Date:         Thu, 12 Oct 2000 12:14:50 -0400
Reply-To:     Big Mar <martin@CINCI.RR.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Big Mar <martin@CINCI.RR.COM>
Organization: OneNet Communications News Hub
Subject:      Re: SAS Stat question-- partially overlapping samples

I think I've managed to do it in a data step using a calculation for t for
overlapping samples I found in one of my stat books, but I still need to
check it by doing a few by hand.  A procedure would still be preferable.

Anyways....Here is what I want to test.

n = 600
I have 11 products.  Respondents rated 5 of the 11 products on a 5 point
scale.

I want to test ALL of the possible combinations . i.e. 1 vs. 2 - 1  vs. 3
... 1 vs. 11, 2 vs. 3 etc....

If they rated all 11 it would be easy...

proc ttest data = one;
paired (q1_1-q1_11) * (q1_1-q1_11);
run;

If this isn't enough info.  I'll be happy to provide more.

Thanks again.



"Jay Weedon" <jweedon@earthlink.net> wrote in message
news:d1gbuscna94jikjehn7lcl3put1tk6eqh8@4ax.com...
> More detail please!
>
> Please provide more information on the ratings, and on what hypotheses
> you wish to test.
>
> On Wed, 11 Oct 2000 16:53:11 -0400, "Big Mar" <martin@cinci.rr.com>
> wrote:
>
> >I am stumped!!!
> >
> >I have data with about 50% partial overlap (i.e. each respondent rated 3
of
> >30 products), so a paired t-test is not appropriate, nor is a standard
> >t-test.  Is there a way, using ANOVA or some other procedure to do this
> >automatically?
> >
> >Any help is greatly apprciated!
> >
> >
> >
>