Dear Statalist,
my dataset contains:
- data regarding several polls predicting the..
- result of parties of
- several countries
My variable of interest (errors in polls) depend on factors related to each of these three "level".
Polls are not crosscountry waves (but countryspecific).
Thus each poll predict the results of more than one party, and
each party result is predicted by more than one poll (though parties are more than polls)
Data on polls and parties seems crossclassified to me, and they are both clearly nested within countries.
a) Is there a way to deal with crossclassified AND nested data?
b) I read that, even if data are not perfectly nested, Stata13 behaves AS IF they were nested. But I cannot really figure out the meaning and the consequence of this stretching (if I got it right).
Thanks a lot
Marco
my dataset contains:
- data regarding several polls predicting the..
- result of parties of
- several countries
My variable of interest (errors in polls) depend on factors related to each of these three "level".
Polls are not crosscountry waves (but countryspecific).
Thus each poll predict the results of more than one party, and
each party result is predicted by more than one poll (though parties are more than polls)
Data on polls and parties seems crossclassified to me, and they are both clearly nested within countries.
a) Is there a way to deal with crossclassified AND nested data?
b) I read that, even if data are not perfectly nested, Stata13 behaves AS IF they were nested. But I cannot really figure out the meaning and the consequence of this stretching (if I got it right).
Thanks a lot
Marco
