? 2003 for Handbook of Entrepreneurial Dynamics;
Appendix B - Data Documentation, Data Preparation and
Weights
Richard T. Curtin
University of Michigan
and
Paul D. Reynolds
Babson College and London Business School
Prepared for the
Handbook of Entrepreneurial Dynamics: The Process of
Organizational Creation
September 2003 Last printed 7/9/2010 1:11:00 PM Page 1 of 25
C:\convert\temp\63364232.doc
? 2003 for Handbook of Entrepreneurial Dynamics; Outline
Introduction
Interview schedules
Codebook Presentations
Overview of Data Sets
Screening Data Set
Detailed Data Set
Representative Samples and Respondent Weights
Adjusting for Bias and Respondent Weights
Initial Weights
U Michigan Survey Research Center Revised Weights
Respondent Rights and Welfare
Commentary
Exhibits and Tables
Exhibit 1 - Example of Phone Administered Interview Schedule: Screening Exhibit 2 - Sample Page from Self-Administered Questionnaire Exhibit 3 - Codebook Variable Example: Initial Screening Items Exhibit 4 - Codebook Variable Example: Mail Questionnaire Item Exhibit 5 - Codebook Variable Example: Response Categories Across Waves Exhibit 6 - Codebook Variable Example: Constructed Variable - Start-up Type Exhibit 7 - Example of Weight Calculation
Table 1 - Source and Count of Screening File Variables
Table 2 - Variable Names First Letter and Follow-up Stage by Cohort Table 3 - Impact of Alternative Weights on Screening Sample Nascent Entrepreneurs Age
Distribution
Table 4 - Weights in the Data Sets
Last printed 7/9/2010 1:11:00 PM Page 2 of 25
C:\convert\temp\63364232.doc
? 2003 for Handbook of Entrepreneurial Dynamics;
Introduction
The analysis of data from the Panel Study of Entrepreneurial Dynamics [PSED]
requires a good deal of information about the design of the research. The wording of questions and how the responses were coded is just as important as how to effectively combine the answers from the different samples while preserving the confidentiality and identity of the respondents. Whereas the prior appendix focused on the sampling
procedures and interview schedules, this appendix details the procedures used to collect and document the data, the development of case weights, and the procedures used to
protect the confidentiality and identity of respondents.
Without a theoretical foundation, mere numbers are meaningless. Most of the
descriptive and theoretical questions addressed in the PSED require multiple questions to assess. As a result, the individual questions must be interpreted within the context in which they were asked, including both preceding and subsequent questions in each sequence. While all response were translated into a numerical code for analysis purposes, some codes have explicit meaning, say dollars or age in years, while other coded values were assigned by convention, say ”1” for a yes response. Even if the coded values do not have an explicit meaning, the simple tabulation of responses will be misleading unless the responses are properly weighted to account for the differences in sample designs and non-response. The proper use of weighted data is critical for developing inferences regarding the population represented by the samples.
Types of Questionnaires
There were seven basic questionnaires associated with different parts of the project: 1) Screening questionnaire: A short set of questions used to locate potential nascent
entrepreneurs or candidates for the comparison groups. 2) Initial Nascent Entrepreneurs Questionnaire: Detailed questions asked in phone
interview schedules of all eligible nascent entrepreneurs. Last printed 7/9/2010 1:11:00 PM Page 3 of 25
C:\convert\temp\63364232.doc
? 2003 for Handbook of Entrepreneurial Dynamics;
3) Initial Self-administered Questionnaires: Mail interview schedules sent to all
nascent entrepreneurs that completed the phone interview and agreed to complete
the questionnaire.
4) Comparison Group Questionnaires: Initial phone interviews conducted with the
comparison group respondents; this is a revised version of the nascent entrepreneur
phone interview more appropriate for those not starting a business.
5) Comparison Group Self-Administered Questionnaires: Mail questionnaires used
with the comparison group respondents; this is a revised version of the nascent
entrepreneur mail interview schedule more appropriate for those not starting a
business.
