dunnet test using Stata

[Jeong-Kweon Park]

I am doing dunnet test with Stata.
When I do the dunnet test with Stata, it seems like I could use pwcompare command for the case of balanced data.
And the user written command, dunnet seems like same as this.
What if I do the dunnet test with unbalanced data, is there any way that I can modify the data?
Please let me know if anybody know about the solution.
SAS and R seems like have the way to modify them.

<dunnett test result>
SAS, R p-value: 0.686, 0.175
STATA p-value: 0.695, 0.179

[Nick Cox]

As already pointed out in http://stackoverflow.com/questions/3...st-using-stata the test concerned is Dunnett's test and the program concerned is dunnett (search dunnett for download location).

[Jeong-Kweon Park]

I tested my file using dunnet command. But I saw error message.
"sample sizes of each group must be equal"
So I can't go to the next step.
Could you help me?

[Nigel Moore]

I also have the same problem, but Nick's response does not actually address it, and the stackoverflow link is broken.

So, how do you run Dunnett's test on unbalanced data in STATA?

[Nick Cox]

The original thread on Stack Overflow isn't broken, but rather just not readable without enough "reputation" (a numerical score) on that forum. But you're missing nothing by not being able to read the original, which has the same content as here.

To summarize what little I know:

The program dunnett is downloadable via

net describe dunnett, from(http://www.stata.com/users/tboswell)

and visible via

type http://www.stata.com/users/tboswell/dunnett/dunnett.ado

Despite the provenance this program qualifies as user-written (and Theresa Boswell no longer works at StataCorp, so the email address you may find is broken).

The program does insist on equal sample sizes,

I can't advise further.