Problems of Response in Enumerative Surveys

D ATA such as age, sex, occupation, acres of cropland, number of cattle, and similar information collected by the Bureau of the Census are sometimes referred to as data. These are frequently contrasted with another class of items known as attitudinal data. It is generally supposed that data can be collected with relatively little difficulty, and that response problems in enumerative surveys are minimal. Aside from the fact that the line between factual and attitudinal data is somewhat nebulous, the studies we have made indicate that response errors in enumerative surveys are not always negligible. There is some indication, for example, of an under-count of population in some enumerative surveys that may amount to as much as two or three per cent. In a factual item such as age it is likely that inaccuracies are minor, but it is common knowledge that too many ages are reported in figures ending in o or 5. The folklore of our culture is rich in jokes about the reluctance of women to tell their age, and the forgetting of a few years is not considered to be particularly bad taste. Actually, many men are also subject to similar memory lapses. Aside from people's reluctance to tell their exact age, other factors affect the data that are collected. There may be confusion as to whether one should give age as of last birthday, or age to the nearest birthday. Also, a moderately large number of persons simply don't know how old they are-there are no birth certificates on file for millions of persons, and some of these persons never knew the exact year in which they were born. The errors which we observe in age statistics do not appear to be uniformly distributed-they seem to cluster around ages