6) Follow-up Nascent Entrepreneur Questionnaire: Phone interview schedules used in
subsequent follow-ups with the nascent entrepreneurs.
7) Follow-up Self-Administered Questionnaires: Mail interview schedule used for all
nascent entrepreneurs that completed subsequent interviews: a reduced version of
the interview schedule provided to nascent entrepreneurs in the initial interviews.
There are two forms of the phone interviews. The Computer Assisted Telephone Interview
[CATI] schedules are relatively complicated computer programs using programming
syntax. To facilitate analysis, a “user friendly form” was prepared that looks like a paper
and pencil interview and attempts to reflect all the features of the CATI version. The self-
administered mail questionnaires are less complex and are only in one format.
Another critical document is the codebook which represents a compilation of the
initial and follow-up data in one 450 page document. The codebook provides a description
of all variables in the data sets, the valid responses to each item, and the numeric values
assigned to each response code. .
These materials are much too lengthy to include in this Handbook. Copies of the
CATI phone interview schedules, the self-completed questionnaires, and the codebooks are
available on the project website [http://projects.isr.umich.edu/PSED] or on a CD provided
by the sponsoring organizations.
Last printed 7/9/2010 1:11:00 PM Page 4 of 25 Most serious researchers will produce a printed copy of the codebook and interview C:\convert\temp\63364232.doc
schedules for reference; they fit easily in ring binders and two sided printing is very useful
? 2003 for Handbook of Entrepreneurial Dynamics;
for reducing the bulk. A separate binder for each facilitates simultaneous cross-referencing; reviewing the item in the interview schedule and then considering the pattern of the
responses in the codebook.
Interview Schedules
There are two types of interview schedules used in this project. The first are phone
interviews where a trained interviewer engages in a carefully constructed interaction with the respondent. These are completed with a Computer Assisted Telephone Interviewing
(CATI) procedure where questions appear on a computer monitor, are read to the
respondent by the interviewer who then enters the responses directly into the computer. The questions asked of each respondent are adjusted based on their responses to prior
questions. The computer program automatically performs consistency checks and provides
immediate feedback to the interviewer in the form of clarification questions or probes
[additional questions to resolve any ambiguity in the answer] so any inconsistencies can be resolved.
The second type of interview schedules are self-completed questionnaires that were
mailed to each respondent.
The screening interview is provided in Exhibit 1 as an example of a phone
interview schedule. The actual interview schedule is a sophisticated and complicated
computer program. This version has been created to guide those planning to analyze the
data. Most of the items are related to the personal and household situation of the
respondent. Only seven questions are related to the PSED screening procedure, starting
with item 1, [variable label BSTART]: “Are you, alone or with others, now trying to start a
new business?”
Note the important skip pattern that follows item 2 [BJOBST]. This indicates that
only if a respondent says “Yes” to either or both items 1 and 2 [BSTART and BJOBST]
will they continue with item 3 [OWNER]. If they say “no” to both questions, the program will exit the PSED section and the interviewer will be instructed ask about the next topic of Last printed 7/9/2010 1:11:00 PM Page 5 of 25 the interview. Those that stay in the section are then asked about ownership [OWNER] of C:\convert\temp\63364232.doc
the start-up and if they have been active in the past twelve months [SUACTS]. If they will
? 2003 for Handbook of Entrepreneurial Dynamics;
not be owners or have not been active the interviewer exits the section and the respondent is not considered eligible for the detailed PSED interview. Those that qualify are given two chances to volunteer for the project [VOLUNT1, VOLUNT2] and those that agree to participate are asked to provide their first name only [NAMEVOL]. All respondents, however, are asked the standard set of a dozen socio-demographic items regarding age, educational attainment, household structure, ethnic background, etc.
