0
$\begingroup$

In my study, participants saw a picture of a man or woman either with or without a cigarette. So I have a 2(Male, Female) x 2(Smoker, Non-smoker) experimental design.

Coding: Male=1, Female=2, Smoker=1 and Non-Smoker=2

The question is whether participants would choose to (1) Date, (2) Not Date or (3) Become Friends (with the person on the picture).

I have put Gender and Smoking Status into the Co-Variate Box in SPSS and put the DV into the DV-Box with (2) Not Date as a reference category.

I used a cutomized model, that is, I put Gender, Smoking Status and Gender*Smoking Status into the model by using the forced-entry method.

After analyzing the data I found an interaction effect for "Date vs. No Date":

b= -1.56, Exp(B)= .340, p= .034.

Questions:

Based on these coefficients, would I be able to interpret the interaction effect? (I don't know how to do this).

Or would I need to do follow-ups to see if there are simple main effects?

$\endgroup$
1
$\begingroup$

An interaction effect means that the relationship between the DV and the IV is different at different levels of the other IV. So, the effect of smoking is different for men and women and the effect of sex is different for smokers and nonsmokers.

You should (very nearly) always include the main effects when you include an interaction

The easiest way to see what is going on is to get the predicted probabilities of each combination. I don't know how to do this in SPSS, but there is surely a way. Different programs code the dependent variable differently and this can reverse the meaning of the interaction.

$\endgroup$
  • $\begingroup$ Thanks Peter! I will do that then. My data also shows a strange pattern. The likelihood ratio test showed that the overall relationship between the predictor variables (gender, smoking status and gendersmoking status) and dating(DV) was not significant (p=.076). However, the likelihood ratio test for the interaction gendersmoking status IS (p = .020)! How can that be? And can I still interpret the interaction even if the overall likelihood ratio test is NOT significant? $\endgroup$ – lost_in_statistic Sep 7 '15 at 12:13
  • $\begingroup$ Don't get too hung up on "significance" and especially not on exact cutoffs. Look at effect sizes. $\endgroup$ – Peter Flom Sep 7 '15 at 12:29

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.