I really do not understand why gender or other demographic
variables could not enter a regression analysis, if they are related to the
dependent variable. You talk about co-variables, and this suggests
multivariate regression analysis with several dependent variables, and in
that case you're probably right: usually one does not include gender or age
as DEPENDENT variables, but there is no problem with them as INDEPENDENT or
PREDICTOR variables.
On the other hand you also mention the fact that gender is
dichotomous, not continuous, which is a completely different issue (and has
nothing to do with the fact that people do not choose their values):
regression analysis requires interval level of measurement in all variables;
dichotomous variables can be treated as interval measures, and are
acceptable as predictors in regression equations, though there are some
objections to them as DEPENDENT variables in the context of regression,
especially because their residuals cannot be normally distributed around
predicted values, and also because their predicted values can be fractional
values or be outside the two-value range of a dichotomous variable.
All in all, I think your question mixes several different and
unconnected issues:
1. Variables chosen or not chosen by subjects
2. Continuous versus discrete variables
3. Dependent versus predictor variables
4. Background variables versus variables directly involved in your
analysis (probably subjective scores)
Hector
-----Original Message-----
From: SPSSX(r) Discussion [mailto:
[hidden email]] On Behalf Of
Annemiek Kamphuis
Sent: 12 July 2007 01:44
To:
[hidden email]
Subject: Regression analysis with demographic background variables
Hello,
I was wondering if it is possible to perform a regression analysis
that includes demographic variables like gender as co-variables. I prefer
not to use these background variables in the regression analysis as people
do not choose to have specific demographics, so the scores of these
variables can only be yes or no/high or moderate or low - not something I
want to base results of my research on.
Thank you,
Annemiek Kamphuis
Locked
