Chapter 10 Further Reading: Reading and Evaluating Polls

Foundational Frameworks

American Association for Public Opinion Research. (2023). Transparency Initiative Standards. AAPOR. The authoritative source for ATI disclosure requirements. Available at transparencyinitiative.aapor.org. Includes a searchable database of ATI-certified polls with full methodology disclosures — the best single resource for practicing poll evaluation on real data.

National Council on Public Polls. (2019). 20 Questions a Journalist Should Ask About Poll Results. NCPP. Available free at ncpp.org. The NCPP framework is complementary to AAPOR standards, written for media consumers rather than researchers. Useful for understanding what a well-informed lay evaluator should demand from published polls.

Asher, H. (2016). Polling and the Public: What Every Citizen Should Know (9th ed.). CQ Press. The most accessible book-length treatment of polling for a general audience. Covers how polls work, how they can mislead, and how to be a critical consumer of survey-based political information. Excellent companion to this chapter.

Margin of Error and Statistical Interpretation

Gelman, A., & Unwin, A. (2013). "Infovis and statistical graphics: Different goals, different looks." Journal of Computational and Graphical Statistics, 22(1), 2–28. Broader context for understanding the visualization of statistical uncertainty, including confidence intervals. Valuable for understanding how to represent MOE in charts without creating the "within the MOE = tied" misreading.

Lau, R. R. (1994). "An analysis of the accuracy of 'trial heat' polls during the 1992 presidential election." Public Opinion Quarterly, 58(1), 2–20. Early systematic study of how polling errors decompose into sampling variability vs. systematic bias. Still relevant for understanding the relationship between MOE and accuracy.

House Effects and Pollster Bias

Silver, N. (2014). "The polls are all right." FiveThirtyEight. Accessible explanation of how FiveThirtyEight estimates and accounts for house effects in its polling averages. A good introduction to the practical methodology of professional poll aggregation.

Erikson, R. S., & Wlezien, C. (2012). The Timeline of Presidential Elections: How Campaigns Do (and Do Not) Matter. University of Chicago Press. Rigorous academic treatment of polling accuracy and house effects across presidential election cycles. Essential reading for understanding how house effects interact with campaign dynamics.

Shirani-Mehr, H., Rothschild, D., Goel, S., & Gelman, A. (2018). "Disentangling bias and variance in election polls." Journal of the American Statistical Association, 113(522), 607–614. The most important recent methodological paper on separating systematic bias from sampling variance in election polls. Demonstrates that total polling error is much larger than the sampling MOE suggests, and characterizes the contributions of different error sources.

Polling Averages and Aggregation

Pasek, J. (2015). "Predicting elections: Considering tools to pool the polls." Public Opinion Quarterly, 79(2), 594–619. Technical comparison of different averaging approaches: simple averages, weighted averages, trend-line methods, and Bayesian aggregation. Good for understanding the methodological choices behind professional aggregators' different approaches.

Lock, K., & Gelman, A. (2010). "Bayesian combination of state polls and election forecasts." Political Analysis, 18(3), 337–348. Foundational paper on Bayesian polling aggregation. More technical than most entries in this list, but essential for understanding how uncertainty is formally propagated in probabilistic forecasting.

Likely Voter Modeling

Erikson, R. S., Panagopoulos, C., & Wlezien, C. (2004). "Likely (and unlikely) voters and the assessment of campaign dynamics." Public Opinion Quarterly, 68(4), 588–601. Detailed analysis of how different likely voter screening models affect measured campaign dynamics. Shows that the choice of LV model can change not only the level of support but also the apparent trend over time.

Traugott, M. W. (2005). "The accuracy of the national preelection polls in the 2004 presidential election." Public Opinion Quarterly, 69(5), 642–654. Post-election accuracy assessment that compares outcomes across polls using different LV methodologies. Useful for understanding the empirical consequences of LV model choices.

Python and Data Analysis

McKinney, W. (2022). Python for Data Analysis (3rd ed.). O'Reilly Media. The definitive reference for pandas, written by its creator. Chapters on time-series data (relevant for poll trend analysis) and groupby operations (relevant for house effects analysis) are directly applicable to this chapter's Python work.

VanderPlas, J. (2016). Python Data Science Handbook. O'Reilly Media. Comprehensive introduction to the scientific Python stack (NumPy, pandas, matplotlib, scikit-learn). Available free online at jakevdp.github.io/PythonDataScienceHandbook. Chapters 3 (pandas) and 4 (matplotlib) are directly applicable.

Hunter, J. D. (2007). "Matplotlib: A 2D graphics environment." Computing in Science & Engineering, 9(3), 90–95. The original matplotlib paper. For understanding the library's design philosophy and capabilities beyond basic charting.

Online Polling Databases

FiveThirtyEight Polling Database (projects.fivethirtyeight.com/polls): Comprehensive archive of U.S. election polls with pollster ratings, historical house effects, and methodology information. Indispensable for practicing poll evaluation.

Ballotpedia's Polling Research (ballotpedia.org): State and local race polling data with methodology summaries. Useful for down-ballot races that FiveThirtyEight does not always cover.

Polling Report (pollingreport.com): Long-running archive of polling data on presidential approval, congressional races, and issue questions. Useful for historical trend analysis.

AAPOR Roper Center Poll Archive (ropercenter.cornell.edu): The most comprehensive academic archive of survey data, with full question text and methodology documentation for thousands of surveys dating to the 1930s.