Search This Blog

Thursday, June 8, 2017

Creative History at 538

In 2012, the guru of political prediction was Nate Silver, then of the New York Times.  He had models and poll readjustments with which he was able to predict outcomes.  Silver became such a celebrity that he was able to start his own site called 538 to make predictions about political and sporting events.  Then came the 2016 election.  Prior to the closing of the polls, the 538 prediction was that Hillary was over 80% likely to win.  We all know what happened after that.  So now that the UK is voting today, it's interesting to see the predictions made for that election by 538.

Today's predictions start with a retelling by Silver of what happened last November in the US elections.  Strangely, he claims that prior to that vote he said that one of three outcomes was likely:  Hillary would win big, Hillary would win in a close election, or Trump would win in a close election.  Silver even claims that these three outcomes were equally likely.  It's a nice defense to his missing the results of the US election, but it's just not true.  You cannot give Donald Trump less than a 20% chance of winning and then claim that you really gave him a one-third chance to win.  Actually, I guess you can make such a claim; you just can't say that truthfully.

Silver is taking no chances with the UK vote embarrassing him again.  He now says that the Tories will 1) win big; 2) win by a small majority or 3) lose the majority.  After reading Silver's analysis, one is left with the main question:  why did I bother to read this nonsense.  Of course the Tories will either win or not.  That is, by definition, 100% likely.

For decades, we have been given constant polling results in election after election.  The polls are usually correct, but they are often wrong.  Now, Silver at 538 is trying something new; he is predicting every possible outcome and announcing that one of them will happen.

No comments: