I am doing a logistic regression (in three steps) in SPSS on radical right support and gender, using data from the European social survey. The outcome variable is whether the person voted for a radical right party or not.
I have some attitudinal variables, as well as sociodemographics like age, marital status etc. These all go on the first regression model. Then in the second one I add 'female_leader' (i.e. whether the party they voted for had a female leader), and on the third an interaction between 'female' (of the voter) and 'female_leader'.
My problem is that it doesn't add the 'female_leader' in the second step, just repeats the first step. The third one it does add the interaction variable. I'm unsure why SPSS doesn't add the 'female_leader' variable.
Here's my code:
LOGISTIC REGRESSION VARIABLES all_votes
/METHOD=ENTER female marriage_recode religion_recode attendence_recode
unemployed
professional_worker skilled_workers culture_undermined gays_free_disagree
democracy_important
secondary_below_education degree_education_plus agea cntry
/METHOD=ENTER female_leader
/METHOD=ENTER interaction
/CONTRAST (cntry)=Indicator
/CONTRAST (marriage_recode)=Indicator
/CONTRAST (religion_recode)=Indicator
/CONTRAST (attendence_recode)=Indicator
/SAVE=RESID ZRESID
/CLASSPLOT
/PRINT=GOODFIT CI(95)
/CRITERIA=PIN(0.05) POUT(0.10) ITERATE(20) CUT(0.5).
Hopefully this is enough info. Thanks.
The likely reason for the female_leader variable not to be entered in this situation is that it's redundant or linearly dependent upon the variables already entered in the first block, which means that it won't add any information to those variables in trying to predict the dependent. Look for a warning message at the top of your output reading: "Due to redundancies, degrees of freedom have been reduced for one or more variables."