NCHILD early 80s

CPS
1

I think you are probably aware but there are some irregularities in the NCHILD var before Jan 1982 that might need to be more clearly flagged in the data description. For example, there seems to a periodic undercounting in the months Jan-Apr and Jul-Aug (inclusive) relative to later years: see table below which is produced with basic monthly sample CPS from 1979 as follows:

gen day=1
gen date=mdy(month, day, year)
format date %tdMon,CCYY
gen haschild=(nchild>0)
table date [aw=wtfinl] c(mean haschild)

Resulting table:

date | mean(haschild)
----------±--------------
Jan,1979 | .1897951
Feb,1979 | .1889169
Mar,1979 | .1882834
Apr,1979 | .1889208
May,1979 | .3994564
Jun,1979 | .4001991
Jul,1979 | .1869851
Aug,1979 | .1858945
Sep,1979 | .3879481
Oct,1979 | .3920952
Nov,1979 | .3929478
Dec,1979 | .3951934
Jan,1980 | .181826
Feb,1980 | .1820925
Mar,1980 | .18112
Apr,1980 | .1813779
May,1980 | .3926153
Jun,1980 | .3936236
Jul,1980 | .1817007
Aug,1980 | .1798945
Sep,1980 | .3860932
Oct,1980 | .387099
Nov,1980 | .3875811
Dec,1980 | .1776806
Jan,1981 | .179806
Feb,1981 | .1790667
Mar,1981 | .1791891
Apr,1981 | .1799098
May,1981 | .3902053
Jun,1981 | .3888804
Jul,1981 | .1766069
Aug,1981 | .176092
Sep,1981 | .3851475
Oct,1981 | .3861556
Nov,1981 | .1762264
Dec,1981 | .3874279
Jan,1982 | .3881936
Feb,1982 | .3861274
Mar,1982 | .3858052
Apr,1982 | .3852535
May,1982 | .3854895
Jun,1982 | .3861552
Jul,1982 | .3874568
Aug,1982 | .3863076
Sep,1982 | .3836292
Oct,1982 | .3860345
Nov,1982 | .3855165
Dec,1982 | .3850844
Jan,1983 | .3854777
Feb,1983 | .3847058
Mar,1983 | .3827938
Apr,1983 | .3846395
May,1983 | .3847411
Jun,1983 | .3856003
Jul,1983 | .3860782
Aug,1983 | .385844
Sep,1983 | .3818506
Oct,1983 | .382155
Nov,1983 | .3811696
Dec,1983 | .382665

2

As noted in the basic monthly sample notes and ASEC sample notes, children under the age of 14 are not included in the ASEC prior to 1968 or basic monthly data prior to 1982–though they are in a few supplement-containing months in the 1976-1981 time period (this aligns with your observation that the undercounts are periodic rather than constant), such as the education supplement in October. Because this is a change at the sample-level, I don’t believe it is documented in many variable descriptions. I will pass along the recommendation to the IPUMS CPS team that they consider if there are variables (e.g., NCHILD) that should flag this change in sample composition more clearly. For more information you may be interested in this working paper on challenges creating linkages to leverage the panel component of the CPS, which discusses this omission of children as a comparability issue.