We are indebted to Mr Peter Seymour of Hertfordshire for bringing our attention to an article published in CNN news by Harry Enten (How Trump has broken the polls 4 May 2020)*
Enten draws an important distinction between what people want to happen, and what they believe will happen. Example: when you tell a pollster “I will vote for Biden”, you want Biden to win. When you tell the same pollster in the next question “I think Trump will win”, that is your belief about what will happen. Unless you have changed your mind.
Enten points out that between 1988 and 2012, voters’ predictions of who they believed would win were a more powerful predictor of electoral success than voters’ declared intentions (“I will vote for Dole,Romney, Dukakis…etc”) However he points out that in recent times this pattern has broken. In 2018, voters called the US House race for the Republicans, but the Democrats won it-as the opinion polls had been correctly predicting.
So what is happening? We at LSS have always been fascinated by the gap between when people say what they think you want to hear, and their true intentions. Especially when these intentions are not certain, and they may be havering between choices that they would really rather not make. As in quantum mechanics, the very act of measuring the system changes it. This may explain odd election results such as that of Britain in 1992, Readers may recall that after large poll leads, Labour was soundly beaten, a fact attributed to “shy Tories“, who told the pollsters one thing, then did another.
Yet we believe this explanation is at once too facile, and misses an opportunity to delve deeper into a fascinating subject-Games Theory. The observation that as one member of a group changes its behaviour, other members react, in turn change theirs is well known. A classic example is the game of Poker, played by some very intelligent people, where the change in tactics by one player will be mirrored by all the others around the table. Michael Karnjanaprakorn posts an interesting guide to poker decision pathways in his beginner’s guide to game theory, which we have linked below.*
Games theory attempts to put mathematical models onto complex systems where the actions of every player modify the actions of all the others. The pioneering work was started by John van Neuman and Oskar Morgenstern in 1944. It is widely applied in biology, information and economics. In our present context it would attempt to answer the fiendish intricacies of a situation where voters declare for Biden, making Trump supporters try harder, leading Biden voters to fear they have caused a Trump win, so they react by…… etc etc ad nauseam)
Before we go mad, or lose ourselves in a whole day in games theory (see excellent Wiki article below) it is worth remembering two things. Firstly, most voters’ attachments are deeply formed, and they will not change their minds. Most elections are decided by a relative handful of swing voters in a few constituencies (we think that’s Districts in the US). Secondly, all attempts to apply over elaborate theories to human behaviour have so far failed. It was always the Communist boast that Marxist analysis was so perfect that the victory of the proletariat was inevitable. So you had better get on the winning side, or look out. Obviously the more people who did that, the more chance there was that the Communists would win.
#enten #DonaldTrump #Biden #gamestheory #opinionpolls