An official website of the United States government
Abstract Presidential elections can be forecast using information from political and economic conditions, polls, and a statistical model of changes in public opinion over time. However, these “knowns” about how to make a good presidential election forecast come with many unknowns due to the challenges of evaluating forecast calibration and communication. We highlight how incentives may shape forecasts, and particularly forecast uncertainty, in light of calibration challenges. We illustrate these challenges in creating, communicating, and evaluating election predictions, using the Economist and Fivethirtyeight forecasts of the 2020 election as examples, and offer recommendations for forecasters and scholars.
more »
« less
Polls, Context, and Time: A Dynamic Hierarchical Bayesian Forecasting Model for US Senate Elections
Abstract We present a hierarchical Dirichlet regression model with Gaussian process priors that enables accurate and well-calibrated forecasts for U.S. Senate elections at varying time horizons. This Bayesian model provides a balance between predictions based on time-dependent opinion polls and those made based on fundamentals. It also provides uncertainty estimates that arise naturally from historical data on elections and polls. Experiments show that our model is highly accurate and has a well calibrated coverage rate for vote share predictions at various forecasting horizons. We validate the model with a retrospective forecast of the 2018 cycle as well as a true out-of-sample forecast for 2020. We show that our approach achieves state-of-the art accuracy and coverage despite relying on few covariates.
more »
« less
- Award ID(s):
- 1845434
- PAR ID:
- 10394195
- Date Published:
- Journal Name:
- Political Analysis
- Volume:
- 31
- Issue:
- 1
- ISSN:
- 1047-1987
- Page Range / eLocation ID:
- 113 to 133
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Abstract Accurate forecasts can enable more effective public health responses during seasonal influenza epidemics. For the 2021–22 and 2022–23 influenza seasons, 26 forecasting teams provided national and jurisdiction-specific probabilistic predictions of weekly confirmed influenza hospital admissions for one-to-four weeks ahead. Forecast skill is evaluated using the Weighted Interval Score (WIS), relative WIS, and coverage. Six out of 23 models outperform the baseline model across forecast weeks and locations in 2021–22 and 12 out of 18 models in 2022–23. Averaging across all forecast targets, the FluSight ensemble is the 2ndmost accurate model measured by WIS in 2021–22 and the 5thmost accurate in the 2022–23 season. Forecast skill and 95% coverage for the FluSight ensemble and most component models degrade over longer forecast horizons. In this work we demonstrate that while the FluSight ensemble was a robust predictor, even ensembles face challenges during periods of rapid change.more » « less
-
Abstract With historic misses in the 2016 and 2020 US Presidential elections, interest in measuring polling errors has increased. The most common method for measuring directional errors and non-sampling excess variability during a postmortem for an election is by assessing the difference between the poll result and election result for polls conducted within a few days of the day of the election. Analysing such polling error data is notoriously difficult with typical models being extremely sensitive to the time between the poll and the election. We leverage hidden Markov models traditionally used for election forecasting to flexibly capture time-varying preferences and treat the election result as a peek at the typically hidden Markovian process. Our results are much less sensitive to the choice of time window, avoid conflating shifting preferences with polling error, and are more interpretable despite a highly flexible model. We demonstrate these results with data on polls from the 2004 through 2020 US Presidential elections and 1992 through 2020 US Senate elections, concluding that previously reported estimates of bias in Presidential elections were too extreme by 10%, estimated bias in Senatorial elections was too extreme by 25%, and excess variability estimates were also too large.more » « less
-
Near-term, ecological forecasting with iterative model refitting and uncertainty partitioning has great promise for improving our understanding of ecological processes and the predictive skill of ecological models, but to date has been infrequently applied to predict biogeochemical fluxes. Bubble fluxes of methane (CH 4 ) from aquatic sediments to the atmosphere (ebullition) dominate freshwater greenhouse gas emissions, but it remains unknown how best to make robust near-term CH 4 ebullition predictions using models. Near-term forecasting workflows have the potential to address several current challenges in predicting CH 4 ebullition rates, including: development of models that can be applied across time horizons and ecosystems, identification of the timescales for which predictions can provide useful information, and quantification of uncertainty in predictions. To assess the capacity of near-term, iterative forecasting workflows to improve ebullition rate predictions, we developed and tested a near-term, iterative forecasting workflow of CH 4 ebullition rates in a small eutrophic reservoir throughout one open-water period. The workflow included the repeated updating of a CH 4 ebullition forecast model over time with newly-collected data via iterative model refitting. We compared the CH 4 forecasts from our workflow to both alternative forecasts generated without iterative model refitting and a persistence null model. Our forecasts with iterative model refitting estimated CH 4 ebullition rates up to 2 weeks into the future [RMSE at 1-week ahead = 0.53 and 0.48 log e (mg CH 4 m −2 d −1 ) at 2-week ahead horizons]. Forecasts with iterative model refitting outperformed forecasts without refitting and the persistence null model at both 1- and 2-week forecast horizons. Driver uncertainty and model process uncertainty contributed the most to total forecast uncertainty, suggesting that future workflow improvements should focus on improved mechanistic understanding of CH 4 models and drivers. Altogether, our study suggests that iterative forecasting improves week-to-week CH 4 ebullition predictions, provides insight into predictability of ebullition rates into the future, and identifies which sources of uncertainty are the most important contributors to the total uncertainty in CH 4 ebullition predictions.more » « less
-
Abstract Traditionally, election polls have asked for participants’ own voting intentions. In four elections, we previously found that we could improve predictions by asking participants how they thought their social circles would vote. A potential concern is that the social-circle question might predict results less well in elections with larger numbers of political options because it becomes harder to accurately track how social contacts plan to vote. However, we now find that the social-circle question performs better than the own-intention question in predicting two elections with many political parties: The Netherlands’ 2017 general election and the Swedish 2018 general election.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Journal Article:
- https://doi.org/10.1017/pan.2021.42
