Charlotte Manly wrote:
> I generally distinguish between nominal ("categorical"), ordinal ("ranked") and
>interval/ratio ("continuous") data. The interval/ratio distinction is not important
>for selecting a statistical test;
Consistent Charlotte's comments, SPSS (and perhaps other statistical packages) allows
a variable type to be defined as nominal, ordinal, or *scale*.
> however, you shouldn't compute a percent change on interval data (e.g. 20% reduction
>in RT is okay, but 10% improvement in SAT scores doesn't make sense since there is no
>true zero). However, I am teaching at the graduate level.
This gets back to a point tht is difficult for students, and probably the best reason
for discussing Stevens' scales in statistics or methods classes. That is, we can
always do mathematical operations with numbers, the question is whether they are
interpretable. The numbers need to be recognized as secondary to the things that they
represent.
--
********* http://www.coe.uca.edu/psych/scoles/index.html ********
* Mike Scoles * [EMAIL PROTECTED] *
* Department of Psychology * voice: (501) 450-5418 *
* University of Central Arkansas * fax: (501) 450-5424 *
* Conway, AR 72035-0001 * *
*****************************************************************
---
You are currently subscribed to tips as: [EMAIL PROTECTED]
To unsubscribe send a blank email to [EMAIL PROTECTED]
