Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
From | Bosco Rowland <bosco.rowland@deakin.edu.au> |
To | "statalist@hsphsun2.harvard.edu" <statalist@hsphsun2.harvard.edu> |
Subject | st: elasticity and odds ratios in xtmelogit |
Date | Tue, 20 Aug 2013 06:44:25 +0000 |
Hi Stata list, I have a question pertaining to elasitcities in an xtmelogit model. I have run an xtmelogit regression with imputed data. I have identified a significant interaction with two continuous variables: mi estimate, or var : xtmelogit ind14_3e saq05 saq06 seq09_r PRfud ind10_8a ind14_3a metro_reg c.saq01##c.package_den || lga1: it estimates consumption in last 30 days(yes or no: ind14_3e) package density (package_den) and age (saq01) are the key IVs, the others are covariates,, LGA is the cluster variable I have used margins (ey/ex) to understand the interaction, however, as the data uses imputed data I have had to write a wrapper program based on the following information on the UCLA website http://www.ats.ucla.edu/stat/stata/faq/ologit_mi_marginsplot.htm The basic margins command is based on the following margins, eyex(package_den) at(saq01= (2 (1) 7)) atmeans asbalanced post predict(xb) This has produced margins which demonstrate proportional change in the IV which is consumption in the last 30 days (ind14_3e ) for a proportional change in package_den (package density outlet) for each value of saq01 (age) ------------------------------------------------------------------------------ | Delta-method | ey/ex Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- package_den | _at | 1 | .5296682 .1764275 3.00 0.003 .1838768 .8754597 2 | .4362936 .1559193 2.80 0.005 .1306975 .7418898 3 | .3027863 .14129 2.14 0.032 .025863 .5797095 4 | .09619 .158365 0.61 0.544 -.2141997 .4065797 5 | -.2661349 .2630258 -1.01 0.312 -.781656 .2493861 6 | -1.066769 .6232084 -1.71 0.087 -2.288235 .1546972 ------------------------------------------------------------------------------ However, a reviewer has stated the following: "When interpreting the results, it would be better to specify the ORs rather than just writing the signs (e.g., how much percentage of alcohol consumption was increased)". I am not sure whether the reviewers suggestion is valid. I do not think you can convert the elasticity into an odds ratio. Is anyone able to shed some insight into what is being asked and how it can be done? Thank you in advance Bosco Rowland, PhD, MAPS Alfred Deakin Post Doctoral Research Fellow School of Psychology, Faculty of Health Deakin University Melbourne Burwood Campus, 221 Burwood Highway, Burwood, VIC 3125 +61 3 92443002 bosco.rowland@deakin.edu.au www.deakin.edu.au Deakin University CRICOS Provider Code 00113B Important Notice: The contents of this email are intended solely for the named addressee and are confidential; any unauthorised use, reproduction or storage of the contents is expressly prohibited. If you have received this email in error, please delete it and any attachments immediately and advise the sender by return email or telephone. Deakin University does not warrant that this email and any attachments are error or virus free. * * 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/