The problem with testing the hypothesis is that most of the traps can only be showed in group comparisons using ordinal test like chi-squared.

I will give some examples of the experiments I want to do (with an online survey):

Anchoring

The group that has to estimate the result of 8x7x6x5x4x3x2x1 generally estimates the result to be 2250 while the group that has to estimate 1x2x3x4x5x6x7x8 genereally estimates the result to be 512.

Omission bias

The decision whether or not to vaccinate if the chance of dying from the disease is 10% and the chance of dying from the vaccination is 5%.

The chance of dying from vaccination has to be substantially lower then the chance of dying from the disease before people want to vaccinate. The experiment would be to vary the levels of the chances ni different editions of the questionnaire.

De-escalating commitment

The problem is whether to complete a project or not. There are two projects which each, when completed, will result in a return of 2.000$. The first project has cost 1.750$ up to now and will take another 500$ to complete. The second project has cost 500$ up to now and will also take 500$ to complete. Which projects should be completed: none, the first, the second or both? People tend to complete the second project more often than the first one.

My intention is to determine the correlation between the occurence of these decision traps. The hypothesis is that within a causal group the correlations between the traps are higher than between decision traps in different causal groups.

Problem is: which test do I use to determine the levels of correlation: Chi-square, Kendall's tau, spearman, Cramers V, odds ratio, anything else?