RE: st: Nominal or ordinal?

["Polis, Chelsea B." <cpolis@jhsph.edu>]

Many thanks to everybody for their responses.  As Steve Samuels suggests, I think I was confusing ordinal logistic regression with something else.  Thanks to those who pointed out that concern about equal spacing is unnecessary in this case.

To John: I am using weighted survey data to assess sociodemographic correlates of self-perceived likelihood of infertility among young adults (i.e., young adults were asked how likely it is that they are infertile or will have difficulty getting [someone] pregnant, and responded not likely, slightly likely, quite likely, or extremely likely).  I am stratifying by gender, and there are 862 males and 854 females with answers to this question.  I ran Pearson chi-squares, corrected for survey design with second-order correction of Rao and Scott, to determine which variables were significantly associated with perceived infertility.  In multivariate analysis, I  had planned to control for variables associated at p<0.10 in the Pearson chi-square tests (between 4 and 8 predictor variables depending on gender).  These variables represent sociodemographic characteristics such as ethnicity, education level, receipt of welfare, age at first sex, etc.  Any comments on my plan would be gr!
 eatly appreciated.  If I am correctly understanding the responses, it seems as though I can try to either use regress or ologit, depending on whether my data fit the assumptions required of these models?  

Many thanks,
Chelsea

-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of John F Hall
Sent: Friday, August 13, 2010 1:17 AM
To: statalist@hsphsun2.harvard.edu
Subject: Re: st: Nominal or ordinal?

Fascinating thread, but do the data actually warrant such sophistication?

I'd like to know much more about the data, even more about the research 
question.  What is being evaluated by the "likely" scale?  Propensity of 
teenage readers of a magzine to buy a beauty product, people stopped in a 
shopping mall to purchase a health insurance scheme or unemployed people's 
perceived likelihood of finding a job?

How many predictor variables are being used and what are they?  How many 
cases are there: 10 or 1,000?  Is it a probability sample or a "sugging" 
exercise?

As John Tukey once said, "All the statistics in the world won't save you if 
you asked the wrong question in the first place."  (Does anyone have a 
reference for that?  I saw it in the front of a (?statistics 
textbook/software manual?) book once.

John Hall
johnfhall@orange.fr
http://surveyresearch.weebly.com

"Sugging" - selling under the guise.


*
*   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/