us. Why would
you say something that you just said fifteen pages earlier? Again,
please don’t. The best thing about that sort of conclusion is
that you can skip it. We’re not saying you don’t conclude (often
by returning to your “hook”), and we’re not saying that you
don’t connect your conclusions (the answers to your research
questions) back to the results/data. But that isn’t repetition; it’s

extension and reflection.
OK, so at this point you might be saying “am I done yet?”
Sorry, but no. While everything is hopefully structured well and
written well, you still have a ways to go. There are three things
that you should address here:
1. Edit for tone. One of the biggest mistakes that even pro-
fessionals make is that they start ranting in their research.
They’re so passionate about their work, and so convinced
by their results, that they dive into what we call “polemic,”
which basically means a kind of verbal attack. This is not only
alienating for your reader, but it’s also a clear sign that you
haven’t given yourself the kind of distance necessary to eval-
uate your project for bias, errors, and unexpected results.
2. Edit for content. You’re probably exhausted by your proj-
ect at the end. So put it down, walk away, and come back
later to edit. You’ll find that you can probably cut a lot of
it out (maybe more than half of it), and tighten the rest of
it up. You also might find that there are some big chunks
missing that you thought you said because they were in your
head, but that you didn’t actually say. The best way to do
this is to have a clear word count for your paper, and write
double that amount in the first pass. If you do your literature
review and your data analysis right, that’s the very least you
should have. If you haven’t written too much, you’ve writ-
ten too little. And if you’ve written too little, you definitely
won’t have a good paper.
3. Peer review! After you edit it yourself, read it aloud to a friend.
That probably sounds horrible, and we understand that. But
if you want to catch all the places you didn’t make sense, read
it out loud to them rather than having them read it. That way
you’ll catch the problems with grammar and structure, but
you’ll also be able to have them stop you and say “hold on.
What? What are you talking about? I’m lost.” That’s the best
thing they can possibly do for you, so buy them pizza or ice
cream as a thank you. If you can then send the paper to some
friends that know something about your method (maybe
someone in your class, or another graduate student or col-
league), do that as well. They’ll give you some valuable insights
that will make the project much stronger, or at least make you
feel a little better about writing something strong.
There’s one final step that is usually optional. This is to present

your research and maybe publish it. That seems daunting, but
it’s also one of the most satisfying things you’ll ever do. Here’s
a quick primer on those things:
1. Presenting: A research presentation is often incredibly bor-

ing, but it doesn’t have to be. Reading your paper is one
way to do it, never looking at the audience and using a Pow-
erPoint that no one will look at unless their phone is dead.
We don’t recommend it. Instead, tell your audience the
story of your project, data, and conclusions. Use the same
hook as your paper does. Show us that it matters. Explain
why it’s so interesting to you, and bring the passion that you
had when you started the project. What are your research
questions, and why are they important? How did the litera-
ture review help shape how you understood the topic, and
why did it turn into your method and sample? What did you
find out, and what are the implications? This could happen
with a beautiful PowerPoint, and that’s not a bad way to go.
It could also happen with just you standing and speaking
earnestly to your audience about something that makes you
passionate. Don’t read your paper, don’t carry note cards,
and don’t look back at your PowerPoint on the screen. Just
be open and honest with us, and take us along for the ride.
That’s it. Drop the mic and go eat a taco.
2. Poster sessions: Presenting a poster at a conference is a

very awkward, but yet low-commitment way to discuss your
research. Basically you make a big poster in which you dis-
play all the elements of your research, focusing the visuals
on your results and why they matter. There are lots of ways
to organize academic posters, and we’ve given you some
ideas in the online resources. But the key again is to ask
good questions, show them the answers you found, how
you found them, and what they mean, and to do it in a way
that is visually engaging. You’ll also in some cases have to
stand next to your poster and talk to people about it, which
can be a great way to get to know people at a conference.
It’s also great practice for “pitching” your project (which
basically means doing a fifteen-minute presentation in fif-
teen seconds). Go eat another taco.
3. Publication: Finally, the holy grail of research. This hon-
estly isn’t that bad, although it sometimes feels so stressful
that people are paralyzed by the thought. Basically what
you’re doing is taking your paper, fitting the length and
citation format to what a journal asks you for on their web-
site, and then sending it to them. They send it out to a set
of scholars in your field who will read it, respond, and then
either accept it outright (which almost never happens), ask
you to revise it either extensively or just a little, or reject it.
Hopefully they’ll give you good comments at that point.
Some people don’t bother to revise if they get a rejection,
because most papers are rejected even if they’re good; there
are only so many articles a journal can print. So there’s a
case to be made for just sending it to another journal at
that point and trying your luck there. This whole process
can take a few months, but it could also easily last a few
years. That’s why people get stressed about it. Librarians
can help with this process and make it a little less stressful;
after all, it’s what they do. One trick is to send your article
to a journal that published something that you cited in your

