My model is as followes:
DV = Wage
IV = Educational level (master vs. doctor)
Moderator = Gender
Interaction = Gender * Educational level
Control variables (covariates):
- Age (scale)
- Supervisor responsibilities (0=yes/1=no)
- Job category (5 categories --> created dummies)
- Full/parttime (0=full time / 1 = part time)
- Contract (0 = permanent / 1 = temporary)
- Marital status (0 = single / 1 = married)
- Children (scale)
If I try do this with the Hayes process method v3.5 in SPSS it doesn't seem to work. It does work if I delete the 'job category' variables and it's 5 dummies, does that mean I only need to include the 'job category' variable and not the dummies?
I get the message "a linear or near linear dependence (singularity) exists in the data" and I don't know how to make it work.
I basically want to figure out if there is a wage gap (women vs. men) and if the wage gap is higher/lower for masters vs. doctors. I will make a model with and without control variables to see how much influence they have on the wage gap.
Thanks!
If you have 5 job categories you should have 4 dummies not 5. The fifth category is represented by the intercept and the other 4 represent deviations from the intercept. If you 5 dummies plus the intercept you are estimating 5 means with 6 parameters and there are infinite number of solutions - hence the singularity ...
Thanks, I also learned that you need to have K-1 dummies, but I used the 'create dummy variables' option in SPSS and it conducted 5 dummies by default. So I should probably just make the dummies myself?
Marieke Vandesteene, what you need to do is include only four of these five dummy variables as predictors in your model and estimate it using linear regression in SPSS. Many of the code-based software (e.g., R, Stata) does this automatically in the estimation command/function.
A much easier approach is given at this link: Mediation moderation research virginia R - Bing Kudos to the research group at Univ of Virginia
D. Booth PS R is a free download and
If you need help with R, this should do it: R for Everyone: Advanced Analytics and Graphics | Lander, Jared P. | download (1lib.us)
Best wishes, David Booth
David Eugene Booth, thank you for this resource for downloading free ebooks.
Mehmet Mehmetoglu Please thank the state of California and all of the other contributors. Best, D. Booth
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS