Professional Documents
Culture Documents
RTBT - Chapter 3 v3
RTBT - Chapter 3 v3
CHAPTER 3:
CASE SELECTION AND
MEASUREMENT
ROBERT ROODERKERK
CHECKLIST FOR CRITICAL EVALUATION OF A REPORT
An example:
Theoretical domain = all alliances, all over the world, in all economic
sectors, at all times.
Population = US alliances of airlines with a total turnover of at least [n]
dollars a year.
Example:
o Study on the effect of workload on employee satisfaction
o Employees with a high workload are too busy to participate in a study
FROM DOMAIN TO DATA SET (DUL & HAK, 2008: 46)
x
x xx
= theoretical domain
= sample of a population
Theoretical domain:
All firms with CEOs, at all times, across all locations & industries.
A sample is studied.
4D. IS THE SAMPLE A PROBABILITY SAMPLE?
YES: There are missing cases, due to for instance data availability
The authors describe the sample sizes of the different models, from which
we can derive the relative numbers of missing data, but there is no
indication of the absolute amount of missing data
Compustat contains census
VISUAL SUMMARY of this population
All firms with CEO’s, at all times, across all locations and industries
1. Domain
X X X X
X
X X
X X
X
X
X
X X
X X
X
X
X X
X
X
The number of cases (i.e. rows in a data matrix) are decided upon prior
to conducting the study. This is why missing cases have to be reported.
The values that each variable can take must be clearly specified by
the researcher (e.g. satisfaction as an ordinal variable).
Hypothesis in theme:
“CEO compensation is positively related to firm performance”
CEO + Firm
overpayment profitability
1. Most (if not all) databases offered by the EDSC (Erasmus Data
Service Center) are considered to be trustworthy.
2. If there is only a limited amount of data available, this may
hamper the intended statistical analysis as well as raise selection
bias issues.
3. The best way to assess whether a database is appropriate for
your research, is to read the database manual and (if necessary)
to make individual appointments.
4. The difficulty lies in understanding what the data represents,
and what processing steps you will take to address your research
question.
X: CEO overpayment