I simply want to create a measure for the percentage of children under age 18 who are
the sons or daughters of a householder—male or female—without a spouse present in the home for the 1990 and 2000 IPUMS and the 2008-2010 ACS (e.g. the percentage of children in single parent families). There are many “family type” variables in the IPUMS so I want to make sure I am using the correct measures. I thank you in advance for your help.
There are a number of different ways you could go about finding the percentage of children living with a single parent. First you will want to create a variable to identify individuals as single-parent householders using the variables RELATE (RELATE==1 identifies householders), MARST to identify individuals who are not currently Married with spouse present, and NCHILD to identify if there are any children present in the household (note that NCHILD counts children of any age).
Since you are only considering single-parents who are householders, you can then drop all non-householders from the dataset and save the resulting dataset as a separate file. Since you have only kept householders, this new file should act as a household-level dataset with your new variable identifying households with single-parent householders.
You can then merge your new household-level dataset back onto the person-level dataset using YEAR, DATANUM, and SERIAL to attach your new variable to each person within a household.
Now, use the RELATE variable to identify all of the householder’s children (RELATE==3 identifies children of householder and RELATE==4 identifies children-in-law of householder) and use the newly merged variable to identify if they are the children of a single-parent householder (and are 18 or under using AGE). Then simply take the weighted total of Children of Single-Parent Householders and divide it by the weighted total of all children.
I hope this helps.