Professional Documents
Culture Documents
ERM Session4
ERM Session4
9 Final Exam
Describe what happened in the past so that we can understand the present or
plan for the future Consumer Electronics
Go through existing records and reports and talk to different people to get as true
Firms
Intel
a picture as possible Fairchild Semiconductor
of
The archives are reviewed in an interrogative manner with a particular
Number
Bell Labs
research question/problem in mind
Transistor
Main problem: trusting human memory which records selective parts of reality
Nanobiotechnology Industry
Firms
106 | Dr. Sandra Pauser 107 | Dr. Sandra Pauser Zevalin®
Rituxan®
of
Number
Hybridoma Drug Delivery
PEGylation Doxil®
Diagnostics
Liposome
Bio-materials
Time
- Coding: data are broken down into component parts, which are given names
Category Core
Causal
- Theoretical saturation: you reach a point where new data are no longer Conditions
Category or Strategies Consequences
illuminating the concept Phenomenon
Category
- Constant comparison (between concepts and indicators) (Ghauri, Intervening
Grønhaug, and Strange, 2020) Category Conditions
In the second phase, axial coding, the grounded theorist selects one open coding
category, positions it at the center of the process being explored (as the core phenome-
non), and then relates other categories to it. These other categories are the causal condi-
tions (factors that influence the core phenomenon), strategies (actions taken in response
CHAPTER 13 Grounded Theory Designs 445
Close to Saturated
Categories
More Refined
Categories
Third
Interview Toward
Refined Saturation
Categories of
Second Categories
Interview
Preliminary (Ghauri, Grønhaug, and Strange, 2020)
Categories
First
Interview
the process by which the researcher collects data, analyzes those data immediately rather
than waiting until all data are collected, and then bases the decision about what data to
collect next on this analysis. The image of a “zigzag” helps us to understand this proce-
dure, as shown in Figure 13.5. As illustrated in this figure, the grounded theorist engages
in initial data collection (e.g., the first collection of interview data), analyzes those data
for preliminary categories, and then looks for clues about what additional data to collect.
These clues may be underdeveloped categories, missing information in the sequence of
Qualitative
the study process, orResearch Methods
new individuals who can provide insight into some aspect of the Qualitative Research Methods
process. The grounded
Grounded Theorytheorist
Example:then returns to the field to gather this additional infor- Case Study
mation. In this procedure, the inquirer refines, develops, and clarifies the meanings of
categories for the theory. This process weaves back and forth between data collection Data sources of a case study can come from verbal reports, personal interviews
and analysis and continues until the inquirer reaches saturation of a category. Saturation and observation as primary data sources or financial reports, archives, as well as
in grounded theory research is a state in which the researcher makes the subjective budget and operating statements, including market and competition reports
determination that new data will not provide any new information or insights for the
developing categories. Case studies are often of an explanatory, exploratory or descriptive nature
Identifying this process in a published grounded theory study requires close exam-
ination of the data collection and analysis process to note whether the researcher seems When research questions concern only ‘what’, e.g., ‘What are the ways in which
an effective firm is operated?’, an exploratory study is justified
to be recycling between data collection and data analysis. For example, in a study of the
processes of men experiencing chronic illness, Charmaz (1990) carefully documented
Example: study a situation or an organization
how she interviewed 7 of the 20 men in her study more than once to refine her emerging which has rarely been studied and is unique
categories. in its nature (Ghauri, Grønhaug, and Strange, 2020)
it into categories, collecting additional information, and comparing the new information
with emerging categories. This process of slowly developing categories of information is
the constant comparative procedure. In grounded theory research, constant comparison
Qualitative Research Methods Qualitative Research Methods
Comparative case studies - we ask or study the same questions in a number of Strategies for Case Selection (Ghauri, Grønhaug, and Strange, 2020)
organizations and compare them with each other to draw conclusions Selection strategy Explanation
Critical case Where an insight into a single critical case can provide maximum and
logical explanation that can be true for all other cases in that
The purpose is to
category
Comparative cases When cases are selected so that these can be compared and
compare the phenomenon (e.g. strategy formation) studied in different cases in a contrasted to achieve maximum understanding
systematic way Criteria based cases When cases are selected that are based on pre-decided criteria.
