Can We, Should We Trust the Polls?

If I had a nickel for every poll which is published which says something different from the poll that was just published five minutes before this poll was published, I wouldn’t have to stop what my father called this ‘foolishness’ and go back to work tomorrow. But what are you going to do, right?

Anyway, last week the eagle-eyed lawyer sister sent me a long article out of one of the left-wing rags she reads, which attempts to explain why political polls are usually wrong. And basically, the degree to which pre-election polls don’t capture what then happens on Election Day is, according to the author, the result of polling having become a business and in the last few years, a big business as well.

And since in business the idea is to make yourself slightly different from the other, competitive products but keep your product within a well-defined group which shares certain basic features with every other similar product, once a few polls begin to say more or less the same thing, then a ‘herd instinct’ develops and all the polls tend to agree.

So, for example, once the polls started showing Hillary with a lead in 2016, then every poll started saying that she would win, even if the statistical probability which showed her in the lead was based on nothing more than a few points here or a few points there.