Professional Documents
Culture Documents
MICHAEL J. ALBERS
Uni ver sity of Mem phis, Tennessee
ABSTRACT
INTRODUCTION
Audience analysis in the past could be performed without extracting a high level of
detail about the individual groups because other factors often rendered much of
that detail less than essential. Although the writers recognized the need to address
multiple groups and did provide hierarchies and categories within a document,
they were constrained by a limitation to create a single document. Whether the
document explained to a patient how to use a new drug, how the new building
would benefit a downtown area, or how to operate machinery or software,
economic and technological limitations often mandated a “one size fits all”
document. The audience analysis could have clearly defined in detail the differ-
ences between the various groups who would be using the document, but with a
“one size fits all” approach, specific details about different levels of audience
263
2003, Baywood Pub lishing Co., Inc.
264 / ALBERS
could not be applied. The economics of printing and storing paper meant only one
version and, when we moved the information to the Web, only one online
document was created. Of course, in most of these writing situations, the text was
developed for either the novice user or the low literacy reader (least common
denominator) segment of the audience. The underlying assumption was that
when a document was written for the novice, “the experts will be able to under-
stand it too.” Yet continuing complaints from more experienced users about the
document being so novice-oriented highlights the difference between being able
to understand the document and having the information in an effective format
from both the design and content perspective.
The era of a “one size fits all” single document is drawing to a close. Finally,
we are getting the technology and toolsets to create different documents that
really focus on various users’ specific needs. With text markup (XML), single
sourcing, text databases, and dynamic text generation, we can address a major
problem for technical communication: trying to write to multiple audiences in a
cost-effective yet more individual manner.
Single sourcing strives to develop text elements that can be reused many times
and merged into multiple documents. Rockley [1] describes four stages:
• Stage 1: Same content, different media. Paper manual and online help are
the same.
• Stage 2: Static customized content. Although developed from the same text,
the paper manual, and online help differ.
• Stage 3: Dynamic customized documents. Document gets created on the
fly based on a user profile or user selected options.
• Stage 4: Electronic performance support system provides “just in time”
information to fit the user’s current situational needs.
Most single sourcing projects currently develop multiple static documents from
a common set of documents or from a text database based on templates (stage 2).
The dynamic documents that would most benefit from the multidimensional
analysis I describe fit into Rockley’s stages 3 and 4. The creation of dynamic
documents which support all the dimensions of a multidimensional audience
analysis requires a textual database, but goes beyond populating static templates to
allowing dynamic expansion and contraction of the information to fit the reader
needs. Quite simply, I foresee and want to write in a world where the custom
documents are not just essentially static template-based documents, but ones
that are dynamically built to conform to each reader’s knowledge level, desired
detail level, cognitive ability, and any other situation-specific aspect the audience
analysis has deemed important for that particular user group.
Figuring out how to write for those multiple audiences requires us to reexamine
how we define audience analysis. However, I do not foresee a need for new
methodologies. The current audience analysis methods should work. Actually,
current methods are good at defining what data to collect, but are weak at
MULTIDIMENSIONAL AUDIENCE ANALYSIS / 265
approaches to analyzing and interpreting that collected data. The multi dimen-
sional analysis of this article provides an additional tool for the audience analysis
toolkit. The basic difference is that the analyst must work from a mindset of
constructing multiple dimensions. Once working from that mindset, they should
gain a clearer picture of how the different user groups fit within a situation’s
goals and information needs.
This article works to build a multidimensional model on which to base the
audience analysis. The article first looks at why we need to consider a multi-
dimensional model and then lays out an explanation of the model. Next, it
explains how a multidimensional model can help develop content that better
suits reader’s needs. Finally, it looks at various pitfalls that can plague a multi-
dimensional audience analysis. While acknowledging the need to operationalize
the model, space constraints force me to leave that issue for later work. Actually,
much of the current audience analysis methodology remains applicable to this
model; the major issue is having the analyst interpret the information with respect
to this model.
problematical since the document has to average across all potential readers. In a
future with dynamic custom text generation, extensive upfront audience analysis
becomes essential to creating high quality documents, since neither writer nor
editor will be able to review the complete document before the user sees it.
Most explanations of how to perform audience analysis operate in a single
dimension. Both of the previous lists (and other audience analysis lists) result
in an unordered linear flow of data about the reader. All points that need to be
considered are listed without considering how they are interrelated or how they
affect each other. For example, Burnett gives a form with questions that the analyst
completes [5]; of course, the questions on that form could be rearranged. There
is nothing intrinsic to order the questions asked as part of the data collection
for an audience analysis. Consequently, making sense of the audience analysis
data requires figuring out how each element relates to others and interpreting/
predicting their effect on the reader. The difficulty of both collecting the data and
performing that analysis prevents many writers from using an audience analysis
beyond a superficial level. In this article, I hope to provide a method for making
those interconnections apparent and improve the extent to which technical com-
municators perform and use audience analysis when creating documents.
