T-test on weighted mean - error weights not allowed

[Metric2009]

I need to do a test on the weighted average. I get an error message saying "weights not allowed". What are my options. How can I do a test on weighted average in Stata 13?

Thanks,

Josh

[Rich Goldstein]

I assume you mean that the data are weighted and that you have individual data; in that case you can use a regression; if you want a two-sample t-test that is a regression where the only two right-hand side variables are the constant and the indicator variable for the group; all 4 kinds of weights that Stata recognizes are allowed with regression

[Metric2009]

I am not sure how to run the regression to do the t-test. Here is my sample data.

GROUP_ID	MV_EQUITY	RETURN
1	303	0.13
1	1338	0.10
1	1994	-0.03
1	191	0.19
1	418	0.00
1	925	0.01
2	758	-0.02
2	196	0.05
2	1045	-0.01
2	537	0.18
2	151	0.07
3	2364	0.00
3	635	0.08
3	198	0.02
3	430	0.08
3	315	0.03
3	911	0.18
3	9796	0.13
4	551	0.00
4	521	0.05
4	1320	0.09
4	3971	0.10
4	1294	0.06
4	334	0.11
4	892	0.04
5	5865	0.04
5	1251	-0.01
5	847	-0.01
5	1755	0.02
5	461	0.00
5	330	0.14
5	1870	0.07

I need to find the following:

1) For each group I need to calculate the weighted average return (using mv_equity as weights) and obtain the "Newey and West" adjusted t-statistic for the weighted average return == 0 .

2) I need to test for the difference in the weighted average return between group 1 and 5. Here again I need to obtain the "Newey and West" adjusted t-statistic.

I have read in the stata archives that the regression approach can be used when we are using weights to obtain the mean. I am not sure how to set up the regression and get the "Newey and West" adjusted t-statistic.

Thank you,

Josh

[Rich Goldstein]

have you investigated the -newey- command in Stata; start with the help file