I would like to use nchild variable (number of own children in household) to choose households that have more than 1 kid, but I noticed that there is a suspicious increase in the average of nchild in 1982.
Attached screen shot shows that average nchild in IPUMS CPS monthly data (all states and DC) dramatically changed before and after 1982.
Could you give me some thoughts for the reason why this happened?
Thank you!
The NCHILD variable is created based on the information contained in the RELATE variable. The information gathered via the RELATE variable has changed slightly over the years, see the comparability tab. Specifically in explanation for your question, in Basic Monthly samples from January 1976 - December 1981 persons younger than 14 years were not included in the original CPS data (except for in October samples from 1976 and 1977 which have an education supplement). In Basic Monthly samples from January 1982 - December 1988, persons younger than 14 years of age are included in the original CPS data, however there is no “child” code for these samples; children are coded as “Other relative.” In Basic Monthly samples from January 1989 - current samples all persons are part of the universe for the RELATE variable.
