EQUALITY GUIdE - KU Leuven

38 Equality Guide
one just has to send an e-mail to everyone and all the responses are collected in a database
that can easily be merged to a statistical programme (immediate data validation). The costs are
low since there are no paperwork and printing costs. Besides, to create the online survey and
design the web pages, software costs can be reduced by using an open source web-design
programme. In theory, everyone at the university has access to the Internet. Students who do
not have access at home can easily use a computer in one of the computer rooms at the university.
For some groups of personnel (e.g. kitchen staff) we decided to print the surveys and
hand them over in closed envelopes that the respondents could send back to our unit. Another
advantage of the survey is that it is a simple quantitative method which can be used on a yearly
basis, and which can easily be applied by other researchers. Weaknesses of this method are
social desirability and not considering the present context in which the respondent fills out the
questionnaire. Furthermore, the same survey will be used for academic staff as well as personnel
with lower educational and literacy levels. Therefore, it is important to create questions that
are understandable for all populations, but that are not too simplistic either.
To put the survey online, Osucre 61 was used, a user-friendly software program that generates
the response in a spreadsheet, which can easily be merged to a statistical programme like
SPSS. To design the web pages, the open source web-design programme NVU 62 was used.
The web survey was completely anonymous: we did not ask for the name of the respondent
and the link to the questionnaire was not linked to their e-mail address. The only personal item
the Osucre programme registers, is the IP address.
3.6. Validating the scales
One of the most important aspects in developing this survey was to test whether the
instrument was valid, i.e. does the instrument measure what it is meant to measure?
One way of doing this is by verifying whether items cluster around a particular construct,
by means of factor analysis. This was the aim of the pre-testing. We measured
which variables loaded highly on which factors (this means: which variables fit to the
different factors/constructs) and then conducted an item analysis on the factors to examine
if those variables constitute a reliable scale. This means: is there an internal
consistency in the different scales? Another aim of pre-testing was to make sure the
survey was user-friendly and understandable for the respondents. The pre-test was
conducted with 315 students (3 random samples + 2 reminders) and 227 members of
personnel who have access to the Internet (2 random samples + 2 reminders). There
was a 56.8% response for members of personnel (N = 129) and 34.6% response for
students (N = 109). Bad timing may be the reason for the lower response from students
since they were in the pre-exam period. A summary of the results of the factor
analysis of both groups is given below.
61
62
Frederik Van Acker ! www.osucre.be.
www.nvu.com.