own paper, and that you admired. It shows that you’re inter-
ested in what they’re interested in, and it shows that you
know what kinds of things the journal publishes. It’s like
going to a party that someone has invited you to rather than
walking in off the street when you hear loud music playing.
Just hope they have tacos.
This is a quick guide, to be sure. However, we hope it’s a useful

one that shows you that research can be an exciting, interesting
experience. Rather than dragging along from step to step, you
can become so invested that it feels like you’re just responding
to clickbait on social media. Research should flow, even if some-
times it gets confusing or frustrating. What good conversation
doesn’t present some sort of a challenge? If it doesn’t make you
think or make you feel, chances are it won’t be worth remem-
bering. We hope that this chapter, and this book, will help you
get another step closer to writing something that impacts others
in that way. And if it does, let us know and send us the article so
that we can be impacted too. Maybe we’ll even buy you a taco.
Glossary
annotated bibliographies are bibliographies with added notes

about each source
ANOVA (ANalysis Of VAriance Test or ANOVA F-test)
determines significance across groups
archival research uses an institution’s records about an event/
text/process to determine how it was created or discussed
archives are physical places usually located in a library or
museum
auto-ethnography is the study of the self
case study is the study of a specific instance to explain a broader
trend or issue
categorical variable (nominal) is a variable with a specific
amount of optional responses, not necessarily in logical
order
content analysis is a brief reading of texts with the aid of explicit
“coding”
convenience sampling selects people close by or who you know
correlation is any mutual association between two or more
variables
critical paradigm asks why things are happening in relation to
the world more broadly
182
Glossary
dependent variable is the variable that changes based on the

variable movement of the independent variable
discourse analysis (critical discourse analysis) is a brief read-
ing of a large number of different kinds of texts to discuss
the ideological landscape surrounding a central topic
empirical paradigm is the belief that the topic studied is discov-
erable and often measurable data with numbers
ethnography is the investigation of the discourses, events, rit-
uals, and communications with a community or a set of
individuals
experimental research is used within a controlled environment,
manipulating variables in one group in order to determine
changes
focus groups are small groups of individuals that are questioned
collectively to gain specific information about a topic
independent variable is the variable selected by the research
whose variation does not depend on other variables
in-depth interviews (IDIs) are structured, semi-structured, or
unstructured interviews often conducted with selective par-
ticipants and cover great depth
interpretive paradigm looks at individuals’ perception, por-
trayal, and understanding of their social world
linear regression examines how predictor variable(s) influ-
ence(s) an outcome—dependent variable(s)
literature review is a review of scholarly literature on a topic
mean (M) is the average numeric value of all the data within a
variable
median (Mdn) is the numeric value in the middle when the data
is ordered from lowest to highest value
metasynthesis brings together scholarly sources to examine
their results all together
mode is the numeric value that appears the most frequently
multiple regression allows inclusion of different prediction
variable to determine outcome of another variable
183
Glossary
Number of cases (N) is the number of participants or data

observational analysis is observing something in the real world
and then drawing conclusions from it
participant observation is when the researcher is both observ-
ing and a part of what they are observing
participatory-action research (PAR) focuses on working
directly within communities
perception/reception approach is how people interpret, view,
and react to things
portrayal approach is how things are constructed or shown
probability sampling method randomly selects participants
p-test tells how significant the differences/correlations are
purposive sampling is when the researcher strategically decides
people with meaningful responses
qualitative (type) analysis is study of qualities of texts
quantitative (number) analysis is counting something in or
across texts
random sampling method gets a representative sample of the
population
range is the lowest and highest number values in the data set
reliability refers to how consistent each question within the
scale is with other questions
research is answering interesting questions rather than agoniz-
ing work 3
skewness (skewed distribution) is when there is an uneven
amount of scores either higher or lower
snowball sampling is for a difficult to locate population
standard deviation (SD) is the incremental distance each data
point is from the mean and the other data points
surveys are taken with a questionnaire given to sample of indi-
viduals for their perspective on a topic
systematic sampling selects one out of every three people
textual analysis/close textual analysis (multimodal semiot-
ics) is an in-depth reading of text or texts
184
Glossary
theory is a way to observe your surroundings to come up with

explanations for how things work and what they mean
truth is someone’s perception, often conveyed through
conversation
t-test determines if there’s a difference in mean scores between
two groups
validity determines if questions actually represent things they
should represent
Index
Page numbers in italics indicate figures.
abstract, and writing research archival research 60; comparison

