Hi guys,

I currently have 4 variables all of which are binary (0,1). They all measure the presence/absence of a psychological characteristic in subjeccts. Variables 1 and 2 belong to one "cluster" of psychological traits and variables 3 and 4 belong to another.

So assuming we have trait presence/absence rates as the follows

Variable 1 - 60%, 40%
Variable 2 - 55%, 45%

Variable 3 - 40%, 60%
Variable 4 - 50%, 50%

What I would like to do is to say that the proportion of showing cluster 1 psychological traits in general is higher than that of cluster 2 traits. By a simple addition, we can say that 57.5% of subjects exhibit cluster 1 traits as opposed to 45% in cluster 2 traits. Is there a way i can test for significance between this 57.5 and 45? The alternative is for me to run a McNemar test for proportions, but this would require 4 runs, ie Var 1 * Var 3, Var 1 * Var 4, Var 2 * Var 3, Var 2 * Var 4.

Many thanks guys!