I have an experiment design with:
Variable 1 - 2 levels
Variable 2 - 3 levels

And demographic information collected about my participants:
Demographic 1 - 2 levels
Demographic 2 - 3 levels
Demographic 3 - 2 levels
Demographic 4 - 4 levels

Dependant variable - Score on a perception test

The experimental variables show homogeneity of residuals but the demographic variables do not (both Levene's and Bartlett's test give the same outcomes). My ideal test would be 2 tailed anova or regression for the different variable types but i can't do this one for the demographic ones. I have looked separately at the experimental variables using Anova/Tukey, and the demographics using Kruskall-Wilk/Dunn. I also looked at the demographics through Anova/Tukey and the significant/not significant markings match for both sets of tests. I'm getting to the regressions next.

I want to see if there are interactions, specifically between Variable 2 and Demographic 4. The variable is categorical interval groups - the length of stimulus (2 syllables, 4 syllables, 6 syllables, accumulative) and the other is categorical ordinal - the self identified level of awareness/exposure of/to the phenomena that I'm testing (I do this myself, i don't do it know someone who does, etc). Is there a test that would allow this? I'm guessing that I can't take a chance on Anova but it would be nice to have this confirmed. I was thinking about making new groups:

Group A = Variable 2, Level 1 + Demographic 4 Level 1
Group B = Variable 2, Level 2 + Demographic 4 Level 1
Group C = Variable 2, Level 1 + Demographic 4 Level 2
Group D = Variable 2, Level 2 + Demographic 4 Level 2 ... and so on

Then do tests on Score~Group A, Score~Group B and so on.

Would this be an option?


Your Answer

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

Browse other questions tagged or ask your own question.