Household analysis

[Lilly Tee]

Hi, I currently have data that is structured at an individual level but is provided by the household. I am interested in how parental income may have an effect on children's health. I separated data on individuals and merged it with a separate dataset on children. With the merge, I have been able to match children and parents from the same household. With the matching, I can see how many children the parents have. For example, some may have 1 or 2 children.

Since I am looking at the effect of parental income on children's health, would it be best to take an average of the children's health if there is more than 1 child related to the parent? In addition, would I have to take household averages than for all variables such as highest qualification, and so on. I'm slightly confused

[George Ford]

I think you're good including multiple kids per household, since you're interested in health-income relationship.

You could use highest qualification for a parent, use parent1qual and parent2qual as regressors, or have a category 0, 1, 2, indicating number of parents with the qualification. You'd have to think about how to handle single parents. Could use % of parents with qualification, and I'd probably include a single parent dummy (which is probably relevant in any case).