Primary and spousal analysis STATA

I am analyzing information on uhrswork Nash eqm. decision matrix for husbands and wives. As of now, I have 2.2million uhrswork(male) observations with 2.2million linked uhrswork_sp(female) observations, and 1.7million uhrswork(female) observations with 1.7million linked uhrswork_sp(male) observations. I assume that the difference in sex of the primary person is due to more male-identifying spouses filling out the survey or being listed as head of household.

I want to somehow combine all male observations with each other (primary and _sp) and all female observations with each other (primary and _sp) so I can tabulate data from all couples together. Right now I only know how to tabulate information for primary husbands and secondary (_sp) wives OR primary wives and secondary (_sp) husbands. I want to combine these 4 variables into just 2, one for each identifying gender, without losing the ability to analyze spousal actions through actions of the primary.

Thank you for the help!


I’m not sure I entirely understand your question here. When using the attach characteristics tool, all observations will be (probabilistically) linked with their spouse, if one exists. That is, in a “traditional” household, the husband will be linked with the wife on one line in the data and the wife will be linked to the husband on another line in the data. This means that the data come to you, by default, as a person-level data set with each couple showing up twice in the data.

If you want to create a couple-level data set, (e.g., so that each couple only shows up once in the data), then you will need to systematically drop half of the couples. There are a number of ways to do this. If you are okay keeping only heterosexual couples, then an easy method is to keep only observations who are male (via SEX) and spouses who are female (via SEX_SP). Alternative methods include only keeping one couple per household or using SPLOC to systematically drop spousal observations from a couple-level data set.