NEWS PUBLICATION 29-OCT-2020
The electoral model treats political influence like contagion
(Click Show Link to watch the video. ~ Cr)
VIDEO: To visualize the uncertain nature of the election predictions, this video shows a random sample of 500 simulated elections by researchers. PAUSE THE VIDEO AT ANY TIME TO WATCH A… More
CREDIT: NORTHWEST UNIVERSITY
- The new model treats determined voters as “infected” and indecisive voters as “vulnerable” to infection
- Democratic and Republican "diseases" spread through a population and "infect" indecisive voters
- The model introduces the possibility of asymmetrical relationships or influences between states
- As of today (October 29), the model predicts a win for Biden 89.03% of the time
EVANSTON, IL – A new approach to election forecasting uses mathematical models to describe how voters in different states can influence one another during an election year.
To simulate how voter interactions can play a role in the upcoming presidential, gubernatorial and Senate elections, a research team at Northwestern University is adapting a model commonly used to study infectious diseases.
The model treats determined voters as “infected” and indecisive voters as “vulnerable” to infection. Two “diseases” (Democratic and Republican voting tendencies) spread through a population and “infect” (or influence) indecisive individuals.
"Experts like the FiveThirtyEight team explain the fact that if you incorrectly specify how Pennsylvania will vote, you may also incorrectly specify how Ohio will vote because these states share similar characteristics," said Northwestern's Alexandria Volkening, who the Conducts research. “Such symmetrical relations between states are important. Using a disease transmission model, we also introduce the possibility of asymmetrical relationships or influences. For example, a candidate fighting in Florida might be featured on the Ohio news and influence voters there. "
The research was published online in SIAM Review yesterday. Viewers can follow the forecast for 2020 here.
Volkening is an NSF-Simons Fellow at the NSF-Simons Center for Quantitative Biology in the Northwest and at the Department of Engineering Sciences and Applied Mathematics at the McCormick School of Engineering. The newspaper is coauthored by Daniel Linder from Augusta University, Mason Porter from UCLA, and Grzegorz Rempala from Ohio State University. Their 2020 predictions are based on Volkening's students (Samuel Chian, William He, and Christopher Lee) studying at the McCormick School of Engineering.
The project began when Volkening and her co-authors aimed to better understand the election predictions.
"My background isn't in election forecasting," said Volkening, who often applies mathematics to biological questions. “But I'm interested in problems in complex systems where individuals come together to create group dynamics. Mathematical models can be used to describe cell behavior in developmental applications and voter interactions prior to elections. "
Volkening and her team wanted to use a data-driven mathematical modeling approach. They decided to adapt what is known as a susceptible-infected-susceptible compartment model that is typically used to study the spread of diseases such as the flu.
By adapting this model to two “diseases” (Democratic and Republican voter propensity), the researchers simulated how determined voters can influence undecided voters. For example, a Republican voter who speaks to an undecided voter can trick him into becoming a Republican. In another scenario, former Vice President Joe Biden could attend a campaign event that affects undecided voters.
"In the future, we may be able to find out how states influence each other and identify more influential states," said Volkening. "We want to study how interactions between states change over time."
To create each of their 2020 predictions, the researchers use survey data from FiveThirtyEight to simulate 10,000 potential election results. At the time of this writing, the model predicts victory for Biden 89.03% of the time and President Donald Trump 10.78% of the time.
"It was exciting to have the model running continuously over time," said he, a sophomore student studying applied math and statistics. "We don't just have a single forecast. We update our website regularly so we can see how opinions change."
While 89% may sound like Biden has a high chance of winning the election, Volkening is quick to point out that turnout and undecided voters could change that.
"In many states, the profit margin we forecast for Biden is lower than the percentage of undecided voters," she said. "If undecided voters stand up for Trump, we could certainly see a Republican result."
The paper “Predicting Elections Using Compartmental Models for Infection” was supported by the Mathematical Biosciences Institute, the National Science Foundation (grant numbers DMS-1440386, DMS-1853587 and DMS-1764421) and the Simons Foundation (grant number 597491-RWC). The students' research was supported by the Northwestern Office of Undergraduate Research and NSF DMS-1547394.
More news at Northwestern Now
Find experts in our Faculty Experts Hub
Follow @NUSources for expert perspectives