> Hi,
> In my research, I study the quality of items which are used in empirical
> studies about fear of criminal victimization. Men and women seem to
> differ in their tendency to respond in a certain way, probably partly
> due to social desirability.
> I want to test this tendency and the item validity among subgroups and
> therefore I am looking for references about methods on this topic. Any
> suggestions?
> Thanks,
> Gabry
Gabry,
The issue that you have identified is that of differential item
functioning. There are a few ways to determine if groups are
responding significantly differently such as item response theory or
multiple group analysis in SEM (for example see Shevlin, M.,
Brunsden, V., & Miles, J. N. V. (1998). The Satisfaction
with Life Scale: Factorial invariance, mean structures and
reliability. Personality and Individual Differences, 25, 911-917).
However, I recently attended a British Psychological Society
conference where a paper was presented which examined the differences
in males in females in terms of the EPQ. The analysis used structural
equation modelling and had some nice features. First, it allowed the
examination of group differences at both the factor and individual
item level. Second, the direction and magnitude of these effects are
easily estimated. Third, the analysis incorporated a test of an a
priori factor structure (effectively a confirmatory factor analysis).
This model could easily be formulated to include social desirability
as a covariate or an exogenous variable.
Send me an e-mail to [log in to unmask] for further information.
Regards
Mark Shevlin
Dr Mark Shevlin
Lecturer in Psychology
University of Ulster at Magee
Northern Ireland
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|