Attrition in administrative data

[chiara.mussida]

Dear all, I estimated a xtcloglog and therefore I controlled for random effects. However, my panel data from administrative source might be affected by the problem of attrition. Is there a way to control for attrition with such a kind of model (xtcloglog)? By looking into the Stata manual and sources there seems not to be a way. My questions therefore are:
Do you know whehter is it possible to deal with the issue of attrition in panel data with a similar model, like a xtlogit? If it is no the case: do you have alternative solutions?

Thank you very much for your kind attention.

[Dave Airey]

I thought Stata 13 had improved censoring capabilities? Maybe this: http://www.stata.com/stata13/censored-outcomes/?

[Steve Samuels]

Inverse probability of attrition weighting might (or might not) be a solution. See: Wooldridge, 2002, p. 28-29 and Woodridge, 2007, p. 1293. There's a Stata example in Baulch and Quisumbing (2010), which I have not read.

References:

Bob Baulch and Agnes Quisumbing (2010) Testing and adjusting for attrition in household panel data
http://www.chronicpoverty.org/upload...N-revfinal.pdf


Wooldridge, Jeffrey M. 2002. Inverse probability weighted M-estimators for sample selection, attrition, and stratification. Portuguese Economic Journal 1, no. 2: 117-139. http://citeseerx.ist.psu.edu/viewdoc/download?rep=rep1&type=pdf&doi=10.1.1.198.5398

Wooldridge, Jeffrey. 2007. Inverse probability weighted estimation for general missing data problems. Journal of Econometrics 141, no. 2: 1281 - 1301. http://www.econstor.eu/bitstream/10419/79298/1/386079048.pdf