Hi,
I am now working with a panel dataset consisting of all members of
households (ID) under study. As this implies a certain degree of
non-independence among the observations within the same household, I
have used cluster option (cluster(HHNr)) in all my regression.
However, I have got the suggestions that clustering on the level of
households may not be enough to account for non-independence since
some individuals in the dataset switched from one household to another
and their previous households may also affect and I now want to group
observations in a way that individuals who lived together at any point
during the observed period belong to the same cluster. This would
imply, in some cases, that some individuals who did not ever live
together are grouped into the same cluster, but there would be a
sequence of individuals (one, in most cases) who "connect" these
persons. Does anyone know how I can make a Stata code using ID and
HHNr variables in order to perform this longitudinal grouping of
individuals, as I call it. Of course, the individual identifier (ID)
never changes regardless which household the person lives in. Best
wishes to all,
Dimitrije
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