An example of one page from the self-administered questionnaire is presented in Exhibit 2. These items were provided to all nascent entrepreneurs and all those in the comparison groups. The specific examples are concerned with preferences for different types of businesses, those with more certain payout but less risk versus those with higher risk and payouts. This is followed by items related to the previous work experiences of the respondent. Because a wide range of respondents with very diverse educational backgrounds must be able to read and interpret these items, substantial effort was devoted to making them as simple and as direct as possible.
Last printed 7/9/2010 1:11:00 PM Page 6 of 25
C:\convert\temp\63364232.doc
? 2003 for Handbook of Entrepreneurial Dynamics;
Exhibit 1 - Example of Phone Administered Interview Schedule: Screening
[Variable labels in brackets.]
[BSTART] 1. Are you, alone or with others, now trying to start a new business?
Yes............................................................................. 1
No .............................................................................. 2
Don‟t know ............................................................... X
Refused ..................................................................... R
[BJOBST] 2. Are you, alone or with others, now trying to start a new business or new venture for your
employer? An effort that is part of your job assignment?
Yes............................................................................. 1
No .............................................................................. 2
Don‟t know ............................................................... X
Refused ..................................................................... R
ASK QU. 3 IF “YES” TO QU. 1 OR QU. 2; OTHERWISE, EXIT INTERVIEW.
[OWNER] 3. Will you own all, part or none of this new business?
(DO NOT READ LIST. ENTER SINGLE RESPONSE.)
All .............................................................................. 1
Part ............................................................................ 2
None 3] ? (EXIT INTERVIEW)
Don‟t know ............................................................... X
Refused ..................................................................... R
[SUACTS] 4. In the past twelve months, have you done anything to help start this new business, such as
looking for equipment or a location, organizing a start-up team, working on a business plan,
beginning to save money, or any other activity that would help launch a new business?
Yes............................................................................. 1
No .............................................................................. 2 ] ? (EXIT INTERVIEW)
Don‟t know ............................................................... X
Refused ..................................................................... R
[VOLUNT1] 5. A national study of those starting new businesses is being conducted through the University
of Wisconsin. Those eligible will receive a cash payment. May they contact you?
Yes............................................................................. 1 ? (SKIP TO QU. 7)
No .............................................................................. 2
Don‟t know ............................................................... X
Refused ..................................................................... R
[VOLUNT2] 6. You need not participate, but those that have find it interesting and very useful. Can the
University of Wisconsin researchers contact you and tell you what is involved? You can
always turn them down.
Yes............................................................................. 1
No .............................................................................. 2 ?
Don't know ............................................................... X ? ? (EXIT INTERVIEW)
Refused ..................................................................... R ?
[NAMEVOL] 7. May I have your first name only please?
__________________________________________
Last printed 7/9/2010 1:11:00 PM Page 7 of 25
C:\convert\temp\63364232.doc
? 2003 for Handbook of Entrepreneurial Dynamics;
Exhibit 2 - Sample Page from Self-Administered Questionnaire
H9. Consider two types of new businesses. Assuming you are the sole owner, which situation
would you prefer? [CHECK ONE BOX ONLY]
? 1. ALPHA - A business that would provide a good living, but with little risk of
failure, and little likelihood of making you a millionaire.
? 2. BETA - A business that was much more likely to make you a millionaire but had a
much higher chance of going bankrupt
H10. If you could obtain more information to make a choice between businesses ALPHA and
BETA, how important would each of the following be?
[CIRCLE ONE FOR EACH ROW]
1=Unimportant 2=Somewhat important 3=Very important
a. The chances of going bankrupt for both ALPHA and BETA ............... 1 2 3
b. The chances of making millions for both ALPHA and BETA ............. 1 2 3
c. The exact amount of earnings if ALPHA and BETA were successful ...... 1 2 3
d. The time and effort required to manage ALPHA and BETA ............... 1 2 3
e. The opinion of family and friends about this choice ............................. 1 2 3
f. The experience of those managing businesses
like ALPHA and BETA ........................................................................ 1 2 3
g. Your feelings about the type of business activity represented
by ALPHA and BETA........................................................................... 1 2 3
I1. Since beginning your work career, how many times have you resigned your job to take a
new position . . .
a. . . . with a new job lined up? _____ times
b. . . . without a new job lined up? _____ times
I2. The last time you had a job working for someone else or in an established organization,
what was your job title?
