Physicist: In a divided society, polls make election results random

In a society divided into two camps, public opinion polls and the media influence voting behavior. In such a situation, the competition between candidates becomes equal, and their support can approach 50 percent, assessed Prof. Piotr Szymczak, a physicist from the University of Warsaw, who has studied the mechanisms behind voter behavior.
The closer the support of both candidates approaches 50 percent, the greater the influence of chance on the final result, and the result itself ceases to be representative, assessed in an interview with PAP. Prof. Piotr Szymczak from the Faculty of Physics at the University of Warsaw. In his opinion, this phenomenon has negative consequences for trust in democracy.
On June 1, Karol Nawrocki won the second round of the presidential election, defeating Rafał Trzaskowski, with a difference of 369,591 votes (1.78 percentage points). This is the smallest difference in the history of the Third Polish Republic in terms of the number of votes cast for candidates in the second round of the presidential election. In 2020, Andrzej Duda won against Rafał Trzaskowski by a margin of 422,385 votes (2.06 percentage points). And in 2015, Andrzej Duda gained 518,316 votes more than Bronisław Komorowski (it was a margin of 51.55 percent to 48.45 percent).
"If the election result in a 40-million-strong society is decided by 200,000 votes, then we can assume that the outcome is largely random - not much more representative than a coin toss. And we probably don't want random fluctuations to decide the president's election. This is not a favorable situation for democracy," assessed Prof. Szymczak. He added that the Polish elections are just one example in which support for both competing candidates was almost equal.
In a paper published in 2024 in the journal Physical Review E, Olivier Devauchelle, Piotr Szymczak, and Piotr Nowakowski decided to examine the mechanisms that cause support for opposing options in some elections or referenda to approach 50 percent. The article was given the telling title: "Aversion to the general opinion leads to equal election results."
"This is a surprising phenomenon - how do members of a huge society know exactly how to divide themselves in half? For this to happen, they must receive feedback on what the dominant tendencies are among the remaining voters. This role is played by pre-election polls," interpreted Prof. Szymczak. The scientist also noted that the media devotes a lot of attention to polls, the election fight and analysis of vote flow before the elections. "I have the impression that even more than the political content itself," commented the physicist.
And he gave an illustrative example of how polls affect voter behavior. "If I know from a poll that my candidate has 70 percent support, I can calmly go mushroom picking instead of voting. But if I know that he is 2 percent short of winning, then I will not only give up mushroom picking, but I will also call my friends to mobilize them," explained Prof. Szymczak. He added that such a mechanism promotes increasingly fierce electoral competition and intensifies social polarization.
The physicist compared a simplified model of an electoral situation in a polarized society to a game of Go, in which the stones are colored black or white, symbolizing the choice of one of two candidates.
"The probability that a given stone will turn white increases if its surroundings - family, friends - are also white. This is an example of conformist tendencies in a group. Such a mechanism does not, however, lead to a 50/50 balance by itself. Another factor is needed: a tendency to nonconformism towards the majority. If the stones 'know' what the general proportion of whites and blacks is, they can change sides and vote against the dominant trend. In this way, support is equalized," the researcher described.
This mechanism promotes increasingly fierce electoral competition and deepens social polarization. In their work, the researchers showed that this leads to the emergence of the so-called divided society phase: "Two political camps of similar strength stand opposite each other, with few connections between them. To change the outcome, it is enough for a small number of undecided voters to change sides, while the majority sticks to their position."
Prof. Szymczak said that the inspiration for the research was the 2015 elections in Poland, when Bronisław Komorowski initially had a large advantage, which gradually melted away until he lost to Andrzej Duda. "We talked about it with friends over a beer and came up with the idea that maybe it could be described with a physical model," he recalled.
Asked if he saw a way to break away from the model of 50/50 support, he replied that a total ban on polls may be difficult to implement, but one can try to limit their impact. He mentioned that in Poland, the election silence lasts two days, while in Italy there is a ban on publishing poll results for up to two weeks before the election. "It's an idea. And we, the citizens, always have one more option - to cheat the pollsters who conduct the polls," he joked. He noted that if the polls cease to be reliable, their impact will automatically decrease.
In conclusion, he emphasized that in order to get out of this logic of polarized politics, it is necessary to overcome divisions. "The role of politicians should be to connect society and build bridges - not only attack opponents, but also to notice those points in which the other side is right. And on this foundation - building a community in which people have a common goal and are not opponents of each other" - summed up Prof. Piotr Szymczak.
Science in Poland, Ludwik Tomal
lt/ agt/ bar/ lm/
naukawpolsce.pl