Identify Outlier

#1
Hi,

I need help to identify outlier for my research project :)

Model: 2x2 moderated mediation
1.) IV: both are nominal (rich vs. poor & public vs. privat) --> 4 scenarios were created -- one of the IV is a moderator (public vs. privat)
2.) two mediators (interval scale)
3.) two DVs (interval scale)

How do I identify outliers in SPSS?
I have no clue because I have two nominal IV's. :yup:
I tried to identify them with mahalanobis.. but I got only 4 values (because of 4 scenarios) - I think thats not right :/
 
#2
Maybe I start with a easy question:

I have 2 independent variables and 2 mediator and 2 dependent variables.

How do I identify outliers?
- Do I analyze the two independent variables and the two mediator variables together?
- Where do I get the cut-offs of Mahalanobis & Cook-Distance and so on? :)
 

noetsi

Fortran must die
#3
You should start with what method you are running :)

Are you running two DV in the same equation, a few method will allow this, or one DV at a time? Are you running linear regression, ANOVA....

More details are required to answer this question. Including if the mediator variables are in the model with the other predictors.

There is disagreement about acceptable values for Mahalanobis distance and Cook's D. Also if the data is not normal you need to run a robust Mahalanobis distance not the general one.