EViews.com

Dear Forum,

I'm trying to find a way to streamline 72 mean/median=0 series distribution tests that I need to perform.
I need to determine for 4 variables (call them centrality measures) if their mean and median returns differ significantly from 0.
Each one of these 4 variables has to be split into subsamples (<1st quartile, between 1 and 3 quartile, >3rd quartile)
I have 3 types of returns (acquiror, target, and combined).

To summarize. For each subsample(3) (quartile) of each centrality variable (4) , i need to perfom a mean= 0 and median=o test (2) of each one of the 3 types of returns.
This amounts to 3x4x2x3=72 tests.

Up till now i have done it manually in this fashion: I create a sample of the 1st quartile of one of the centrality measure and perform a t-test for this sample for mean/median and then proceed to the next one.
This is very long and tedious however and I was wondering if there is any way to do this faster?

Many Thanks in Advance!,

ConstantijnCrans

Yep, you'll need to write a loop that loops over the series you want to perform the test on. Then inside that loop there will be code that sets the three samples and freezes the tests into tables.

Yep, you'll need to write a loop that loops over the series you want to perform the test on. Then inside that loop there will be code that sets the three samples and freezes the tests into tables.

Dear Gareth,
I realize that. Any idea what this would look like? My problem is that i find the manual and the programming reference kind of confusing sometimes with respect to finding commands such as to freeze tests into tables or to even perform this particular test. Any possibility you could write a sort of blueprint for what it would need to look like?

much thanks!

If you have EViews 9, use the command capture feature to figure out the commands needed.