Alana’s reference to the Pew Research poll that used the membership of the American Association for the Advancement of Science as a proxy for American scientists generally, which it is not, reminds me of perhaps the most famous poll in American History, the Literary Digest poll for the presidential election of 1936.
The Literary Digest, a middlebrow magazine perhaps roughly analogous to the Atlantic or Harper’s today, polled a huge number of people — 2,500,000 all told — and came up with the prediction that Alf Landon, the Republican candidate, would decisively defeat FDR. He didn’t. In fact he won only two states and eight electoral votes, tying the lowest total of any major party candidate since modern party politics began around the time of the Civil War.
What went wrong? As with the Pew Research poll, the problem was the pool from which the sample was drawn. The Digest used its own subscription base, people with registered automobiles, and those with telephones. In the Depression, that skewed the sample way up the socioeconomic scale. (It would not be until 1946 that half of American households had a telephone.)
When the poll was first published, it had a gloss of believability about it. Republicans had done well in the Maine gubernatorial and congressional races (which were then held in September), and Maine had a reputation a being a bellwether state. “As Maine goes,” went the political proverb, “so goes the nation.”
After the election, Democrats gleefully changed the saying to “As Maine goes, so goes Vermont.” And the Literary Digest soon folded.