I am trying to study the relationship between person level demographic and income factors, and homeownership. I have turned homeownership into a binary variable, in order to utilize logistic regression. However, homeownership is a household level variable, whereas the demographic factors are person level data. Homeownership is my dependent variable, whereas the other demographic factors are explanatory. Which weights (person, household, or both) would I use so that I am able to run this model? Thanks.

# Household versus person weights in analysis

**Tim_Moreland**#2

Your decision of which type of weight to use depends on your unit of analysis. It sounds as if you are estimating the effect of certain person-level characteristics on the likelihood of a person owning a home. In this case, your unit of analysis is the person, and you should use person weights.

If, on the other hand, you were creating a model that was determining the likelihood of a household being owned, based on the characteristics of its occupants, then your unit of analysis would be the household. In this case, you would use household weights and restrict your sample to one observation per household (i.e. PERNUM=1).

Hope this helps.