papers 174 of methods and 46; example
academic research see research; 61; historians and 90 – 91;
research questions (RQs) librarians’ support 61 – 62,
action and activism plans 136 – 137 92 – 93; limitations 61;
advertisements: content metasynthesis and 60
analysis and 86 – 8 7; textual archives 61; librarians’ support
analysis and 77 – 7 8; see also and 61 – 62, 92 – 93; see also
specific advertisements archival research
Alcoholics Anonymous (AA) 101 arts: discourse analysis and 95;
Amazon, and Mechanical textual analysis and 75 – 76
Turk 122 Associated Press 22 – 23
American Psychological Association auto-ethnography 63; ethics and
(APA) 22 – 23, 40, 129 – 130 63 – 64; impact and evidence
annotated bibliographies for projects and 10 – 11, 62 (see
24; scholarly sources and also ethnography); scholars
6 – 7 , 24 – 25, 26; see also and 80
bibliographies awkward pause, and in-depth
ANOVA (ANalysis Of VAriance interview 111
Test or ANOVA F-test) 172;
quantitative data analysis and bad sources 15 – 16, 21
164, 165, 166 Bechdel, Alison 87
APA (American Psychological Bechdel Test 87 – 88
Association) 22 – 23, 40, bias: ethnography and 100,
129 – 130 102, 117, 118; focus groups
186
Index
and 115, 117, 118; in-depth citations: citation export tools

interviews and 103 – 104, 105, 23, 40; scholarly sources and
111 – 112, 117, 118 22 – 23
bibliographies: citation closed questions 124 – 125
styles and 22; scholarly close textual analysis see textual
sources and 22; see also analysis/close textual analysis
annotated bibliographies are (multimodal semiotics)
bibliographies with added cloud, the 23
notes about each source codes and coding scheme:
Big Little Lies (television show) content analysis and 82 – 85,
147 – 148 89 – 90, 100; discourse analysis
“bipolar trap,” and in-depth and 93 – 94, 96, 97; textual
interview 111 analysis and 100
Black Lives Matter collaboration, and participatory-
demonstration 73 action research (PAR) 137
blind men and the elephant conclusion, and writing research
parable 31, 91 papers 176 – 177
“blue eyes brown eyes” study 64 content analysis 108, 50;
body language, and in-depth advertisements and 86 – 87;
interviews 115 Bechdel Test and 87 – 88;
body of research papers 174 – 176 body of research paper and
Boston, Bernie 73 – 74 175; codes and coding
brand perception, and discourse scheme 82 – 85, 89 – 90, 100;
analysis 95 – 96 comparison of methods and
business studies, and discourse 44; example 50; Excel and
analysis 94 84, 85; IRL work situations
67, 85 – 86; limitations 50,
Carroll, Dave 95 90; number, depth, sample
case studies: comparison of and 108, 82; overview and
methods and 46; example 60; objectives 81; and people,
limitations 60 study of 88 – 89; qualitative
categorical questions 124 – 126, (type) analysis and 81, 82;
125, 126, 127 qualitative data analysis
categorical variable (nominal) and 147, 148, 150, 151;
163; t-test and 164 quantitative (number) analysis
central tendency measures and 81, 82; social media and
161 – 162 83 – 84, 88; SPSS and 85;
cherry-picking 12, 20, 21, 29, television shows and 87 – 88;
49, 50, 93 (managed) textual data and
Chicago Manual of Style 22 – 23 147, 148, 150, 151
187
Index
continuous variable 163, 164, descriptive statistics, and

166 quantitative data analysis
control group, and experimental 158 – 162, 159, 161
research 139 – 140 design and flow of surveys 132
convenience sampling 121 – 122 discourse analysis (critical
correlation 172; correlation discourse analysis) 108,
between variables 50 – 51; arts and 95; brand
167 – 168; negative (inverse) perception and 95 – 96;
correlation 168; positive business studies and 94; codes
correlation 167 and coding scheme 93 – 94,
Cosby Show, The (television show) 96, 97, 100; comparison of
77, 87 methods and 44 – 45; creation/
CRAP test 15, 16 – 17, 18, 19 portray/representation
creation/portray/representation approach and 50 – 51, 91 – 92;
approach: discourse analysis example 51; and history
and 50 – 51, 91 – 92; “text” studies, critical 94 – 95; IRL
and 48; textual analysis and work situations 67 – 68, 73 – 74;
49; see also portrayal librarians’ support 92 – 93;
approach limitations 51; number, depth,
critical discourse analysis sample and 108; overview and
(discourse analysis) see objectives 82, 96; qualitative
discourse analysis data analysis and 150;
critical paradigm 7; embodiment timeframe and 90
of change and 137; IRL Pepsi DMV offices 78 – 79
advertisement 74; topics and Dolce & Gabbana 77 – 78, 86
6 – 7, 44 – 47 Doodle 116
duplication of research, and
data analysis see qualitative data methods for analysis 43
analysis; quantitative data
analysis EasyBibb 23
data and data sources: EbscoHost 37
metasynthesis and 141; theory editing, and writing research
and 142, 143 papers 177
De Beers 87 Ehrenreich, Barbara 80
Deloitte services network 122 Elliott, Jane 64
demographic questions 124 empirical paradigm 5; critical
dependent variable 5, 171; paradigm and 7; independent
quantitative data analysis and variable 5, 7; topics and 6 – 7,
163, 164, 165 44 – 47
Descript 54 EndNote 23
188
Index
ethics: auto-ethnography and 65, 140 – 141; experimental

