Undergraduate
Experiences of University of Virginia Alumni 1992
Survey of alumni and alumnae to assess the impact of their UVa undergraduate
education two, five, and ten years after graduation. IAS conducted the
survey in the summer and fall of 1992. The mail survey of randomly selected
alumni from the classes of 1982, 1987, and 1990 requested detailed information
about what graduates were doing in both their personal lives and in their
careers. The survey asked graduates about their academic pursuits and
their extracurricular activities while they were undergraduates, their
current satisfaction with various aspects of the University, and how
their undergraduate experiences have influenced their lives after graduation.
1,439 alumni and alumnae participated in the survey, out of an initial
sample of 2,500 for a response rate of 62% (excluding respondents who
could not be contacted).
Questionnaire (.pdf)
Data
(Stored in IAS Data
Catalog)
Summary Tables
Frequencies
Overall
Text Comments
Reports
A Study of UVa Alumni: The Undergraduate Experience and Beyond
Methodology
The objective of the Undergraduate Experiences
of University Alumni survey was to measure the impact of a U.Va.
undergraduate education on the personal lives and careers of alumni
two, five, and ten years after graduation. The office of Institutional
Assessment and Studies asked the Center for Survey Research (CSR)
to design a study to help us meet this objective. In the summer
of 1992, CSR administered a mail survey based on the “Total
Design Method,” a set of related techniques that has been
shown to optimize cooperation, rates of response, and accuracy
in mail surveys without compromising the confidentiality of the
process.
Study Population and Sample
We defined the
study population as all U.Va. students who earned an undergraduate
degree in 1982,
1987, or 1990. In order to make inferences with confidence about
U.Va. alumni and differences among subgroups in the population,
we sought to obtain completed questionnaires from about 400 alumni
in each class and from at least 100 alumni in each of the University’s
six schools. The University’s Alumni Association provided
CSR with a list of names and addresses of alumni from the three
selected classes from which to prepare the survey sample. Based
on CSR’s past experiences with return rates, we arrived at
an initial sample size of 2,500. These potential respondents were
selected randomly with quotas established for each class and school.
Data Collection and Data Entry
In May 1992, respondents
received a personalized letter from Vice President and Provost
Thomas H.
Jackson explaining the subject of the study, its importance to
future students, the importance of each respondent to the success
of the study, and his guarantee of complete anonymity and confidentiality.
Alumni who wished to respond to the questionnaire by computer disk,
were asked to return a postage-paid postcard indicating their preferences.
In June, respondents received a cover letter from CSR’s associate
director, Steven E. Finkel, a 16-page questionnaire (and IBMcompatible
floppy diskette, if requested), a postage-paid return envelope
in which to return the questionnaire, and a U.Va. decal as a small
token of appreciation. The study questionnaire was written by Steven
Finkel of CSR and W. Edmund Moomaw of the Office of Institutional
Assessment and Studies and accompanied by a cover letter reiterating
the points included in the initial letter from the provost and
providing instructions on completing the questionnaire. Again,
the cover letter discussed in detail the procedure for maintaining
respondents’ anonymity and confidentiality.
We guaranteed
the anonymity of our respondents by placing a label with the respondent’s
number on the return envelope rather than the questionnaire. CSR
used this respondent number to track respondents’ participation
in the computer database system as questionnaires were received.
No identifiers of any kind remained on the questionnaire, making
it impossible to link respondent answers with their name. CSR staff
entered the responses into a database customized to emulate the
questionnaire format as closely as possible. Each set of responses
was assigned a unique study number, a number different from the
one used for tracking returns. There is no way that a respondent’s
tracking and data respondent numbers can ever be linked. Standard
data-cleaning procedures were used in the analysis phase to eliminate
data entry errors and logically invalid entries.
11See Don A. Dillman.
Mail and Telephone Surveys: The Total Design Method, (New York:
Wiley, 1978). 45
About two weeks after the initial mailing, every
respondent was sent a postcard reminding them to return their questionnaire
and thanking them if they had already done so. In July, CSR administered
a second mailing addressed only to those who had not responded
to date. The second mailing included a new copy of the questionnaire,
another postage-paid return envelope, and a more urgently worded
personalized cover letter from Mr. Finkel.
By the end of the field
period, which ended around Labor Day, CSR had received 1,436 identifiable
and completed questionnaires for a response rate of 57%. Thirteen
percent of the questionnaires were completed on computer diskette.
CSR also received 185 unopened mailing packets marked “return-to-sender” which
were undeliverable to the address provided by the Alumni Association.
Excluding these 185 respondents produces an impressive overall
response rate of 62%.
Weighting
Because we initially sampled the
three classes at different sampling rates, oversampled the smaller
schools, and the groups varied in their rates of response, the
surveys actually received did not represent the proper proportion
of each sampling group among the study population (all U.Va. students
who earned an undergraduate degree in 1982, 1987, or 1990). For
example, College of Arts and Sciences graduates comprised about
70% of the study population, yet they were intentionally underrepresented
in our initial sample so we could obtain at least 100 responses
from Architecture, Education, and Nursing school graduates. As
a result, College graduates comprised only 48% of our actual respondents.
In order to generalize back to the population, we needed to apply
weights to the data, giving greater weight to respondents who were
underrepresented in the sample and giving lesser weight to those
who were overrepresented.
All results reported for the total
sample throughout this report are based on the statistically weighted
data to ensure
representativeness of the results. All results reported by class
are based on data weighted only by school while all results reported
by school are based on data weighted only by class. We employed
this procedure because using the full sample weight would have
produced a weighted N less than 50 for the smaller schools, compromising
the reliability of our measures that indicate statistically significant
differences among groups.
Report Preparation
Analysis of the data
from the survey was completed by the Office of Institutional Assessment
and Studies,
particularly by Girish J.“Jeff” Gulati, Research Assistant
in the Office of Institutional Assessment and Studies and graduate
student in the Department of Government and Foreign Affairs. In
the Office of Institutional Assessment and Studies, John M. Clayton
compiled additional information for the report and also assisted
with data analysis. Laura F. Hawthorne provided editorial and design
support, and Frances Marsh provided data entry, typing, and other
clerical support services. W. Edmund Moomaw and Mr. Gulati wrote
the report and take responsibility for conclusions reached.
Margin of Error
At a 95% confidence level,
the margin of error for the results of this survey is ±2.6%.
Put simply, 95 out of 100 samples of this size, drawn from the
same population, will generate a sample result that is within plus
or minus 2.6 percentage points of the population value. On certain
questions which were answered by smaller numbers of respondents
or were reported by sub-population groups, the margin of error
is correspondingly greater. For example, the margin of error is ±3.2%
for results based on 1,000 respondents; ±3.6% for results
based on 750 respondents; ±4.4% for results based on 500
respondents; ±6.2% for results based on 250 respondents; ±9.8
for results based on 100 respondents; and ±13.9% for results
based on 50 respondents. These statistics do not measure non-sampling
sources of error which can occur in any survey or poll.
Questions regarding the administration of the
survey, the questionnaire, or the results can be directed to Jonathan
Schnyer, Assistant Director and Assessment Coordinator, Office
of Institutional Assessment and Studies, P.O. Box 400427, Charlottesville,
VA 22904-4727 (telephone: 434/924-3417 or schnyer@virginia.edu).
|
