Simple mean test by category

[oksanakim]

Hello everyone

Please help with the following, should be a simple program but I do not do any programming with Eviews. For the data series called Ret collected over 25-month period, I need to perform a simple hypothesis test that the mean Ret for each month is not equal to zero. So 25 separate mean tests are required, but is there any more efficient way to do this via say programming? The separate Month categorical variable was crated. How can I do that if at all? Much appreciated!

[startz]

You can get the mean, standard deviation, and number of observations from View/Descriptive statistics/Stats by classification and using month as the classifying series. That gets most of the way.

[oksanakim]

That's right but then how can I test that each of those 25 monthly means is statistically different from zero? My original question was more about how to perform a simple mean equality test for each month...

[startz]

I'm not sure of the question. Wouldn't you create a t-statistic for each month the way you would for another other mean test?

[oksanakim]

I'm not sure either we are on the same page. A simple hypothesis test allows me to test whether the mean of the entire population is equal to, say, zero - t stats with probability is produced in this test. But what if I need to perform the same mean test for 25 sub samples from this population - separately for each month? Can I do this in Eviews without the need to manually break down the population into 25 sub samples and then separately performing a simple hypothesis test that each of 25 means is equal to zero? Thanks!

[startz]

So you want a joint test that all 25 means equal zero?

[oksanakim]

I want 25 independent tests that each of monthly means is equal to zero.

[startz]

Code: Select all

ls Ret @expand(month)

will give you a t-statistic for each month, assuming a common variance across months. Or you can use the stats-by-classification output to compute a t-statistic for each month

[oksanakim]

Thanks a lot for your help! Much appreciated.