______________________________________________________________________
I3. What did you do?
______________________________________________________________________
______________________________________________________________________
I4. How long did you have this job? _____ years _____ months
I5. Following the chain of command, how many people
were between you and the Chief Executive Officer?
If you were the CEO, write “0.” _____ people
I6. How many people worked for this organization? _____ people
Last printed 7/9/2010 1:11:00 PM Page 8 of 25
C:\convert\temp\63364232.doc
? 2003 for Handbook of Entrepreneurial Dynamics;
Codebook presentations
The beginning of all analysis should start with an examination of the unweighted
frequency distribution of the variables under consideration. For example, the answers of
64,622 respondents approached in the screening process to the two basic items are
presented in Exhibit 3. This data is un-weighted and combines the responses of all screened
respondents, including those selected for the comparison groups. These questions were
answered by almost all the respondents, the interviewers were instructed to get a “yes” or
“no” answer if at all possible. For the few that gave “don‟t know” and “refuse” responses,
they were recoded to a “no” responses; thus all answers in the data set are either “yes” or
“no.” This practice of coding “don‟t know” and refusals as “no” was confined to the
screening data set as the other data bases include specific codes for these responses.
Exhibit 3 - Codebook Variable Example: Initial Screening Items
BSTART Are you, alone or with others, now trying to start a new business?
Frequency Code Response
4,465 1. Yes
60,157 2. No
BJOBST Are you, alone or with others, now trying to start a new business or new venture for your employer? An
effort that is part of your job assignment? Frequency Code Response
2,339 1. Yes
62,283 2. No
Last printed 7/9/2010 1:11:00 PM Page 9 of 25
C:\convert\temp\63364232.doc
? 2003 for Handbook of Entrepreneurial Dynamics;
An example from the self-administered questionnaire, item QH9, as provided in the
Codebook is presented in Exhibit 4. The letter “Q” has been added to the variable label to
indicate the first wave of data collection. In this case a total of 6 respondents did not
provide useful information on this item compared to 899 that were able to answer the
question. Again, both nascent entrepreneurs and control group respondents are combined in
this un-weighted presentation.
Exhibit 4 - Codebook Variable Example: Mail Questionnaire Item
QH9 Consider two types of new businesses. Assuming you are the sole owner, which situation would
you prefer?
Frequency Code Response
744 1. ALPHA - A business that would provide a good living, but with little risk of failure,
and little likelihood of making you a millionaire.
155 2. BETA - A business that was much more likely to make you a millionaire but had a
much higher chance of going bankrupt.
6 9. NA (No answer)
To facilitate comparisons across the different waves of data collection--and
consistent with the practice followed in other longitudinal studies--the answers to the same
questions asked at different times are presented together. For example, respondents
classified as nascent entrepreneurs were asked a number of questions about what they had
done to get ready to start the business, including the question shown in Exhibit 5, which
asks about saving money to invest in the business. If they said “Yes” at any point, they
were asked when this saving behavior began and were not asked the question in subsequent
interviews. The results, taken from the most recent codebook, indicate that 830 were asked
this question in Wave 1 and 574 answered “yes.” Only those that did not say “yes” in Wave
1 were asked this question a second time in Wave 2. Apparently 99 could be located and
interviewed in Wave 2, another 37 said “yes” and 62 said “no.” Of the 62 that were contacted for the third interview, another 15 said “yes.” [The Wave 2 and Wave 3 may
not be the same 62, as efforts are made at each wave to locate those that could not be
contacted in the previous efforts.] Such a presentation has obvious advantages for
understanding the response patterns associated with each item.
Last printed 7/9/2010 1:11:00 PM Page 10 of 25
C:\convert\temp\63364232.doc