|
|
Hello all,
I hoped someone might be able to help me understand the correct interpretation of my SPSS output for interaction effects in logistic regression. I have included an example situation from James Jaccard's Interaction Effects in Logistic Regression to illustrate my question.
A researcher is interested in how gender differences in the odds of teens engaging in sex differ as a function of the employment status of the mother. So, mother's employment status is the moderator variable and teen's gender is the focal independent variable. For mother's employment dummies, the reference category is unemployed.
Predictor
Logistic Coefficient
Exponent of Coefficient (a.k.a. odds ratio)
Male
0.1015
1.1068
Full-time
0.2922
1.3394
Part-time
-0.8539
0.4257
Male*Full-time
0.0769
1.0799
Male*Part-time
0.9511*
2.5886
Jaccard says, "In the case of Male*Part-time, the exponent of the coefficient reflects the odds ratio for gender for adolescents of part-time mothers (2.8668) divided by the odds ratio for gender for adolescents of unemployed mothers (1.1068). This equals 2.8688/1.1068 = 2.5886 which is the value of the exponent of the logistic coefficient for Male*Part-time. This contrast is statistically significant."
Here's my take on how it should be interpreted, keeping in mind that mother's employment status is the moderator variable: Compared to ADOLESCENTS of unemployed mothers, BOYS of mothers who work part-time are 2.5886 times more likely to engage in sex.
Have I correctly interpreted the odds ratio for the interaction effect?
Any help you could provide is greatly appreciated! Barbara
|
Locked
