st: Weighting xtreg regressions

["Mirnezami, Oliver" <O.Y.Mirnezami@warwick.ac.uk>]

Dear Statalist 

I have an unbalanced panel at the individual level over 9 time periods (1994-2010 in 2 year increments). I'm investigating the relationship between health and unemployment, attempting to analyse the causal effect of job loss on health. 

I'm trying to generalise the standard 2 period differences-in-differences method to multiple time periods:

health = constant + job loss + controls + year effects + individual fixed effects + error term 

The coefficient on job loss would be the effect of 'treatment'. I am assuming the health of those individuals that lost their jobs would have evolved in the same way as those that didn't lose their jobs in the absence of job loss. 

The dataset is unbalanced due to some missing values and also the fact that certain individuals enter in different waves and sometimes individuals are not re-interviewed in subsequent waves. 

I wondered about weighting my regression by the number of times an individual appears in the dataset. How can I adapt my code to incorporate this weighting? My unweighted regression code is as follows: 

xtreg  health  jobloss age married income wealth smoking bmi  y1996 y1998 y2000 y2002 y2004 y2006 y2008 y2010, fe vce(cluster id)

My reason for trying this was in case there happened to be some kind of selection issue whereby there could be a systematic relationship between individuals' frequency in the dataset and their health or likelihood of experiencing job loss. 

I would appreciate any advice that you can offer.

Thank you

Kind regards

Oliver 


*
*   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/