63 – 64; “ethics alerts” 42; group and 139 – 140; IRB
ethnography and 63 – 64; (Institutional Review Board)
experimental research and and 65, 138 – 139; IRL work
64 – 65, 138 – 139; IRL situations 69; limitations
work situations and 52 – 53; 65 – 66; number, depth, sample
observational analysis and and 11
59 – 60; scholarly sources and
22
field notes: ethnography and
ethnography 10 – 11, 62; bias
62 – 63, 99 – 100; in-depth
and 100, 102, 117, 118;
interviews and 114
codes and coding scheme
and 100; ethics and 63 – 64; films, as “text” 48; see also specific
example 62 – 63; field notes films
and 62 – 63, 99 – 100; IRB flow: and design of surveys 132;
(Institutional Review Board) research and 180
and 64, 101, 102, 117 – 118; “Flower Power” photograph
IRL work situations 69; 73 – 74
librarians’ support 117 – 118; focus, and research 3, 5, 6 – 7, 8,
limitations 63; non-participant 13
observation 101 – 102; focus groups 10, 55, 98; bias and
number, depth, sample and 115, 117, 118; comparison
11; overview and objectives of methods and 45; example
98, 99 – 100; participant 55 – 56; groupthink and 116,
observation 98, 102, 117, 136; 117; individual responses and
textual analysis as compared 116 – 117; interview guides
with 80; “thick description” and 116; IRB (Institutional
and 80; topics and 47; truth Review Board) 117 – 118; IRL
and 99, 103, 107; undercover
work situations 68; librarians’
ethnography 100 – 101; see also
support 117 – 118; limitations
auto-ethnography
56; number, depth, sample and
Evans, Ieshia 73
Excel: content analysis and 84, 10; online signups and 116;
85; probability sampling and overview and objectives 98;
120; quantitative data analysis perception/reception approach
and 156, 164 and 55 – 56; qualitative data
experimental research 11, analysis and 146, 151; truth
64, 135; control group and 99
and 139 – 140; ethics and formative research 52
64 – 65, 138 – 139; example frequencies, and statistics 158
189
Index
F-test see ANOVA (ANalysis Of 54, 113, 117 – 118; IRL

VAriance Test or ANOVA [F] work situations 68, 103 – 104;
test) leading questions problem
and 111 – 112; librarians’
Gallup, Inc. 122 support 117 – 118; limitations
Geertz, Clifford 80 55, 115 – 116; number, depth,
Google Docs 40 – 41 sample and 108 – 10, 11 – 12;
Google Drive 23 overview and objectives 98;
Google Scholar 18 – 19, 24 participant observation and
(managed) grounded theory, 117; qualitative data analysis
and qualitative data analysis and 146, 151; questions and
151 – 153 107 – 108, 108, 109, 110,
groupthink, and focus groups 110 – 112; recording interviews
116, 117 and 113 – 114; recruiting
participants and 112 – 113;
historians 48, 72, 90 – 91 rehearsals 112; transcribing
historiography 91 interviews 115; truth and 99,
history studies, critical 94 – 95 103, 107; willing participants
and 113
IDIs (in-depth interviews) see industry publications, and
in-depth interviews (IDIs) sources 15 – 16
ILL (Interlibrary Loan) 19, 40 inferential statistics 162 – 164,
independent variable 5, 163, 171; 167, 169
ANOVA and 166; empirical in real life (IRL) work situations
paradigm and 5, 7; linear see IRL (in real life) work
regression and 169, 172; tests situations
and variables 165; t-test and Institutional Review Board (IRB)
163 – 164 see IRB (Institutional Review
in-depth interviews (IDIs) Board)
108 – 10, 53, 98; awkward Interlibrary Loan (ILL) 19, 40
pause and 111; bias and interpretive paradigm: dependent
103 – 104, 105, 111 – 112, variable 5; topics and 6 – 7,
117, 118; “bipolar trap” 44 – 47; truth and 99; see also
and 111; body language portrayal approach
and 115; comparison of interview guides 105 – 107, 116
methods and 45; example interviews see in-depth interviews
53 – 54; field notes and 114; (IDIs)
interview guides and 105 – 107; inverse (negative) correlation,
interview skills and 102; IRB and quantitative data
(Institutional Review Board) analysis 168
190
Index
IRB (Institutional Review leading questions, and in-depth

