One-sided t test regression

[Ursa Ferjancic]

Hello everyone,

I don't have much experience with STATA and I'm working on my master's thesis. I would like to calculate one-sided t tests as "automatic" as possible. I found some instructions how to calculate it manually. Is it possible to get an output table like you get when using function regress, only with both, one and two sided t tests?

Thank you for your answers in advance.

[Jean-Claude Arbaut]

What do you mean by "manually"? Stata has the ttest command, which I think does what you want.
You could automate a bit if you need to apply the test to many variables, for instance:

foreach v of varlist * {
    ttest `v'==0
}

If you require something more specific, you must tell us more about your setting. For instance, it's possible to write a program to (silently) apply the test to several variables, store results and print a single table of t- and p-values. Is that what you need?

[Ursa Ferjancic]

Thank you so much for your response. Yes, I would need that what you wrote.

Also, I will need to provide a regression output table (the one that appears when you simply do for example: reg y x1 x2...), only that some variables will have t- and p- values for one-sided test and others for two-sided test?

(Right now I'm working on a list which variables would be tested one-sided and which two-sided, that is why I can't give you more specific info, when I'm done I can post it)

Thank you.

[Jean-Claude Arbaut]

Ok, I'm not sure there is exactly this in base Stata or in SSC, so unless someone comes up with a solution, I'll have a look tonight.

There are ttable and ttable2 on SSC, but apparently only for mean comparison between two groups.
There is also this in the FAQ: The results from estimation commands display only two-sided tests for the coefficients. How can I perform a one-sided test?

Just to be sure: you are comparing several variables (or coefficients in the case of regress) to the same constant, but with either two-sided or one of the two one-sided comparisons?

[Ursa Ferjancic]

Ok thank you. No worries, I will try to do it following instructions from FAQ.