Often to compare the cases that meet criteria and those that do not
meet criteria.
explore different dimensions of our research issues Heterogeneous cases When we select cases that are clearly different from each other. For
example, to understand the impact of a certain factor on the success
of manufacturing versus services firms.
examine different levels of research variables Homogeneous cases When we select cases that are clearly similar. For example, cases
(Ghauri, Grønhaug, and Strange, 2020) from manufacturing as well as the same industry such as
automobiles.
Typical cases When we select cases that are considered normal/average to
achieve better insight and understanding.
Snowball case selection When we select cases that can lead us to further cases, suggested
by respondents, that are difficult to approach on our own or are
connected to earlier cases.
114 | Dr. Sandra Pauser 115 | Dr. Sandra Pauser
Entail listening and watching other people’s behavior in a way that allows some
type of learning and analytical interpretation
Advantages
- collect first-hand information in a natural setting
- can interpret and understand the observed behavior, attitude and situation
- capture the dynamics of social behavior
Disadvantage
- difficult to translate the events or
(Ghauri, Grønhaug, and Strange, 2020)
happenings into scientifically useful
information (Ghauri, Grønhaug, and
Strange, 2020)
Gain an accurate and clear picture of a respondent’s position or behavior Open questions are useful in exploratory
research, when the researcher may not have a
Use open-ended questions and respondents are free to answer according to
clear idea of possible responses, and/or when
their own thinking the researcher wants to elicit detailed
responses
Suitable for exploratory and inductive types of study (Ghauri, Grønhaug, and
Often difficult to code the responses to such
Strange, 2020)
questions consistently (Ghauri, Grønhaug, and
Strange, 2020)
Demand a skilled and cautious interviewer Focus groups take many different forms, e.g., discussion groups, focused
interviews, group interviewing and group research for new product development,
Expect the interviewer have a complete understanding of the research problem,
program evaluation, etc.
its purpose and what information is being sought A group of 6 to around 10 individuals discuss a particular topic/issue under
May take a long time the direction of a moderator
Should be some homogeneity among the individuals in one specific focus
Difficult to interpret and analyze
group
Coding of in-depth interviews can be a difficult Last from half an hour to around two hours, relatively cheap and convenient way
task (Ghauri, Grønhaug, and Strange, 2020) Focus groups are mostly used for collecting qualitative data
The records and notes are analyzed, for example by using content analysis
(Ghauri, Grønhaug, and Strange, 2020)
Produce very rich and in-depth data expressed in respondents’ own words and Difficult to summarize and categorize the information gathered
reactions
Difficult to get useful information in the case of an unskilled moderator
A quick, flexible and inexpensive method of data collection
Difficult to gather people at a location
Observe reactions of people in open and free conversation with each other
The responses of the group members are not independent of one another
Allow the researcher to interact directly with respondents, and to react and build
The live participation and observation may lead the researcher to have greater
upon the discussion as it goes
faith/belief than is warranted
(Ghauri, Grønhaug, and Strange, 2020)
The moderator may bias the respondents
(Ghauri, Grønhaug, and Strange, 2020)
Secondary data are data that have already been collected by government agencies,
market research agencies, firms, or other organizations and/or individuals (Ghauri,
Grønhaug, and Strange, 2020)
The data were collected for purposes other than those of the researcher When secondary data sources are not available or are inappropriate
Sometimes there are some uncertainties relating to the accuracy and reliability of When the researcher requires data for a new group of firms or individuals
the raw data
Need to collect primary data though surveys, interviews, or experiments (Ghauri,
No way of checking how (if at all) the raw data were checked and cleaned Grønhaug, and Strange, 2020)
(Ghauri, Grønhaug, and Strange, 2020)
Be able to collect data with the exact objectives/questions of the project The process is often slow and expensive
Can choose exactly who to approach, and how many units of observation to be Difficult to obtain suitable access and/or build a dataset of suitable size
surveyed
Response rates are often low, which may lead to small sample sizes and/or non-
Be able to collect accurate responses and thus have greater confidence in the response bias (Ghauri, Grønhaug, and Strange, 2020)
final results (Ghauri, Grønhaug, and Strange, 2020)
Analytic surveys may be used to test a theory (e.g., the relationship between
accounting control systems and business performance) Study over Time Study at One Point in Time
Longitudinal Cross-sectional
Descriptive surveys are concerned with identifying the phenomena whose
variance we wish to describe (e.g., consumer attitudes towards a certain product)
