A Check on Underenumeration in the 1940 Census

U NDERENUMERATION of underregistration is the bane of every census and of every Bureau of Vital Statistics the world over. Possibly the main difference in this connection between a good census set-up and a poor one is that the first will attempt some estimate of the proportion underenumerated while the latter will go blithely on in ignorance of the fact that it is missing some of the population. Two checks are possible: a sample check or a mass enumeration which would be both compulsory and subject to penalties. Spot samples, to anticipate our data by a glance forward at Tables i and 2, can vary so much from area to area that a mass census is indicated as the only valid check on enumeration. For the first time in history we are provided by the First Selective Service Registration with a check that fulfills the requirements, for it was compulsory on all males between 2i and 35 in I940 and was backed by adequate penalties. Other difficulties were present, however, and should not be underestimated as the following account will show. After estimating the amount of underenumeration for all classes and then for Negroes separately an attempt is made to evaluate the factors associated with the Census errors. The First Selective Service Registration was on October i6, I940, just six and one half months after the date of the Sixteenth Census. This registration stated that: