The Unemployment Report Card
1937 · Double Reply Postcard · Work 704 of 1,871
The sorting table of a small-town post office in the United States in the early morning of a November day in 1937, on the worn wooden table rows of tall stacks of oversized folded double cards of pale card stock tied into bundles with white cotton string and seen only from their edges, beside them a mail carrier's brown leather shoulder satchel packed full with more bundles, behind the table a wooden pigeonhole sorting case with dozens of open compartments for the delivery routes, a ball of string on the table, warm light from enamel pendant lamps, dark windows before dawn, plank floor, the mail carrier seen from behind in a dark blue-grey wool coat and peaked cap.
- What it was used for
- This was a double postcard with fourteen questions that the United States Post Office delivered to every household on 16 November 1937. In August 1937 Congress had authorised a nationwide voluntary registration of the unemployed. Franklin D. Roosevelt explained it in a radio address on 14 November 1937: the postman would leave the card at the door, and it was to be returned without stamp or envelope by midnight on 20 November. Only the unemployed and partly employed who were able to work and seeking work were to fill it in. Roosevelt expressly called the exercise "a wholly voluntary census" and "purely and simply a fact-seeking census", that is, neither an application for relief nor registration for a job.
- What was new
- It was not the postcard that was new but its check. A voluntary self-registration cannot be read as a rate, because nobody knows who failed to answer. In David Card's account the Census Bureau's statisticians saw in this "serious problems in interpreting the results". So the same postmen were sent out a second time, now as interviewers for a sample, the Enumerative Check Census. For it the roughly 92,000 postal routes were grouped into 9,596 blocks of 50 routes each, and block number 42 was selected. Card calls the result the Census Bureau's first scientifically constructed national sample, the first survey using the modern labour-force concept and the Bureau's first publication with confidence intervals: in January 1938, an unemployment rate of 20.2 per cent, plus or minus 0.9. The control sample became the basis of unemployment statistics.
What the sources leave open: Card attributes the sample design to no individual but generally to statisticians at the Census Bureau, naming Calvert Dedrick and Morris Hansen as authors of the final report. He does not say how many cards were returned. The word postcard does not appear in his paper, but it does in Roosevelt's address.