Dear Listmembers,
I have a question concerning the areg command and missing values.
Assuming I am looking at different studies conducted in 15 countries on newspaper readership. All studies are being conducted slightly differently, but the all have age as the main independent variable and all found that age has a big positive impact. They are all based on OLS regressions with varying control variables.
I now want to conduct some kind of meta analysis, where I get some sense as to how much age really matters across all countries. From all I know, taking the results (i.e. equations) of all 15 countries and trying to find an 'average coefficient' for age won't work due to the different control variables. If I have the raw data sets, I could create a pooled data set and then use the areg command with a categorical country variable. My problem now is that I have different control variables and they have missing values if they were not assessed in a respective country, e.g. in Germany, we used education as an important control variable, but the French data set does not have that, but they included employment status and so forth.
Are there other ways to do this kind of analysis that allow me to use differing control variables and that are more appropriate?
Any suggestion is welcome.
Thanks,
Matthias Kretschmer
*********************************************************************************
Dr. Matthias Kretschmer
Projektleitung Zeitungsmonitor
ZMG ZEITUNGS MARKETING GESELLSCHAFT mbH & Co. KG
Schmidtstra�e 53, 60326 Frankfurt am Main, Germany
Telefon +49 69/973822-65 Fax +49 69/973822-529 65
E-Mail [email protected]
http://www.zmg.de
http://www.zeitungsmonitor.de
*
* 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/
