st: quantile regression - shifting values for effect of school norms

[Lisa Marie Yarnell <lisa.yarnell@utexas.edu>]

Hi all,

I used quantile regression to associate school norms for BMI with
individual BMI, at select moments of the gender- and grade-based BMI
distribution (.25, .50, .75, .85, and .95 quantiles in each grade in
the study, by gender,).

With no covariates, the effect of school norms for BMI on individual
BMI, both overall in a basic correlation, and for the select moments,
is positive (what we'd expect).

But when I add covariates such as age, dummy-coded ethnicity,
socioeconomic status, and stratum (to control for sampling effects
within stratum), the effect of BMI norm scores becomes negative.

I am not sure why the sign of the effect reverses upon entering
covariates.  I have tried removing each covariate separately, and have
tried different combinations of covariates, but the sign always
switches.

But again, with no covariates, the effect of school norms is positive.
 I could really use advice on why this is happening.

Thank you,
Lisa M. Yarnell

-- 
Lisa M. Yarnell
Department of Psychology
University of Southern California
3620 McClintock Ave.
Mailroom SGM 501, Office 826B
Los Angeles, CA 90089-1061
Mobile, preferred: (617) 548-2893
Office: x0-0850 or (213) 740-0850
Fax: (213) 746-9082
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/