(Ghauri, Grønhaug, and Strange, 2020) Changes in a
Changes
Trends in subpopulation
in the Attitudes
the same group identified Community Program
same and
population by a common needs evaluation
people practices
over time characteristic
over time
over time
Causal-
Between-group designs True Quasi- Comparative Repeated
Experiment Experiment Research Factorial Time Series Measures Single Subject
- True experiments (pre- and posttest, posttest only)
Random Yes No No May be No No No
- Quasi-experiments (pre- and posttest, posttest only) assignment used
- Causal-comparative research Number Two or Two or Two or more Two or One group One group One individual
of groups/ more more more studied at a
- Factorial designs individuals time
compared
Number of One or One or One or more Two or One or Two or One or more
Within-group or individual designs 312 3 Research
PART interven- Designs more
more interventions more more more interventions
tions used interventions interventions interventions interventions interventions
- Time-series experiments (interrupted, equivalent)
Number of Once Once Once Once After each After each Multiple points
- Repeated-measures experiments TABLE 10.3
times the
dependent
intervention intervention
True Experiments
Random True Control groupcomprise the most
experiments Norigorous
treatment Posttest
and strong experimental designs because
TABLE 10.3 assignment of equating the groups through random assignment. The procedure for conducting
Quantitative Research
Types of Between-Group Designs Methods Quantitative Research Methods
Random major forms of true group
Experimental experiments and quasi-experiments, viewing them
Experimental in terms of activ-
Posttest
ities from the beginning of the experiment to the end, is shown in Table 10.3. In true
assignment treatment
experiments, the researcher randomly assigns participants to different conditions of the
True Experimental Designs experimental variable. Individuals in the experimental group receive the experimental
Quasi-Experimentaltreatment,
Designswhereas those in the control group do not. After investigators administer the
Pre- and Posttest Design Time treatment, they compile average (or mean) scores on a posttest. One variation on this
Pre- and Posttest
design is Design Timemeasures or observations. When experi-
to obtain pretest as well as posttest
menters collect pretest scores, they may compare net scores (the differences between
Random Control group Pretest No treatment Posttest the pre- and posttests). Alternatively, investigators may relate the pretest scores for the
Select control Pretest No treatment Posttest
assignment control and experimental groups to see if they are statistically similar and then compare
group the two posttest group scores. In many experiments, the pretest is a covariate and is sta-
Random Experimental group Pretest Experimental Posttest tistically controlled
assignment treatment Select experi- Pretestby the researcher.
Experimental Posttest
mental group Because you randomly assign individuals
treatmentto the groups, most of the threats to internal
validity do not arise. Randomization or equating of the groups minimizes the possibility
Posttest-Only Design Time
Posttest-Only Design Time
Random Control group No treatment Posttest
assignment Select control group No treatment Posttest
Random Experimental group Experimental Posttest Select experimental group Experimental treatment Posttest
assignment treatment M10B_CRES9364_06_SE_C10.indd 310 13/12/17 3:50 PM
Causal-Comparative Research
Quasi-Experimental Designs
Pre- and Posttest Design Time
Pre- and Posttest Design Time
140 | Dr. Sandra Pauser 141 | Select group
Dr. Sandra Pauser A Pretest Grouping variable of interest Posttest
Select control Pretest No treatment Posttest (e.g., no intervention)
group Select group B Pretest Grouping variable of interest Posttest
Select experi- Pretest Experimental Posttest (e.g., received intervention)
mental group treatment
Select control Pretest No treatment Posttest
group
Select experi- Pretest Experimental Posttest
mental group treatment
Pre- and Posttest Design Time Generate data that capture the relevant theoretical constructs
Select group A Pretest Grouping variable of interest Posttest Isolate the key mechanisms that the researcher wishes to study, and remove the
(e.g., no intervention) confounding effects that are inevitably present with real-world data
Select group B Pretest Grouping variable of interest Posttest
(e.g., received intervention)
Introduce additional controls to take into account factors that may interact with
Posttest-Only Design Time the main effects
Select group A Grouping variable of interest Posttest Establish causal relationships more conclusively
(e.g., no intervention)
Select group B Grouping variable of interest Posttest
Can replicate experiments (Ghauri, Grønhaug, and Strange, 2020)
(e.g., received intervention)
is used, it introduces all these factors as possible threats to validity. Instrumentation exists
142 | Dr. Sandra Pauser 143 | Dr. Sandra Pauser
as a potential threat in most experiments, but if researchers use the same or similar
instrument for the pre- and posttest or enact standard procedures during the study, you
hold instrumentation threats to a minimum.
