t-test on categorical/factor variables

[Juliane Dold]

Dear community,

I once again have to ask for your support.
I use Stata 18 on a Windows System.

As the caption suggests, I want to carry out a t-test comparing the mean values of a categorical variable.
The data looks like this:

i.Education	mean control	mean treatment	mean difference
Education == 1 (no secondary Education)	0.08	0.08	0
Education == 2 (secondary Education)	0.7	0.61	0.09
Education == 3 (tertiary Education)	0.22	0.31	-0.09

Additionally, the test has to be restricted by sex (either ==1 or ==2) and post_reform == 0
for "regular" variables, the ttest command sufficied, but I have not been able to find a solution for the categorical variables.

Thanks in advance!

[Maarten Buis]

What is the null hypothesis you want to test?

[Maarten Buis]

In addition to the exact null-hypothesis, could you give us an example from your data? How to do so is discussed in the FAQ (black bar near the top of this page).

[Juliane Dold]

thanks Maarten Buis for the reply.
My data looks roughly like that:

PersonalID	Year	Education	sex	treatment	post_reform
1	2005	[1] no secondary education	[1] male	[1] yes	[0] no
1	2006	[1] no secondary education	[1] male	[1] yes	[1] yes
2	2005	[3] tertiary education	[2] female	[1] yes	[0] no
3	2005	[2] secondary Education	[1] male	[0] no	[0]

i want to test the mean difference of the treated women/men vs the women/men in the control group.
My null Hypothesis for this case would be H: mean treated and mean control are significantly different.