Election Forecaster Explains Methodology for Predicting Outcomes
Translated & summarized from JDN by baba
The "Barometer" election forecasting model uses a multi-stage process to analyze polls, weighting them by past accuracy and correcting for biases. It incorporates demographic and geographic models to predict seat allocations, projecting a drop for the Likud party to 25 seats and a potential shift of votes within the coalition bloc. The model excludes pre-2019 data due to methodological changes in polling. It forecasts Yisrael Beiteinu to secure 6 seats and analyzes the impact of vote wastage and turnout, especially in the Arab sector, on the final bloc size.
The story in 6 lines · by baba
- The "Barometer" election model filters polls by past accuracy and uses demographic/geographic data for predictions.
- Likud is projected to fall from 32 seats in 2022 to 25 in the upcoming election.
- Vote wastage on the right, with parties like Am Israel failing to cross the threshold, impacts the coalition bloc's size.
- The model excludes election data prior to 2019 due to changes in polling methodologies.
- Yisrael Beiteinu is forecast to receive 6 seats, above the electoral threshold.
- Voter turnout, particularly in the Arab sector, significantly influences the coalition bloc's final seat count.
The "Barometer" election forecasting model employs a rigorous three-stage professional filtering process to analyze and weigh all published polls, aiming to mitigate political bias and improve accuracy. The model assigns weight to polling institutes based on their precision in the last three election cycles (2020, 2021, 2022), with Direct Polls and Next Data receiving the highest weighting. It also corrects for consistent biases observed in each institute's past predictions for specific parties and utilizes demographic and geographic models based on past election results, population growth, and settlement trends to establish floor and ceiling limits for coalition blocs, particularly for Haredi and Arab parties that are difficult to poll accurately. These models account for half the weight in estimating these parties' results.
The Barometer's final forecast predicts a significant drop for the Likud party from 32 seats in 2022 to 25, while Otzma Yehudit is projected to run alone and receive 9 seats, and Shas is expected to decrease from 11 to 10 seats. This projection is based on the model's analysis that the coalition bloc's overall share of the public has remained consistent since 2022, with the decline in Likud seats attributed to shifts within the bloc and votes being lost to parties failing to cross the electoral threshold. The model suggests that if Ofer Vinter's Am Israel party crosses the threshold, the coalition bloc could return to 60 seats.
The model's methodology deliberately excludes election data prior to 2019, citing changes in polling institute methodologies, team compositions, and sampling techniques over the past decade. It also notes that several current polling firms did not exist or were not active in 2015, and others that were active then are no longer publishing. The exclusion of older data is also justified by the fact that while 2015 is remembered as a significant miss for the Likud, the overall bloc's performance was closer to predictions, with errors mainly in vote distribution within the bloc.
Regarding specific parties, the Barometer forecasts Yisrael Beiteinu to receive 5.1% of the vote and 6 seats, a position well above the electoral threshold, despite a historical downward trend. The model attributes the loss of votes for Yisrael Beiteinu since 2019 to specific cities with large populations of former Soviet Union immigrants, but acknowledges that tracking the exact destination of these votes requires further polling analysis. The forecast for Haredi parties, Shas and United Torah Judaism, is based on their surplus vote agreement, with Shas projected to receive 10 seats and UTJ 7, though a significant increase in voter turnout among Haredim could shift an additional mandate to UTJ, potentially bringing the coalition bloc to 60 seats.
The model's analysis indicates that vote wastage, particularly on the right, is a primary factor influencing the bloc's size. Three lists, Am Israel (Ofer Vinter), HaMeyuladnikim/HaKalkalit (Yoaz Hendel and Yaron Zelicha), and Noam, are projected to fall below the electoral threshold, collectively taking about 6.5% of the votes, with a significant portion coming from coalition voters. The model suggests that if Am Israel crosses the threshold, the bloc could reach 60 seats. The analysis also addresses the impact of voter turnout, particularly in the Arab sector, noting that higher turnout in this sector tends to reduce the coalition bloc's seat count, while lower turnout can increase it. The Barometer's election night page will provide real-time updates on turnout by sector compared to previous elections.