Census Completeness and Accuracy

[rp76226]

The 1950 Census has been published. However, can't find very many relatives on it including myself and my parents. Makes you wonder how they can ever come up with an accurate census. For example, people not home when the census taker knocks on the door. Not every door can physically be knocked on, so they send you census letters to fill out. How many actually fill them out and mail them in? Not everyone for sure. Those that do - how do you find them on any census publication since it appears that only the one form - the one reporting the results of streets where census takers knocked on doors and took down the information on a specific form common to all other census results shown?

Ron

[greglam]

Great questions. I have no answers.

I found my paternal grandparents on the 1950 Census, and I can tell you that my grandfather's age and place of birth were both incorrect. His error? The census taker's? Again, no answer.

-Greg

[darkerhorse]

I'm not sure of any particular census, but when the census was done by census takers, as in 1950, the Bureau made follow-up visits and inquiries to reach people who were missed by census takers. Of course, most responses are now by mail-in or by computer, but there are still follow-up efforts.

The Bureau publishes estimates of census coverage for geographies and population groups which tend to be very high, but that wouldn't apply to any particular individual.

Of course, census takers are human and make errors, intentional and accidental. I'm sure respondents also did. Language and literacy are also barriers.

At times, the Bureau also imputes or allocates some missing answers based on other answers by the respondent or the neighborhood.

My favorite anecdote is when, in the rural, poor South, a census taker, in the interest of complete coverage, made his way up a mountain to a small shack in the woods to reach a lone household. The woman answered the door and asked what he wanted. The census taker said he was here to take the census. She asked what is that? He said he was visiting homes to find out how many people lived in the United States. She responded, you'd better move on to the next house then, because I have no idea!

[darkerhorse]

I forgot to mention, the Bureau also reports estimates of accuracy, and makes some corrections, but in aggregate data, not in records for individuals.

For example, there's always age heaping. People and census takers tend to report ages ending in 0 and 5, say 50 and 55 years old rather than actual ages of 49 or 54. They tried asking year of birth in addition to age. In aggregate data, the Bureau may smooth the data to correct for age heaping - but not change individual records.

Keep in mind, the purpose of the census is to report aggregate data, not individual data.

[darkerhorse]

To be clear, when the Bureau imputes or allocates missing data or corrects data, it's only a statistical exercise to improve the aggregate data they report, like totals, means, and percentages.

They don't change the information on the individual census records. What we view are (unaltered) originals.

The only "changes" are made in cyberspace where the magic happens.

To gauge accuracy of individual records I would look at other related records for comparison, and triangulate.