Statistics – chi square test in a crosstable

Hi, if you want to test if the frequency distribution differs between groups in a crosstable you can use the most common statistical test called chi2-test or chi-square-test.

How could you understand the results, and when is it “forbidden” to use the test?

You get a lot of help in the results, but you just have to know where you find that. I will learn you that today.

In this example I would like to see if the distribution of employee categories differs between male and female (variable: gender).

I start just to describe it with percentages.

Male and female seems to differ in the distribution, but is the difference statistically significant?

Then I go back to test that and a tip is to use this toolbar command, where you find your latest commands.

Here I add the test by click on the button : “Statistics”
and then click in the box: “chisquare”

Finish by clicking the buttons: “Continue” and then “OK”.

Results:

Look below the test result and check if it’s forbidden to use the test or not. This rule is there to avoid testing on too small materials. Just check so the percentage-number (here:  0%) is NOT bigger than  20%.  And the smallest expected values (here: 5.84 ) is not allowed to be under the value 1.

Here the both rules is OK so it’s OK to do chi-square test, so let’s have a look on the results: Pearson Chi-Square has a significance value that is less then 0.05 (it is very small so you can just see 3 decimals:  0.000).

In some research report you write like this: “The frequency distribution of job categories differs significantly between males and females (p<0.05)”.