Board): ethnography and interview 111 – 112
64, 117 – 118; experimental LexisNexis 37
research and 65, 138 – 139; LibGuides 35, 130
focus group and 117 – 118; Library, Information Science, and
in-depth interviews and Technology Abstracts (LISTA)
54, 113, 117 – 118; IRL with Full Text 39 – 40
work situations and 53; Library & Information Science-
librarians’ support 117 – 118; specific database 39 – 40
perception/reception Likert-type scale questions 123,
approach and 55 – 56 125 – 126, 126, 127, 132
IRL (in real life) work linear regression 169, 172
situations: comprehensive LISTA (Library, Information
report on company Science, and Technology
worldwide 4; content analysis Abstracts) with Full Text
67, 85 – 86; critical paradigm 39 – 40
74; discourse analysis 67 – 68, literature review 29; annotated
73 – 74; ethics and 52 – 53; bibliography vs. 35; approaches
ethnography 69; experimental from topic to literature review
research 69; focus groups and 32, 33 – 34; citation
68; in-depth interview 68, export tools 40; conversation
103 – 104; job promotions about an issue and 29 – 30;
and 21, 133 – 134; literature conversations about an issue
review and 27 – 28; and 31; IRL work situations
perception/reception 27 – 28; Josh’s method 37 – 39,
approach 52 – 53; research 40, 41; librarians’ support 35;
and 4, 21; scholarly sources Maridath’s method 39 – 41;
and 21; surveys 69, 122 – 123, organization of literature
133 – 134; textual analysis 67, and 31 – 32; overview and
73 – 74, 79 objectives 26, 27, 29;
irrelevant sources 17 – 18, 20 qualitative data analysis and
151 – 152; setting the stage and
Japanese culture 79 31 – 32; Ted’s method 36 – 37,
JD Power and Associates 122 39; themes and 37, 38 – 39,
Jenner, Kendall 73 40 – 41
job interviews 103 – 104
journalism 16 – 17, 23 Mad Men (television show) 175
Magritte, René 75 – 76
Kimata, H. 140 – 141 mean (M) 159, 171; quantitative
Kinsey, Alfred 136 data analysis and 160, 161
191
Index
Mechanical Turk 122 analysis/close textual analysis

median (Mdn) 171; quantitative (multimodal semiotics)
data analysis and 159, 161, Microsoft Excel see Excel
162, 170 Microsoft’s OneDrive 23, 36
metasynthesis 11, 60, 135; MLA (Modern Language
archival research and 60 – 62; Association) 22 – 23
data and data sources 141; mode 159, 171; quantitative data
example 142; and methods, analysis and 162, 170
different 141 – 142; number, Modern Language Association
depth, sample and 11 (MLA) 22 – 23
methods for analysis: case studies Moonlight (film) 175
and 46, 60; choices and multimodal semiotics (textual
comparison of methods 6 – 7, analysis/close textual analysis)
43, 44 – 47, 48; different way 74 – 75; see also textual
using different methods and analysis/close textual analysis
43; duplication of research and
(multimodal semiotics)
43; IRB (Institutional Review
multiple regression 169,
Board) and 53, 54, 55 – 56,
170, 172
65; IRL work situations and
52 – 53, 67 – 69; librarians’
negative (inverse) correlation,
support and 58 – 59, 61 – 62;
and quantitative data analysis
number, depth, sample and
168
108 – 11, 11 – 12; pet methods
Netflix show 12 Reasons Why
and 58 – 59; research questions
(RQs) and 8; scholarly sources study see 12 Reasons Why
and 24; “text” and 48; 12 (Netflix show) study
Reasons Why study 43, 50, newspaper articles, and sources
51, 55 – 56, 57, 60, 62, 65; 16 – 17, 23
see also archival research; Nielsen Media box 133
content analysis; creation/ non-participant observation
portray/representation 101 – 102
approach; discourse analysis non-probability sampling 120,
(critical discourse analysis); 121 – 122
ethics; ethnography; focus nonprofit organizations project:
groups; in-depth interviews scholarly sources and 15 – 16,
(IDIs); observational analysis; 21; topics 6 – 7
perception/reception NoodleTools 23
approach; portrayal approach; “normal distribution” 158, 159
research; surveys; textual null hypothesis 166
192
Index
Number of cases (N) 159, 171; Pepsi advertisement 73 – 74

