Choosing a test for questionnaire analysis I want to analyse the data from the following experiment setup:
Participants were divided into three groups, each group read a story and had to anwser a question by choosing option A or B. The age, education and gender were also recorded.
Each group had approximately 50 participants.
I want to find out if there were significant differences between the ratios of A/B in all three goups (how people from different groups answeared). What test(s) would be best to find that out?
Cheers!
 A: You can use logistic regression to do this.
Let the answer $A$ denote a positive outcome, and $B$ a negative outcome (in logistic regression, 1 and 0 respectively).  We can examine how group membership changes the probability of answering $A$ adjusting for age, education, and gender.
The model would be
$$  \operatorname{logit}(p) = \beta_0 + \beta_1 \mbox{age} + \beta_2 \mbox{edu} + \beta_3 \mbox{gender} + \beta_4 \mbox{is in group 2} + \beta_5 \mbox{is in group3} $$
Here, the groups have been treated as a categorical variable, using group 1 as the reference group.  I suspect you might also want to make education a categorical variable, but I'm not sure how you coded that.
In this model, the reference group is group 1, and so the tests for coefficients 4 and 5 will tell you if there is a difference between groups 1 and 2 and groups 1 and 3.  If you're interested in differences between 2 and 3, you'd need to do an adjustment to the tests to account for multiple comparisons.  If you're using something like R, the emmeans package is helpful for this.
A: The Chi-squared test will do the thing. If there are differences found, a post-hoc pairwise comparison will be required.
