ATUS issue with married and unmarried workers


I have been using the ATUS data to look at how much more unmarried and married full-time workers spend on work. Using the variables SPOUSEPRES and ACTIVITY, I am finding that unmarried workers are working more than married workers. This contradicts what BLS has told me. I am using WT06, and I can’t figure what I am doing wrong. All my other results seem reasonable.


For this type of calculation, it is a bit easier to use a standard, person-level RECTANGULAR dataset (I presume you aren’t using this since you are using the ACTIVITY variable, which isn’t in the rectangular datasets). To get your data in this form, on the variable selection page, there will be a button on the left that says CHANGE DATA STRUCTURE. Rectangular, person-level data will give you one row per person. Then you can select particular activities of interest, and your dataset will have a column for each activity, and the number of minutes spent by that person on that activity in the day.

There are different ways to code the activities, which can be found under the TIME USE variable menu. You could choose bls_work_working from here to use the BLS’s definition of working. Or you can create a new work variable using the detailed components by clicking the yellow CREATE TIME USE VARIABLE button.

You might also consider using MARST instead of SPOUSEPRES for more detail on marital status.

Using the BLS definition and SPOUSEPRES, married men work more than single men (in 2018). And slightly more for women:

Spouse or unmarried partner in         |        Sex
household                              |     Male    Female
                        Spouse present | 4.201753  2.881319
             Unmarried partner present | 5.184365  3.487805
No spouse or unmarried partner present | 3.035134  2.527202

Thanks so much. This worked really well. You guys are the greatest.