How to write this kind of ttest syntax

[Eddie Pengma]

Dear All,

Thank you sincerely and I run the following program.

sysuse nlsw88.dta, clear
tab industry collgrad, sum(wage) means

stata shows that:

tab industry collgrad, sum(wage) means

Means of hourly wage

| college graduate
industry | not colle college g | Total
-----------+----------------------+----------
Ag/Forest | 5.3122251 7.0626361 | 5.6211211
Mining | 15.349593 . | 15.349593
Construct | 6.3992254 17.66774 | 7.5649338
Manufactu | 6.8861197 13.730761 | 7.5015779
Transport | 11.195142 13.225958 | 11.443353
Wholesale | 5.8526317 8.2472956 | 6.1258966
Finance/I | 9.4961749 11.875596 | 9.8431737
Business/ | 6.4778803 11.728484 | 7.5157904
Personal | 4.3603053 5.679114 | 4.4010932
Entertain | 5.9703888 8.5340567 | 6.7244088
Professio | 6.2705725 10.093487 | 7.8711858
Public Ad | 8.2137595 12.039293 | 9.148407
-----------+----------------------+----------
Total | 6.9259835 10.543932 | 7.7834633

.
end of do-file

I want to test for a specific industry, whether the salaries between those who attend the universities and those who do not are different or not. For example, for Ag/Forest, the salary of people who do not attend colleges and who do are

5.3122251 and 7.0626361 respectively. I want to write a ttest command to see whether it has any difference or not.

How can I write down this syntax?

Best regards,

eddie

[daniel klein]

Well, the two means are obviously different, as you have already established. However,

ttest wage if (industry == 1) , by(collgrad) unequal

provides no evidence that the difference is statistically significant at any conventional level.

Best
Daniel