Fig ure 1. Infor ma tion areas appear along the mul ti ple dimen sions.
A spe cific infor ma tion need maps out a shaded area; the dynamic
con tent gen er a tion must match the text to a par tic u lar reader’s
loca tion within that shaded area. In most real is tic sit u a tions,
more than one shaded area would arise from the anal y sis.
Then, for specific reader goals or information needs, the area of interest can be
mapped out. The writer’s task becomes one of developing content which provides
clear communication throughout the shaded area. As an added gain, rather than
having to work in terms of expert/novice or other classification, the different user
groups and their needs will emerge as part of the analysis.
Figure 1 contains shaded areas which provide examples for specific reader
needs. Although Figure 1 shows the shaded areas as cubes, in reality the actual
shape can vary to fit the information requirements. I can easily envision times
when it would be more of a squished football ball shape or a pyramid with the top
point being the high knowledge and consistent detail level and expanding as the
audience has less knowledge and wants varying detail levels. Figure 1 provides a
good visual of why dynamic content generation will be required to effectively
address this issue. The actual text appropriate for the top/bottom limits of a shaded
area may be very different. On the other hand, this mapping is part of a qualitative
analysis, not a quantitative one. The collected data does not allow for a concise
268 / ALBERS
plot of points that perfectly define a shape, nor do they allow for a precise and
meaningful value to be assigned to each axis. But the overall plot does provide
a clear visual of what information must be provided and what the user expects
to gain from it.
I’ll examine the problem in more detail later in the “Points to Consider” section,
but let me point out that the dimensions are not a direct mapping to the lists most
sources give for audience analysis. Burnett’s list given earlier has six points
(and her list is very short compared to other sources), but should not be taken
as requiring a six-dimensional model. Instead, the related elements in the list
must be examined and grouped into each dimension based on how they relate
to each other.
Knowledge dimension is the subject knowledge the user possesses about the
topic. This influences word choice and determines how much supporting infor -
mation must be provided. Much of the current work in audience analysis tries
to pinpoint user knowledge level.
Subject knowledge differs from cognitive ability in that a person may be highly
literate and can follow a complex explanation, but has minimal knowledge of
the subject. For example, consider a person who has been diagnosed with a chronic
medical condition. This person could be an electrical engineer or a high school
dropout, and yet their levels of specific medical knowledge might be comparable
(very low). However, the most effective way of writing for these two people would
be quite different.
Detail dimension is the amount of detail the user wants about the specific
situation. This can range from basic explanations to highly detailed explanations
about the underlying physical process.
The detail a person wants is independent of their knowledge about the topic.
How do you write content for the person who says “just tell me what I need to
know?” The writing style and word choice used to answer a reader’s question are
connected to the topic knowledge. To present the same level of detail to a person
with a low knowledge level and a person with a high knowledge level, the amount
of text and, potentially, the overall information design would be different because
of differences in the technical level of the writing, supportive text such as the term
definitions, and use of graphics. But, in the end, although the amount of text varies,
the detail level presented remains the same. For example, the verbiage devoted
to term definition and background explanations would be higher for the person
with the lower knowledge level. While these may be essential for understanding,
they do not change the detail level of the document.
MULTIDIMENSIONAL AUDIENCE ANALYSIS / 269
As the person interacts with the text or interacts with the overall situation across
time, the amount of detail they need or want can change. Also, the amount of detail
people want does not remain constant across an entire subject; they want more at
points they deem of interest and less at other points. With the multidimensional
audience analysis, potential changes can be tracked and examined as part of the
information development process.
Continuing with the previous example, the detail level desired by two people
with engineering degrees can vary. They have the same knowledge level, but want
different amounts of information. The first person wants to understand their
condition at a level that approaches that of a physician and another only wants the
basic details. To complicate the detail level even further, the amount of detail
desired can vary within the overall explanation itself. For example, a person may
be content with a high level description of the overall medical condition, but wants
very detailed information on medication side effects and required dietary changes.
allowed for during text development [7]. Failure to allow for social context
consistently appears as one of the reasons which explains systems which function
technically but fail because of lack of user acceptance. A basic problem behind
designs which fail to reach an audience is they tend to be ahistorical and generic.
