Re: st: Comparing two response variables

[Debs Majumdar <debs_stata@yahoo.com>]

Hello,

I should have been more clearer with the question. The regressions  were 
performed for case-mix adjustment. The predicted values from these  regressions 
were then used to to generate ranks.  The process is as  follows:

The data is collected from 10000 students who had answered questions  (ordinal 
items) regarding their teachers (500). There were two total  scores computed out 
of their responses, one with 5 items and one with 7  (same 5 + additional 2). I 
have been told that this new one (total score  out of the 7 items) makes more 
sense qualitatively. I just want to  figure out if one can quantify this.

Initially, case-mix adjustment is done for each of the scores with  respect to 
student characteristics (gender, major etc.) using `-areg'. Then the dataset is 
collapsed by teachers and in the last step the teachers are ranked.

I have the same opinion as Joerg that one one really need ans external variable 
to justify this. The other thing I was thinking if I can use any resampling 
techniques where I generate data corresponding to the null hypothesis that both 
are equally good and so on. I don't know if that's feasible for this case 
though.


Thank you,

Debs


----- Original Message ----
From: Joerg Luedicke <joerg.luedicke@gmail.com>
To: statalist@hsphsun2.harvard.edu
Sent: Mon, February 28, 2011 4:52:15 PM
Subject: Re: st: Comparing two response variables

On Mon, Feb 28, 2011 at 6:20 PM, Debs Majumdar <debs_stata@yahoo.com> wrote:

> Is there anyway to prove Y1 is a better measure for the trait we are measuring
> when compared to Y2?

"Better" with regard to what? As far as I understand your question you
mean something like "better" in the sense of higher validity of your
measurement instrument. Only if you know exactly how the proposed
relation to x1 and x2 is supposed to look like in this multivariate
context, e.g. if your measure was validated before in other studies,
the model could provide some evidence in favor of the one or the
other. In the absence of such knowledge, there is no model that would
tell you that. Basically, you need external and conceptual information
to decide on what the better measure is.

J.
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



      
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/