Everyday History

When it comes to politics, as well as lots of other things, people are always trying to figure out what’s going to happen before it does.  Because gypsies are less than reliable when it comes to divining the future, we’ve come up with polls.  The first poll used in a presidential election was in 1824, when The Harrisburg Pennsylvanian, along with other newspapers, randomly asked people who they thought was going to win in the upcoming election, Andrew Jackson or John Quincy Adams, among others.  The straw poll saw Jackson easily beating Adams 2 to 1.  Adams ended up winning the election... well, he became the president.  However, nobody won a majority of the electoral votes, which meant the House of Representative got to choose, and they chose Adams even though Jackson had more electoral votes and popular votes.  So you tell me:  How accurate was that poll?

It was on September 16, 1916, that The Literary Digest came up with the first of their many presidential polls.  They mailed millions of postcards to people from lists made up from telephone directories and car registrations.  The people filled them out and sent them back.  In 1916 the poll was pretty spot on.  Woody Wilson was the narrow winner.  And their polls were pretty accurate until 1936, when they completely blew it.  In 1916 through 1928 there wasn’t a Depression.  In 1936 there was.  Sure, in 1932 the Depression was going strong.  But everybody blamed Hoover, so it was no surprise that he lost to Roosevelt.  In 1936 those postcards were sent from the same lists as before, only now the only people who had cars or telephones were the wealthy, who heavily favoured Alf Landon.  The poor folk, who far outnumbered the wealthy then, as now, heavily favoured Roosevelt, who won.  Whoops.

And that led to a guy named George Gallup who came up with a scientific method for doing polls, which were first used on October 20, 1935.  Gallup used demographics.  For instance, if 25% of all voters are white male with a high school education or less, then you make sure out of those people you poll only 25% are white males with a high school education.  It can get a whole lot more technical than that, but you get the general idea.  If a poll such is that is accurate, you only need to poll around a thousand people to get a pretty good idea of everybody.  And those polls worked fine, pretty much until recently.

It’s become increasingly difficult to accurately poll people, more than likely because of cell phones.  There no longer is a directory that you can pick up and use to randomly call people.  And if people are called, there’s a good chance they’ll simply hang up, or not take the call to begin with, for any number of reasons.  And then, what’s the point of taking a poll if nobody is going to believe it?  With all that in mind, Gallup stopped doing presidential polls altogether, not because they were no longer accurate, they claimed, but so they could track other stuff, like long term economic trends.  Perhaps the gypsy is more reliable after all.

​