60. POLLS. Can we trust their forecasts?

Pollsters must get used to egg-on-face and eating-humble-pie. Why wrong so many times?

Elections are coming up in Australia, PNG, Libya, Kenya, Angola, Philippines, Lebanon, Colombia, Switzerland, Denmark, and New Zealand.

Elections are high season for pollsters. They prepare years ahead. Pollsters attempt to tell us the results in advance of actual polling day. Ahead of other pollsters — and ahead of other newspapers.

So how come they can get their forecasts so wrong? Not all the time, but a few dramatic disasters make the case.

1936. Four days before the US election The Literary Digest told its readers Alf Landon would defeat Franklin D Roosevelt. FDR won in a landslide, 523 electoral college votes to 8. Reason: TLD polled 10 million readers, but they asked only affluent TLD readers who were mostly Republicans with cars, telephones and high incomes.

1948. The Chicago Daily Tribune declared “Dewey Defeats Truman”! But he didn’t. The weekly New York Commerce Journal published a series of articles forecasting what wonderful things President Dewey would do for US businesses. He lost.

2019. Australians were told “the election is Labor’s to lose”. Pollsters predicted Shorten would defeat Morrison. In fact the result was Morrison’s Liberal/National coalition 77, Shorten’s Labor 68, Others 6. Egg on face time again for pollsters.

But one pollster, professor Bela Stantic of Griffith University, has correctly forecast three recent elections which went against the odds. Trump, Brexit, and Morrison. His technique involves analysing millions of social media comments posted by eligible voters.

By contrast, other pollsters still rely on robot calls to random mobile and fixed line telephones. (Notice the absence of human contact.)

If you really want to know who will win the next election, ask Bela Stantic.

#####BM

This is my SubStack Report #60.

Footnote: Cavewoman said to her caveman: “Now you’ve discovered fire, I suppose you’ll expect hot meals.”