Pauser, S., and Ebster, E. (2015). The depiction of individuals and bundled presentations in online- Pauser, S., and Wagner, U. (2020). Judged by its look: Assessing the comprehensibility and
perceived appearance of sign language avatars, Marketing ZFP – Journal of Research and
catalogues. AIMS International Journal of Management, 9(3), 101-108.
Management, 42(3), 48-62. doi: 10.15358/0344-1369-2020-3-48.
Characteristics:
- Associations between scores (direction, form, and strength)
- Displays of scores (scatter plots and matrices)
- Multiple variable analysis (partial correlations and multiple regression)
The two primary correlation designs are explanation and prediction (Creswell &
Guetterman, 2019, p. 343-347)
Pauser, S., and Wagner, U. (2020). Judged by its look: Assessing the comprehensibility and
perceived appearance of sign language avatars, Marketing ZFP – Journal of Research and
Management, 42(3), 48-62. doi: 10.15358/0344-1369-2020-3-48.
Explanatory designs consist of a simple association between two variables The investigator identifies one or more predictor variable and a criterion (or
(e.g., sense of humor and performance in drama) or more than two (e.g., outcome) variable. A predictor variable is a variable used to make a forecast about
pressure from friends or feelings of isolation that contribute to binge drinking) an outcome in correlational research.
(Creswell & Guetterman, 2019, p. 345)
Depression Scores
Regression Line
50
41
40
30 Slope
(d) Curvilinear (e) Curvilinear
20
10
Intercept
5 10 14 15 20
(Creswell & Guetterman, 2019, p. 348) (Creswell & Guetterman, 2019, p. 355)
150 | Dr. Sandra Pauser 151 | Dr. Sandra Pauser
The calculation of this line holds value for predicting scores on the outcome (i.e.,
M11_CRES9364_06_SE_C11.indd 348 13/12/17 4:18 PM
depression) with knowledge about the predictor (i.e., hours of Internet use per week).
Based on a mathematical formula, a researcher can calculate an equation that expresses
this line:
Y (predicted) = b (X ) + a
where
Y = predicted score on depression
X = actual score on number of hours of Internet use
b = slope of the regression line (called the unstandardized regression coefficient)
a = the intercept or a constant, the value of the predicted Y (depression) score
when X = 0
We would expect an individual who uses the Internet 14 hours per week to have a
depression score of 41. This score can be estimated by drawing a vertical line from the
score for the X-axis variable up to the regression line and over to the Y-axis variable.
Alternatively, using the regression formula,
If a = 6, b = 2.5, and X = 14,
Then Y (predicted) = 2.5(14) + 6 = 41.
Consider a more complicated situation where multiple independent variables may
combine to correlate with a dependent variable. Multiple regression (or multiple
correlation) is a statistical procedure for examining the combined relationship of multiple
independent variables with a single dependent variable. In regression, the variation in
the dependent variable is explained by the variance of each independent variable (the
relative importance of each predictor) as well as the combined effect of all independent
variables (the proportion of criterion variance explained by all predictors), designated by
R2 (Kline, 2016). Similar to the regression equation mentioned earlier, predicted scores on
an outcome can be generated using an equation that is similar to the simple regression