By ignoring specific social or cultural factors and focusing on requirements
applicable to a generic audience, the text fails to provide information needed to
match user goals and information needs. The generic design allows the people
factors to be ignored, but only at the expense of overall design usability.
Fig ure 2. Map ping of per so nas for mul ti di men sional anal y sis.
The shaded areas would come from the audi ence anal y sis and the
named points are per so nas cor re spond ing to each area.
Of course, if scenarios and personas are both used or if other audience analysis
methods are used, they should map into the same areas. If they don’t, then at least
one of them is missing a major information requirement. This type of mapping
also provides a means to respond to statements of “someone might want this
information.” Once the characteristics of that “someone” are more clearly defined,
they can be shown to be inside the shaded area, and thus already considered,
or outside the shaded areas. If they are outside, then the “someone” is either
outside the intended audience or an audience element has been missed. Thus,
adding this analysis step provides an additional check on ensuring the overall
analysis matches reality.
Achieving all of these points requires a certain level of creativity on the user’s
part, but it’s a creativity that must be supported from the beginning of the
development project by the audience analysis to assist in the transformation of
information into knowledge. A well-designed system allows a user to get on with
mentally understanding and analyzing information, while a poorly designed one
bogs a user down in simply locating it. For example, one system provides an
integrated text while another provides a collection of short articles, each primarily
on one particular topic but with significant information duplication and formatted
differently (think of aWeb page that links to several different sites).
POINTS TO CONSIDER
This section discusses three points to consider when developing the multi-
dimensional audience analysis model: 1) create independent dimensions; 2) create
non-binary dimensions; and 3) analyze a full information space.
MULTIDIMENSIONAL AUDIENCE ANALYSIS / 273
Binary dimension only have two choices. The expert/novice distinction com-
mon in audience analysis is one such binary. Most users are neither expert nor
novice and their skill level varies with different aspects of the situation. Likewise,
divisions based purely on aspects such as job title (i.e., warehouse worker, white
Table 1. Exam ples that Vio late Inde pend ent Dimensions
AN EXAMPLE
I conclude with an extended example that considers the interface operation in
the gray area between matching the user goals and information needs as revealed
by the audience analysis and task analysis and actual (underlying) system imple-
mentation. The multidimensional audience analysis I’ve described in this article
MULTIDIMENSIONAL AUDIENCE ANALYSIS / 275
gives the development team a chance of understanding how to relate the infor -
mation to the various user groups and to produce a useful system. This example
does not explain how to perform a multidimensional audience analysis, but,
rather, helps demonstrate the type of dynamic information for which it will be an
essential element.
Assume you are working on the development of a medical system which
provides information to patients. For example, a person has been diagnosed
with high blood pressure and high cholesterol, and told to follow a new diet
that is very different from their current diet. The system needs to present the
new diet and also present effects of failing to follow it. Rather than sweeping
statements, this information should exist at the level of specific food groups and
specific types of food.
Audience analysis must focus on the various audience group’s goals and their
information needs. As part of the analysis, the information needs of each group
need to be clearly understood. Unfortunately for the analyst and the technical
communicator, the interrelations of information are complicated because people
are better viewed, not as someone with clear goals waiting to be uncovered (an
unstated assumption of many discussions of audience analysis), but as someone
with fluid, ill-formed goals [16], which they strive to achieve via multidimensional
strategies.
Consider the following:
• User groups have different views. A middle aged person may have different
concerns and attitudes toward the healthcare system than an older person.
Issues such as of out-of-pocket expense or physician trust vary across age
groups and can affect how a person reads and interprets the information.
• Different people within the same user group want different amounts of infor-
mation and need it presented differently.
• Different factors influence individual needs. For example, in some cultures
diet recommendations should exclude pork products.
One way to perform the audience analysis would be to combine ethnographic
studies with a good set of personas and scenarios. The ethnographic study needs
to capture the goals and attitudes of various user groups. Interactions with the
user groups would also allow basic metatagging that relates the dimensions to
the specific content.
Likewise, the data stored in the system requires proper metadata to allow it to be
related to the sliders. However, from the technical communication point of view,
the dynamic user profiling design and collection is external to the immediate
concerns of information presentation and ensuring the system supports user goals
and information needs. In the end, the user profile and dynamic adaptation gets
translated into virtual operation of these slider controls.
As the user moves the slider controls, the total document is dynamically
changed to correspond to the slider positions. The final text is a hypertext
document with links to associated information. In other words, the user receives a
complete (virtual) Web site dynamically generated on request. After the sliders are
set, the user would be able to browse the site without touching them again.
