Chapter 17 Further Reading: Poll Aggregation

Foundational Works

Silver, Nate. The Signal and the Noise: Why So Many Predictions Fail — But Some Don't. Penguin Press, 2012. The book that introduced sophisticated poll aggregation to a mass audience. Silver explains the logic of combining imperfect signals, the challenge of distinguishing signal from noise, and the probabilistic mindset necessary for forecasting. The chapters on political forecasting remain the clearest non-technical introduction to the field.

Erikson, Robert S., and Christopher Wlezien. The Timeline of Presidential Elections: How Campaigns Do (and Do Not) Matter. University of Chicago Press, 2012. A rigorous examination of how polling data evolves over the course of a campaign, providing the academic foundation for why recency weighting makes sense: early polls carry real information, but their signal about the final outcome is weaker than polls taken close to Election Day.

Gelman, Andrew, and Gary King. "Why Are American Presidential Election Campaign Polls So Variable When Votes Are So Predictable?" British Journal of Political Science 23 (1993): 409-451. A classic paper demonstrating that polls fluctuate far more than final vote shares do — evidence that much of what drives polling variation is noise, not signal. This paper provides important context for why aggregation is necessary and why individual polls should be treated with skepticism.

On Pollster Quality and House Effects

Enten, Harry. "How FiveThirtyEight Calculates Pollster Ratings." FiveThirtyEight, 2014. A detailed explanation of 538's pollster rating methodology, including how they measure historical accuracy, compute simple average error, and assess transparency. Essential reading for understanding quality weighting.

Shirani-Mehr, Houshmand, et al. "Disentangling Bias and Variance in Election Polls." Journal of the American Statistical Association 113 (2018): 607-614. Academic decomposition of polling error into its bias and variance components across hundreds of election polls, finding that bias (systematic error) is often larger than variance (random error). Directly relevant to the limits of aggregation discussed in this chapter.

On Herding

Jennings, Will, and Christopher Wlezien. "Election Polling Errors Across Time and Space." Nature Human Behaviour 2 (2018): 276-283. A comparative analysis of polling errors across multiple countries and election cycles, examining patterns of systematic bias and providing evidence for the herding hypothesis in some contexts.

On Aggregator Comparisons

Pasek, Josh. "When Will Nonprobability Surveys Mirror Probability Surveys? Considering Types of Sample Representativeness and Weighting Strategies." Political Communication 33 (2016): 399-424. Examines how online opt-in polls compare to probability-based telephone polls, relevant to how aggregators should weight the proliferating universe of non-probability surveys.

Aggregator Resources

FiveThirtyEight Pollster Ratings Database (fivethirtyeight.com/interactives/elections-forecast/) The full database of pollster ratings, updated after each election cycle, with methodology documentation. Essential for understanding how quality weighting is operationalized.

RealClearPolitics Polling Averages (realclearpolitics.com/epolls) The most-used simple averaging aggregator, with complete poll lists for all major races. Useful for comparison with weighted models.

Ballotpedia Polling Archive A comprehensive archive of election polls going back multiple cycles, useful for historical analysis of aggregation accuracy.

On the Influence of Aggregators

Westwood, Sean J., Solomon Messing, and Yphtach Lelkes. "Projecting Confidence: How the Probabilistic Horse Race Confuses and Demobilizes the Public." Journal of Politics 82 (2020): 1530-1544. An experimental study finding that probabilistic forecasts (e.g., "70% chance of winning") reduce voter turnout among the favored candidate's supporters compared to margin-based presentations (e.g., "leading by 5 points"). Directly relevant to the aggregator influence discussion.

Healy, Andrew, and Gabriel Lenz. "Presidential Voting and the Local Economy: Entitlement Programs, Credit Claiming, and Sociotropic Voting." American Journal of Political Science 61 (2017): 322-336. Examines media effects on political behavior, with implications for how aggregator coverage shapes campaign decisions.

International Comparisons

Sturgis, Patrick, et al. "Report of the Inquiry into the 2015 British General Election Opinion Polls." British Polling Council/Market Research Society, 2016. The official post-mortem of UK polling failures in 2015, with detailed analysis of herding, house effects, and the limits of aggregation in a different electoral context.

Jennings, Will. "Polling Accuracy in Real Elections: Assessing and Understanding British Polling Errors." British Journal of Political Science 48 (2018): 937-960. Comparative analysis of aggregation approaches in British elections, providing useful perspective on what works differently across electoral systems.

Blogs and Ongoing Commentary

Andrew Gelman's Statistical Modeling, Causal Inference, and Social Science blog (statmodeling.stat.columbia.edu) Gelman and collaborators regularly examine polling methodology, aggregation, and forecasting, often with direct criticism of widely-used practices. Technically demanding but extremely rigorous.

Patrick Ruffini's EchoChamber.com and related polling commentary Provides practitioner perspective on poll methodology from the Republican polling side, useful for understanding how internal campaign polling relates to public aggregations.