Hi

On Wed, 16 Mar 2005, Claudia Stanny wrote:
> Jim Clark wrote:
> >Do the t-test that Christopher Greene originally suggested.  The
> >idea that ordinal data cannot be used for such analyses is a myth
> >in many (most?) people's minds.
> 
> True.
> The numbers don't know where they came from and the
> statistical procedures don't ask (or care). The problem is in
> the interpretation of the results.

> Suppose "problems in school" is measured on a 5 point scale (0=none;
> 5=serious).
> The mean for one group is 2.3.
> This might be taken to imply that the typical student has a moderate number
> (or
> severity) of problems in school.
> However, although large numbers of students might have no problems in school
> there might be enough students in the sample with serious problems to distort
> the value of the mean.
> A Chi-square analysis would not lead you into this incorrect interpretation.

This is not a problem that arises from the t-test analysis of
ordinal data, nor is the problem limited to ordinal data.  If one
school has an average height of 5.9 and another an average height
of 6.4, it could arise either from one school having people who
overall are taller than students in the other school or from one
school having a larger percentage of very tall students.

Claudia is right of course that interpretation requires a good
dose of common sense.

Best wishes
Jim

============================================================================
James M. Clark                          (204) 786-9757
Department of Psychology                (204) 774-4134 Fax
University of Winnipeg                  4L05D
Winnipeg, Manitoba  R3B 2E9             [EMAIL PROTECTED]
CANADA                                  http://www.uwinnipeg.ca/~clark
============================================================================


---
You are currently subscribed to tips as: [email protected]
To unsubscribe send a blank email to [EMAIL PROTECTED]

Reply via email to