I have a survey dataset with an ordinal dependent variable that is left hand
censored. The dependent variable measures a latent construct (y*) that is
linearly related to all x's in the model. I am using the svyintreg command
to estimate the effect of a binary independent variable and several ordinal
independent variables. My question regards the interpretation of the
coefficients. Ideally I would like the metric to be: for a one unit
increase in (or being in the category coded 1), the dependent variable will
increase or decrease by z%, holding all other variables constant.
I read J. Scott Long's 1997 book on Regression Models for CJDV's and he
presents an equation on page 22 for log-linear models which states that
percent change=100[exp(B1)-1]. Will this transformation work for the
svyintreg coefficients as well? If not, any suggestions would be
appreciated.
Thank you.
Tom Brewer
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
