Hi,
I have a problem calculating a mediator effect with the help of PLS.
My model looks like this:
Experimental Situation ---> Leads to certain Emotions ----> Leads to certain Behavior.
I would now like to control for the indirect effect of the situation on the behavior. Therefore, I tried to make dummy variables out of the experimental situations, which initially have been 3 different situations (control group, experimental group A and experimental group B).
I made one variable with value 1 for the control group and value 0 for the two other groups and one variable with value 0 for the control group and value 1 for the two other groups.
When now running the analysis, it says that "a singular matrix occured" (path weighting scheme) or that an error "during the outside estimation occured" (factor weighting scheme). And I really don't know what it does there when choosing for the Centroid Weighting Scheme.
The other two variables consist 6 or 5 items respectively and are on a metric level.
Who can help me in that matter? Thank you in advance!
Mediating effect with Dummies
-
christian.nitzl
- PLS Expert User
- Posts: 248
- Joined: Sat Jul 25, 2009 1:34 pm
- Real name and title:
Hi,
In my experience such an error message often occurs when one or more variable has a to low variance especially when there are one or more items with almost the same number for all cases. Therefore it could be possible that one of your dummy variable cause the problem. Delete one by one of the variable to find out which one cause the problem. Then throw a closer look if there is enough variance in the variable. Another problem could be also that your sample is too small.
Greetings,
Christian
In my experience such an error message often occurs when one or more variable has a to low variance especially when there are one or more items with almost the same number for all cases. Therefore it could be possible that one of your dummy variable cause the problem. Delete one by one of the variable to find out which one cause the problem. Then throw a closer look if there is enough variance in the variable. Another problem could be also that your sample is too small.
Greetings,
Christian