Campaign polls influence election results - new research

Popular budget coffee chain Cofix is holding their own poll of who will be the next prime minister using coffee cups, 2019. (photo credit: MARC ISRAEL SELLEM)
Popular budget coffee chain Cofix is holding their own poll of who will be the next prime minister using coffee cups, 2019.
(photo credit: MARC ISRAEL SELLEM)
A team of researchers from Ben-Gurion University of the Negev and Technion-Israel Institute of Technology attempted to answer some questions about voting with the help of artificial intelligence (AI).
Prof. Kobi Gal and master’s students Roy Fairstein and Adam Laoz of BGU, and the Technion’s Dr. Reshef Meir presented thousands of study participants with a range of polling data about fictional characters and then collected information on how the subjects decided to vote. The reward for participants was determined by the position of the winning candidate among their preferences. The participants had no knowledge of the voting preferences of other people in the group.
At the same time, the researchers built a theoretical model to allow for predicting how each participant would vote in a wide range of situations, after observing only a small number of votes. The model proved highly accurate, and gave insight into the considerations that guide people’s voting patterns, according to a statement from the researchers.
Gal said that most people tried to balance their personal preferences for a particular candidate and the candidates’ chance of winning the election, illustrating the “bandwagon effect,” a psychological phenomenon in which people do something primarily because other people are doing it, even to the extent that they may ignore or override their own personal beliefs or preferences.
The bandwagon effect is commonly seen in politics, Gal said.
Most striking, however, is that “When the respondents were asked to describe their decision-making process during actual election campaigns, they claimed not to have done this, in direct contrast to their behavior during the research project,” Gal told The Jerusalem Post. “It is obviously not conscious.”
He explained that often individuals assume that the “majority” has information or knows something that he or she does not, but which might be beneficial to know.
“If the majority of people eat at a particular restaurant, this is likely to affect whether I wish to go to that restaurant,” said Gal. “This is even if I have no additional information about why they are going there. It is because I assume they have some form of information about the quality or attractiveness of this restaurant that I am not aware of but maybe should be.”
Gal said that while it would be too early and too much of a leap to determine how polls being conducted between now and the April 9 Israeli elections will affect results, he said survey conductors should take into account that the way they present information could affect people’s decision-making.
Further, he said that while AI could play a role in determining election results in the near future, first, there are other challenges to address in this arena, such as why poll results are becoming increasingly inaccurate. On this, he said, “People are becoming more and more strategic about how they report who they vote for.”
“The reason is that as people become more partisan in their views, they see pollsters as their enemies rather than collaborators,” Gal continued. “Some people falsify who they vote for. It is also more difficult to get a representative sample of the population today than it used to be. Ironically, there is a lot more connectivity, but it is more difficult to understand what kind of population you are tapping.”
He said the government and media should invest in making people understand that “polls are here to help, not to sway and or affect results, but to provide people with a more accurate description of the heartbeat of the public with respect to important decisions.”
Then, he added, AI could play a greater role in shedding light on the way people make strategic decisions and help further determine the connection between election polls and actual voting.