Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
| From | Donald Spady <dspady@ualberta.ca> |
| To | Statalist Statalist <statalist@hsphsun2.harvard.edu> |
| Subject | st: Type I errors and multivariate logistic regression |
| Date | Fri, 2 Aug 2013 12:59:58 -0600 |
Dear Statalisters This is not so much an issue directly related to Stata but it is one that I need help on. I have a data set of about 200 cases. The dependent variable is binary and by using logistic regression and after an initial univariate analysis of the variables, my final equation has 6 variables that are significant predictors, and all have reasonable theoretical grounds for inclusion. A reviewer has ask me to comment on how I address the possibility of a Type-I (false positive) error (i.e. accepting a variable as significant when in fact it is not) since if you do enough tests some will be significant by chance alone, and 6 variables may be pushing the envelope.) I must admit I had not even thought about it, as I guided my analysis on the basis of theory and on the univariate analyses. I would appreciate some advice as to how to answer this? Many thanks Don Spady -- dspady@ualberta.ca * * 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/