quantitative data analysis percentages, and statistics 158
and 160 perception/reception approach
NVivo data analysis software 54, 42, 51 – 52; ethics and 52 – 53;
153 – 154 formative research and 52; IRL
work situations and 52 – 53;
Obama, Barack 83 topics 44 – 47; see also in-depth
observational analysis 59; ethics interviews (IDIs)
and 59 – 60, 63 – 65; “text” and persuasion, and writing research
59; see also archival research; papers 173 – 174
case studies; ethnography; pet methods 58 – 59
experimental research; philosophy, and theory 143
metasynthesis; participatory- places, and textual analysis
action research (PAR); theory 76 – 77
OneDrive 23, 36 population vs. sampling
out-of-date sources 17 – 18 120, 121
Owen, W. F. 152 portrayal approach 42; “text”
and 48, 57; topics 44 – 47;
paradigms 5, 6 – 7, 7 – 8, 44 – 47; see also creation/portray/
see also critical paradigm; representation approach;
empirical paradigm; interpretive paradigm
interpretive paradigm positive correlation, and
participant observation 136; quantitative data analysis 167
ethnography and 98, 102, poster sessions, and research
117, 136; focus groups and papers 173, 179
117; participatory-action post-structural discourse
research (PAR) 136 analysis 91; see also discourse
participatory-action research analysis (critical discourse
(PAR) 135; action and analysis)
activism plans and 136 – 137; PowerPoint presentation 48, 191
collaboration and 137; presentation of research papers
example 137 – 138; outcomes 178 – 179
and 137; participant prisons study 64 – 65
observation and 136 probability sampling method
peer review: scholarly sources 120 – 121
and 15, 16 – 17, 104; writing Prolific Academia 122
research papers and 177 – 178 p-test 172; quantitative data
people, study of: content analysis analysis and 167, 170;
and 88 – 89; textual analysis significance level (p-value) test
and 76 and 164, 166
193
Index
publication, of research papers 165, 166, 169, 171; inferential

179 – 180 statistics and 162 – 164, 167,
purposive sampling 120, 169; linear regression and
121 – 122 169; mean (M) and 159, 160,
161, 171; median (Mdn) and
qualitative (type) 82; content 159, 161, 162, 170, 171;
analysis and 81, 82 mode and 159, 162, 170, 171;
qualitative data analysis: multiple regression and 169,
overview and objectives 170, 172; negative (inverse)
146, 147; content analysis correlation and 168; “normal
and 147, 148, 150, 151; distribution” and 158, 159;
discourse analysis and 150; null hypothesis and 166;
focus groups and 146, 151; Number of cases (N) and
(managed) grounded theory 159, 160, 171; overview and
and 151 – 153; in-depth objectives 156; percentages
interviews and 146, 151; and 158; positive correlation
literature review and 151 – 152; and 167; range and 159, 160,
NVivo data analysis software 171; regression 167, 169 – 171;
153 – 154; silences in data and relationship between concepts
154 – 155; software programs and 166 – 171, 171 – 172;
and 153 – 154; textual analysis significance level (p) test and
and 146, 150, 151; (managed) 164, 166, 167, 170, 172;
textual data and 147 – 151; skewness (skewed distribution)
themes and 147, 152 160, 161, 162, 171; SPSS and
Qualtrics 123 156, 164; standard deviation
quantitative (number) analysis (SD) 159, 160, 171; statistical
82; content analysis and 81, 82 terms 159, 171 – 172; statistics
quantitative data analysis: and 157 – 158; t-test and
ANOVA (F) test and 164, 163 – 164, 165, 170, 172;
165, 166, 172; categorical variables described 158 – 162,
variable (nominal) 163, 164; 159, 161
central tendency measures questions, and in-depth
161 – 162; continuous variable interviews 107 – 108, 108, 109,
163, 164, 166; correlation 110, 110 – 112; see also research
and 167 – 168, 172; dependent questions (RQs)
variable 5, 163, 164, 165, questions, and surveys:
171; descriptive statistics and categorical questions 124 – 126,
158 – 162; Excel and 156, 125, 126, 127; closed
164; frequencies and 158; questions 124 – 125; common
independent variable 163 – 164, survey question mistakes
194
Index
131, 131 – 132; demographic sampling, and research 12, 21

