Monday, May 1, 2017

The Evil President, Round 2

The French presidential election is drawing to a close, and it is down to Marine Le Pen, a right wing nationalist, vs. Emmanuel Macron, a more centrist candidate.  There are a lot of people looking at this thinking that it is a rematch of Trump vs. Hillary, business as usual vs. blame the foreigners, xenophobia vs. consensus building.

In some ways it is, of course, but the polls are in a completely different place.  Macron was ahead by 26 points just a short time ago, and his lead currently is 20 points.  People are worried that Le Pen will do what Trump did and sneak in a victory despite polling behind.  News sources are of course spinning that chance as hard as they can because it is far more interesting to say that the unlikely is possible than that the likely is, in fact, likely.

I can see why some people would be worried, but it is almost entirely an illusion.  The experts mostly said that Trump was going to lose.  There were reputable sources putting his chances at 1%.  But he was only 2% behind in the polls.  48-46 in polls is not that far behind, and in fact he ended up losing by 48-46 but won the election anyway because of the complexity of US voting systems.

However, Le Pen is not losing 48-46.  She is losing 60-40.  The difference between those things is outrageous.  Everyone who looks at polls knows that they are going to miss by a percent or two.  That is inevitable.  But missing by 10 points is ridiculous.

The trouble is that people don't actually understand numbers and the math behind polls and predictions.  They listen to the 'experts' on TV or in print and hear that Trump is rated to lose and Le Pen is rated to lose and think that those are equivalent when in fact they aren't even in the same ballpark.

There are so many problems that lead to this false equivalence.  For one, polling and odds making groups aren't perfect.  They ask the questions that will give them the answers they want, shape their methods to tilt things their way, and hunt for results that confirm their biases.  Odds makers do similar things, for similar reasons.  They are by far the most reliable part of the chain though, because the news sources that report information and analysis are horribly biased towards a good story even if it has nothing to do with the truth.

Just as it is hard to get a scientific paper published if you say "We learned that an obscure technique doesn't actually work" and easy if you say "We have developed a new way to cure cancer!" it is tough to get into the public eye by saying "Too close to call" when you can come up with "The underdog has closed the gap and threatens to win it all!"  The supposed experts spend a lot of time twisting the numbers to suit their purposes and that makes it really hard for the random person to know what the real deal is.

That random person is a problem too though.  They don't understand the math well enough to recognize that 60-40 and 48-46 aren't even the same ballpark.  They aren't even going to understand the math, so they have to rely on experts, and since all the experts are just saying what they think will get views or which supports their biases the average person has no idea what is going on.  We are stuck in a place where people are terrified of things that aren't going to happen for reasons that make no sense.

Now of course either candidate could, in the next week before the election, declare a longstanding habit of having sex with dogs, or perhaps an intent to use nukes on Canada.  That one of them completely botches it and the odds shift is possible.  But when the polling is as horribly one sided as it is right now (thankfully in the right direction) it would take something really massive and unexpected to change the result.  Could it be that Le Pen ends up with 42% of the vote?  Sure, that is well within the margin of error.  51%?  Hah!

I would love it if I could educate random voters on these topics, but I suppose if I am hoping for impossible education I should probably start by teaching them about how voting for people like Trump and Le Pen is contrary to their desires and interests, rather than educating them about how to properly figure out the odds that such a leader will end up running the show.

4 comments:

  1. Eric Grenier made a very interesting comment about Trump's chances during the US election. He said that while people had consistently counted Trump out in the primaries and been proven wrong, they were counting Trump out by using rationalizations to justify ignoring the polls. The polls always said Trump was in front, people just didn't think he'd hold on, or thought that the other candidates would drop out and Trump would lose one on one.

    Nate Silver, who admitted he had wrongly counted Trump out in the primaries, had Trump at 14% to win the day before the election. He said that the majority of scenarios that had Trump winning he would win while losing the popular vote. People say, "The pollsters got it wrong" but the most famous poll analyst in America didn't get it wrong; people just don't know the difference between 14% and 0%.

    I think people are scared of Le Pen because Le Pen seemed much better poised to win last fall than she does now. I have to wonder if the French are looking at America and doing some nervous collar tugging at the thought of electing their own Trump.

    ReplyDelete
  2. This in no way changes your point but as a correction 538 (on the day of the election) had trump at 29%

    https://www.nytimes.com/interactive/2016/upshot/presidential-polls-forecast.html?_r=0#other-forecasts

    ReplyDelete
  3. The polling does not look as certain to me. The headline on the polling says 60-40... but the details show Macron 42% - Le Pen 28% - Neither 27% - Undecided 3%. A week before the vote, more that a quarter of people are saying "Neither".

    ReplyDelete
  4. Yeah, there were a lot of 'experts' who kept counting out Trump for reasons that had nothing to do with math and polls, and everything to do with 'common sense' and guesses. The polls and the good analysts weren't far off at all, as someone with a 29% win chance is going to win about, oh, 29% of the time.

    ReplyDelete