Notice that nothing in this description actually describes how the text appears on
the screen. The focus of this example is that a high level description of the
audience analysis can be used as part of the design of a system meeting the user
goals and information needs, not defining the information architecture. Depending
on the actual information, visualization techniques, various graphics, and various
text formats would provide the most effective presentation. The actual presen-
tation formats would be derived as part of the analysis. For that matter, how the
Fig ure 3. Sketch of a pos si ble under ly ing design of a med i cal
infor ma tion sys tem. The doc u ment is dynam i cally updated
with infor ma tion designed to fit the needs of the user who
matches the slider posi tion.
MULTIDIMENSIONAL AUDIENCE ANALYSIS / 277
information gets into the system is also outside of the immediate scope. The large
quantity of information required may be too much for a group of people to write
(several medical writers could address one disease, but not a large number of
diseases). Instead, the information development may require assistance from
some level of software agent technology or other such ways of dynamically
and systematically searching for and retrieving information from the Web. But
the final result must still be an integrated presentation and not a list of links
to various Web sites.
CONCLUSION
My intent in this article was to present a framework for thinking of audience
analysis as being composed of several independent dimensions which work
together to give a clear view of the various user groups. An underlying assumption
is that audience analysis must be performed in more detail than current practice
often supports and that analysis and design must become a much greater part of the
technical communicator’s job [1, 17]. With the move to single sourced dynamic
information, simple views of audience or “one size fits all” approaches will not be
acceptable. I admit that missing from this article was an explanation of how to
perform the analysis and how to transform the analysis results into a usable format
for dynamic information generation.
278 / ALBERS
The first point, how to perform the analysis, does not really require any new
developments. The current audience analysis methods should work as was touched
upon when I discussed scenarios and personas. The basic difference is that the
analyst must work from a mindset of looking for and working with multiple
dimensions. Once working from that mindset, some of the problems of how to
sort out and integrate the collected data may be resolved more clearly than in
current audience analysis.
The second point, how to write and format content for dynamic information
generation, remains an open question which still requires extensive research [18].
Plus, understanding the topic extends far beyond audience analysis. We are only
now gaining the tools to truly deliver dynamic information. How they will be used
and what writing techniques we will need to develop should occupy technical
communication research for the next several years.
Effective communication requires the information to fit a minimum of three
dimensions:
1. Reader knowledge of the topic. How much the reader already knows influ-
ences the technical level of the document and the allowable terminology.
2. Reader desired detail level. Regardless of how much readers already know,
the amount of knowledge they expect to gain can run from a brief overview
to an in-depth understanding.
3. Reader cognitive ability which includes factors such as the person’s reading
ability, education level, and physical/mental limitations.
Depending on the actual situation, other dimensions may be required. Data
only becomes information when it fits within the user’s context and situation. As
we gain the ability to create custom documents on-demand, we need to develop
more in-depth methods of audience analysis to ensure those custom documents
fit a reader’s needs.
REFERENCES
1. A. Rockley, The Impact of Single Sourcing and Technology, Technical Com muni-
cation, 48:2, pp. 189-193, 2001.
2. K. Schriver, Dynamics in Document Design, Wiley, New York, 1996.
3. J. Hackos and J. Redish, User and Task Analysis for Interface Design, Wiley,
New York, 1998.
4. G. Alred, W. Olin, and C. Brusaw, The Professional Writer: A Guide for Advanced
Technical Writing, Bedford/St. Martins, 1991.
5. R. Burnett, Technical Communication, Heinle & Heinle, Boston, 2000.
6. J. Caricato, Visuals for Speaking Presentations: An Analysis of the Presenter’s
Perspective of Audience as a Partner in Visual Design, Technical Communication,
47:4, pp. 496-514, 2000.
7. L. Bannon, Issues in Design: Some Notes, User Centered. System Design: New
Perspectives on Human-Computer Interaction, D. Norman and S. Draper (eds.), LEA,
Hillsdale, New Jersey, 1986.
MULTIDIMENSIONAL AUDIENCE ANALYSIS / 279
Albers, M. and Kim, L. Information Design for the Small-Screen Interface: An Overview of
Web Design Issues for Personal Digital Assistants, Technical Communication, 49 :1,
pp. 45-60, 2002.
Albers, M. and Mazur, B. (eds.), Content and Complexity: The Role of Content in Infor -
mation Design, Erlbaum, Hillsdale, New Jersey, 2002.
Kim, L. and Albers, M., Web Design Issues When Searching for Information Using
Handheld Interfaces, Technical Communication, 49:3, pp. 314-329, 2002.
Webb, M. and Albers, M. J., The Design Elements of Medieval Books of Hours, Journal
of Technical Writing and Communication, 31:4, pp. 353-361, 2001.