questions 124; Likert-type sampling, and surveys:
scale questions 123, 125 – 126, convenience sampling
126, 127, 132; open questions 121 – 122; Excel and 120;
and 130 – 131; overview and non-probability sampling 120,
objectives 123 – 124; scaled 121 – 122; population vs. 120,
questions 124; semantic 121; probability sampling
differential questions 126, 127 120 – 121; purposive sampling
120, 121 – 122; random
random sampling method 121 sampling 121; snowball
range 159, 171; quantitative data sampling 122; SPSS and 120;
analysis and 160 systematic sampling 121
recording interviews 113 – 114 scaled questions 124
recruiting participants, and scholarly sources: annotated
in-depth interviews 112 – 113 bibliographies and 6 – 7,
regression, and quantitative data 24 – 25, 26; bad sources and
analysis 167, 169 – 171 15 – 16, 21; bibliography
rehearsals, for in-depth interviews of sources and 22; cherry-
112 picking and 20, 21, 29;
reliability; cherry-picking and 12, citation export tools and 23;
21; dependent variable and citations 22 – 23; CRAP test
5; flow and 180; focus and 15, 16 – 17, 18, 19; ethics
3, 5, 6 – 7, 8, 13; IRL work and 22; Google Scholar
situations 4, 21; librarians’ and 18 – 19, 24; industry
support and 12, 13; methods publications and 15 – 16;
overview 8; number, depth, Interlibrary Loan (ILL) and
sample and 8, 108 – 11, 11 – 12; 19, 40; IRL work situations
overview and objectives 2, 3, and 21; irrelevant sources and
180; paradigms and 5, 6 – 7, 17 – 18, 20; librarians’ support
7 – 8; research questions (RQs) 15, 16 – 17, 18 – 19, 22 – 23;
and 2, 5, 6 – 7, 8; sampling methods for analysis and 24;
and 12, 21; starting research newspaper articles vs. 16 – 17,
projects 3 – 4; timeframe and 3, 23; nonprofit organizations
4 – 5; topics and 5, 6 – 7; see also project and 15 – 16, 21; out-
methods for analysis; truth; of-date sources and 17 – 18;
writing research papers overview and objectives 14,
research questions (RQs) 2, 5, 15; peer review 15, 16 – 17,
6 – 7, 8 104; topics and 17 – 18, 19,
RQs (research questions) 2, 5, 24; Wikipedia and 18, 19; see
6 – 7, 8 also metasynthesis
195
Index
SD (standard deviation) see see also dependent variable;

standard deviation (SD) independent variable; variables
semantic differential questions SurveyMonkey 123
126, 127 surveys 10, 56 – 57; comparison
sex lives research 136 of methods and 45 – 46;
significance level (p-value) test example 57; flow and design
164, 166; see also p-test of surveys and 132; IRL
silences in data, and qualitative work situations 69, 122 – 123,
data analysis 154 – 155 133 – 134; librarians’ support
situations, and textual analysis 58 – 59, 129 – 130; limitations
78 – 79 57 – 58; number, depth,
skewness (skewed distribution) samples and 8, 10, 11;
171; quantitative data analysis overview and objectives 11,
and 160, 161, 162 118, 119 – 120; pet methods
snowball sampling 122 and 58 – 59; survey types 133;
social media: body of research television shows and 133; see
paper on 175; content analysis also questions, and surveys;
and 83 – 84, 88 sampling, and surveys
sources see scholarly sources systematic sampling 121
Speechlogger 54
SPSS: content analysis and 85; television shows: body of
probability sampling and 120; research papers and 175 – 176;
quantitative data analysis and content analysis and 87 – 88;
156, 164 surveys and 133; textual
standard deviation (SD) 159, analysis and 77; see also specific
171; quantitative data analysis television shows
and 160 “text” could be a poem, novel,
statistics: categorical variable photograph, film or person
(nominal) 163; continuous 48, 59
variable 163; descriptive textual analysis/close textual
statistics 158 – 162; frequencies analysis (multimodal semiotics)
and 158; inferential 108, 74; advertisements
statistics 162 – 164, 167, and 77 – 78; art and 75 – 76;
169; percentages and 158; codes and coding scheme
p-test and 164, 167, 170; 100; comparison of methods
quantitative data analysis and and 44; creation/portray/
157 – 158; significance level representation approach and
(p-value) test and 164, 166; 49; ethnography as compared
statistical terms 159, 171 – 172; with 80; example 49; IRL
t-test 163 – 164, 165, 170, 172; work situations 67, 73 – 74,
196
Index
79; limitations 49, 81, 82; topics: approaches to literature

multimodal semiotics defined review from 32, 33 – 34;
74; number, depth, sample nonprofit organizations project
and 108; and people, study of 6 – 7; paradigms and 6 – 7,
76; overview and objectives 44 – 47; research and 5, 6 – 7;
72 – 73, 74 – 75; places and research questions (RQs) and
76 – 77; qualitative data analysis 6 – 7; scholarly sources and
and 146, 150, 151; situations 17 – 18, 19, 24
and 78 – 79; television shows transcribing interviews 115
and 77; “thick description” Trint 54
and 80, 96 truth is 99; in-depth interview
(managed) textual data, and and 99, 103, 107
qualitative data analysis t-test 172; quantitative data
147 – 151 analysis and 163 – 164, 165,
thematic analysis, and qualitative 170, 172
data analysis 147, 152
themes: literature review and 37, undercover ethnography
38 – 39, 40 – 41; qualitative data 100 – 101
analysis and 147, 152 United Airlines 95 – 96
theory 135, 142 – 143; data and United States culture 79
data sources and 142, 143;
example 143 – 144; philosophy validity 128
and 143 validity and reliability, and
“thick description” 80, 96 surveys 127 – 130, 128
12 Reasons Why (Netflix show) variables: ANOVA F-test 164,
study: case studies and 165, 166, 172; categorical
60; content analysis and variable (nominal) 163;
50; discourse analysis and central tendency measures and
51; ethnography and 62; 161 – 162; continuous variable
experimental research and 163, 164, 166; correlation
65; focus groups and 55 – 56; between 167 – 168; description
methods for analysis and 43, of 158 – 162, 159; “normal
48; surveys and 57; (managed) distribution” 158, 159; skewed
textual data and 147, 148 distribution 160, 161, 162,
timeframe: discourse analysis 171; tests and variables 165;
(critical discourse analysis) 90; t-test and 163 – 164, 165,
research and 3, 4 – 5 170, 172; see also dependent
Tinder 78 – 79 variable; independent variable;
titles, and writing research statistics
papers 174 Vietnam War protesters 73 – 74
197
Index
WeJoinIn 116 180; peer review 177 – 178;

