measuring same variables for unrelated topics within subjects


New Member
Hello everyone,

Purpose of the study: I want to find out how the perceived competence and warmth of an activist group relates to whether someone (the participant) wants to become active in that activist group.

Questionnaire structure:
A participant will receive an instruction "please imagine environmentalists" and then fill out questions about whether these are competent and warm and whether they would go to a demonstration against climate change.
Next the participant will receive an instruction "please imagine feminists" and fill out the same questions.

Each participant gives information for six different groups f.e. climate change, sexism. These groups are presented in random order.

nature of the data:

independent variables:
1) group
2) Competence of a group
3) Warmth of a group

dependent variable: protest intentions (f.e. going to a demonstration) for a certain group

in the data the variables look like this

The variables are measured via questionnaire questions, for each variable 3 questions.

My question:
What I can do: I calculated reliability for each scale (competence_group1, warmth_group1,...) per group. Also calculated correlations between competence, warmth and protest intentions for each group, and regression analysis per group. so far everything okay.

What I don't know how to do: I want to calculate an "overall" correlation between competence, warmth and protest intentions (or a regression) ACROSS all groups. Also I would like to check reliability of the overall "Competence"-Scale and not just of the competence_group1, comeptence_group2, ... separately.

Does anyone have an idea what kind of statistical model or analysis I am dealing with? It seems somewhat like repeated measures, but I am not measuring the same variable more than once, since it always concerns a different political group...

I hope someone can just give me a clue where I can start looking.




TS Contributor
You could consider multivel modeling. It permits regression of a dependent variable on a predictor, if DV and predictor have been measured several times in a participant.

With kind regards



New Member
Hello People!
I am still stuck. I read up on multilevel modelling somewhat, but most of what I find refers to different levels of subjects, e.g. pupils in different classes in different schools --> 3 levels. What I did not really find is what the term is for what I want to do. If I had the term I could read up on it better.

What I found as a fitting description: "Repeated Measures Nested Within Individuals" - BUT this is usually repeated measures over time. What I need is not repeated measure over time, but IV1 and DV measured 6 times, because IV2 has 6 levels.

Please help me! I just need a name for what I am doing so I can research it....