Chi-squared for multiple groups

Dear all

I have data on gender (male/female) for 4 groups that I would like to compare.

I found this previous post that shows how to compare 2 group (or proportions): Chi squared

Does anyone know how to do this?

Best,

Eric

Answer 1

[Jeff Miller]

The crosstab function will do this--it is not limited to 2x2 tables. Example:

% Make up example data for 200 participants, coding gender as 1/2 and group as 1/2/3/4:
n = 200;
Gender = randi(2,n,1);
Group = randi(4,n,1);

% run the chi-square test:
[tbl,chi2stat,pval] = crosstab(Gender,Group)
% output looks like this:
tbl =
    34    24    22    26
    21    30    23    20
chi2stat =
        3.838
pval =
      0.27949

Table shows you the number in each of the 2x4 combinations of group and gender.

chi2stat is the value of the chi-square statistic, here with 3 degrees of freedom. The null hypothesis being tested is that the proportions of males vs females are the same for all four groups. Or, equivalently, that the proportional distribution across the 4 groups is the same for males vs females.

The null would be rejected if pval is small, typically less than 0.05.

Answer 2

[Star Strider]

Consider Perform One-Way ANOVA instead.