[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
st: Re: xtlogit vs gllamm
This message is a follow-up to the one I posted on May 15, 2006.
Following the recommendations on the use of xtlogit that were made to me, I performed a few tests on my data, this time using all the occasions I had for each subject (on average 150 occasions for each respondents) instead of only 8 and I increased the number of integration points. As a result, my data is less correlated. This time I get similar results between gllamm and xtlogit, which leads me to conclude, as suggested, that xtlogit may pose a problem when the intraclass correlation is high highly correlated but otherwise works well.
Mr. Linhart said he was interested in seeing the data. If he still is, I would be glad to send it to him.
* For searches and help try: