Polling bias and undecided voter allocations: US presidential elections, 2004–2016

Joshua J. Bon, Timothy Ballard, Bernard Baffour

Research output: Contribution to journalArticle

Abstract

Accounting for undecided and uncertain voters is a challenging issue for predicting election results from public opinion polls. Undecided voters typify the uncertainty of swing voters in polls but are often ignored or allocated to each candidate in a simple deterministic manner. Historically this may have been adequate because the undecided voters were sufficiently small to assume that they do not affect the relative proportions of the decided voters. However, in the presence of high numbers of undecided voters, these static rules may in fact bias election predictions from election poll researchers and metapoll analysts. We examine the effect of undecided voters in the 2016 US presidential election compared with the previous three presidential elections. We show that there were a relatively high number of undecided voters over the campaign and on election day, and that the allocation of undecided voters in this election was not consistent with two-party proportional (or even) allocations. We find evidence that static allocation regimes are inadequate for election prediction models and that probabilistic allocations may be superior. We also estimate the bias attributable to polling agencies, which is often referred to as ‘house effects’.

Original languageEnglish
Pages (from-to)467-493
Number of pages27
JournalJournal of the Royal Statistical Society. Series A: Statistics in Society
Volume182
Issue number2
Early online date8 Oct 2018
DOIs
Publication statusPublished - Feb 2019

Fingerprint

Polling
Vote
Elections
presidential election
voter
election
trend
election result
opinion poll
public opinion
candidacy
campaign
Presidential elections
Voters
uncertainty
regime
Prediction Model
Proportion
Directly proportional
evidence

Cite this

@article{eb68228b82e84a7aa070817767c43fac,
title = "Polling bias and undecided voter allocations: US presidential elections, 2004–2016",
abstract = "Accounting for undecided and uncertain voters is a challenging issue for predicting election results from public opinion polls. Undecided voters typify the uncertainty of swing voters in polls but are often ignored or allocated to each candidate in a simple deterministic manner. Historically this may have been adequate because the undecided voters were sufficiently small to assume that they do not affect the relative proportions of the decided voters. However, in the presence of high numbers of undecided voters, these static rules may in fact bias election predictions from election poll researchers and metapoll analysts. We examine the effect of undecided voters in the 2016 US presidential election compared with the previous three presidential elections. We show that there were a relatively high number of undecided voters over the campaign and on election day, and that the allocation of undecided voters in this election was not consistent with two-party proportional (or even) allocations. We find evidence that static allocation regimes are inadequate for election prediction models and that probabilistic allocations may be superior. We also estimate the bias attributable to polling agencies, which is often referred to as ‘house effects’.",
keywords = "Bayesian modelling, Election polls, Total survey error",
author = "Bon, {Joshua J.} and Timothy Ballard and Bernard Baffour",
year = "2019",
month = "2",
doi = "10.1111/rssa.12414",
language = "English",
volume = "182",
pages = "467--493",
journal = "Journal of the Royal Statistical Society Series A: Statistics in Society",
issn = "0964-1998",
publisher = "Wiley-Blackwell",
number = "2",

}

Polling bias and undecided voter allocations : US presidential elections, 2004–2016. / Bon, Joshua J.; Ballard, Timothy; Baffour, Bernard.

In: Journal of the Royal Statistical Society. Series A: Statistics in Society, Vol. 182, No. 2, 02.2019, p. 467-493.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Polling bias and undecided voter allocations

T2 - US presidential elections, 2004–2016

AU - Bon, Joshua J.

AU - Ballard, Timothy

AU - Baffour, Bernard

PY - 2019/2

Y1 - 2019/2

N2 - Accounting for undecided and uncertain voters is a challenging issue for predicting election results from public opinion polls. Undecided voters typify the uncertainty of swing voters in polls but are often ignored or allocated to each candidate in a simple deterministic manner. Historically this may have been adequate because the undecided voters were sufficiently small to assume that they do not affect the relative proportions of the decided voters. However, in the presence of high numbers of undecided voters, these static rules may in fact bias election predictions from election poll researchers and metapoll analysts. We examine the effect of undecided voters in the 2016 US presidential election compared with the previous three presidential elections. We show that there were a relatively high number of undecided voters over the campaign and on election day, and that the allocation of undecided voters in this election was not consistent with two-party proportional (or even) allocations. We find evidence that static allocation regimes are inadequate for election prediction models and that probabilistic allocations may be superior. We also estimate the bias attributable to polling agencies, which is often referred to as ‘house effects’.

AB - Accounting for undecided and uncertain voters is a challenging issue for predicting election results from public opinion polls. Undecided voters typify the uncertainty of swing voters in polls but are often ignored or allocated to each candidate in a simple deterministic manner. Historically this may have been adequate because the undecided voters were sufficiently small to assume that they do not affect the relative proportions of the decided voters. However, in the presence of high numbers of undecided voters, these static rules may in fact bias election predictions from election poll researchers and metapoll analysts. We examine the effect of undecided voters in the 2016 US presidential election compared with the previous three presidential elections. We show that there were a relatively high number of undecided voters over the campaign and on election day, and that the allocation of undecided voters in this election was not consistent with two-party proportional (or even) allocations. We find evidence that static allocation regimes are inadequate for election prediction models and that probabilistic allocations may be superior. We also estimate the bias attributable to polling agencies, which is often referred to as ‘house effects’.

KW - Bayesian modelling

KW - Election polls

KW - Total survey error

UR - http://www.scopus.com/inward/record.url?scp=85054474854&partnerID=8YFLogxK

U2 - 10.1111/rssa.12414

DO - 10.1111/rssa.12414

M3 - Article

VL - 182

SP - 467

EP - 493

JO - Journal of the Royal Statistical Society Series A: Statistics in Society

JF - Journal of the Royal Statistical Society Series A: Statistics in Society

SN - 0964-1998

IS - 2

ER -