We are analyzing a dataset where we are looking at the probability of individuals moving house, and the factors that predict moving.

The problem is that a lot of people are married, and (usually) if one moves house, one's spouse also moves house (although not everyone is married). We have two options, neither of which seems satisfactory:

  1. Model at the household level: but then including predictors at the individual level is hard (not everyone is married).
  2. Model at the individual level, and use something which corrects for clustering. Doesn't seem to account for the nearly complete clustering, and I don't feel like the stimates will be appropriate.

In addition, the probability of moving is curious. If a couple moves, and a single person doesn't, was that a 0.5 probability of moving (because 1/2 households moved - as in (1) above, or 0.66, because 2/3 households moved, as in (2) above.



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.