Wikipedia 18, 19 persuasion and 173 – 174;
Will & Grace (television show) poster sessions and 173,
97, 175 – 176 179; presentation of paper
willing participants, and in-depth and 178 – 179; publication of
interviews 113 paper 179 – 180; titles
writing research papers: abstract and 174
and 174; body of paper and
174 – 176; conclusion and Zimbardo, Philip 64 – 65
176 – 177; editing and 177; Zootopia (film) 78 – 79
overview and objectives 173, Zotero 23, 36

Ted Gournelos - Joshua R. Hammonds - Maridath A. Wilson - Doing Academic Research - A Practical Guide To Research Methods and Analysis-Routledge (2019)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ted Gournelos - Joshua R. Hammonds - Maridath A. Wilson - Doing Academic Research - A Practical Guide To Research Methods and Analysis-Routledge (2019)

Uploaded by

Copyright:

Available Formats

Doing Academic

Doing Academic Research is a concise, accessible, and tightly

Joshua R. Hammonds is Assistant Professor at Rollins

Maridath A. Wilson is Assistant Professor and E-Resources

1 Learn to love, not dread, research 2

3 Now what do I do?

5 Close textual and thematic analysis 72

6 Content and discourse analysis 81

7 Ethnography, interviews, and

9 Observational methods in practice

10 Analyzing qualitative data and

11 Analyzing quantitative data and

12 Persuasion, presentation, and publication 173

1.1 From topic to paper 6

Two professors and a librarian walk

Research doesn’t usually start in an office, or as an assignment, or

OVERVIEW AND OBJECTIVES

This chapter goes through what research actually is

Starting a research project always sounds a lot harder than it actu-

IRL: A BOSS THAT WANTS THE WORLD

Time is an important factor in research because, in theory, all

Figure 1.1 From topic to paper

Method: Textual Method: Method: Content

Figure 1.1 Continued

and reacted to by the people in that room. For example, you

Choosing a paradigm is important because it allows you

• Content analysis (brief reading of a large number of

• Discourse analysis (brief reading of a large number

• In-depth interviews (structured, semi-structured, or

Figure 1.2 Choosing a method

• Focus groups (small groups of individuals that are tar-

• Surveys (a questionnaire given to a sample of individ-

• Ethnography (observation and investigation of the

Figure 1.2 Continued

a community or a (set of) individual(s) within that

• Metasynthesis (brings together a bunch of scholarly

• Experiment (within a controlled environment, manip-

Figure 1.2 Continued

people think about or react to something, you need a high N (the

might want to do interviews with a high D (depth) that could take

A NOTE FROM YOUR FRIENDLY

Librarians know that your first (or tenth!) research project

OVERVIEW AND OBJECTIVES

This chapter teaches you how to effectively and efficiently

Problem 1: Bad sources

“industry” publications. These might give you current perspec-

A NOTE FROM YOUR FRIENDLY

What’s peer review and why should

scholarly journals and newspapers; it also depends on dis-

Problem 2: Out-of-date or irrelevant sources

A NOTE FROM YOUR FRIENDLY

But are Google and Wikipedia BAD?

Many researchers choose to begin their research on Goo-

you to follow the link to the full-text PDF of the arti-

“Cherry-picking” is a pitfall of one’s method and use of primary

IRL: RESEARCH CAN GET YOU

One of the most common pieces of feedback we get from

Keeping a consistently formatted bibliography of sources will

A NOTE FROM YOUR FRIENDLY

Chicago Manual of Style, and the Modern Language Associ-

1. Topic (what’s their general concept?): Don’t try to get too

5. Implications (why does it matter?): Implications are both

For example, in Ted’s paper on disability in theater (topic), he

1 Learn to love, not dread, research 2

5 Close textual and thematic analysis 72

6 Content and discourse analysis 81

12 Persuasion, presentation, and publication 173

1.1 From topic to paper 6