Estimating the political center from aggregate data: an item response theory alternative to the Stimson dyad ratios algorithm

Research output: Contribution to journalArticle

15 Citations (Scopus)

Abstract

This article provides an algorithm to produce a time series estimate of the political center (or median voter) from aggregate survey data, even when the same questions are not asked in most years. This is compared to the existing Stimson dyad ratios approach, which has been applied to various questions in political science. Unlike the dyad ratios approach, the model developed here is derived from an explicit model of individual behavior – the widely used item response theory model. I compare the results of both techniques using the data on public opinion from the United Kingdom from 1947-2005 from Bartle, Dellepiane-Avellaneda and Stimson (2011a). Measures of overall model fit are provided, as well as techniques for testing model’s assumptions and the fit of individual items. Full code is provided for estimation with free software WinBUGS and JAGS.
LanguageEnglish
Pages115-129
Number of pages15
JournalPolitical Analysis
Volume22
Issue number1
Early online date4 Dec 2013
DOIs
Publication statusPublished - 2014

Fingerprint

political center
aggregate data
dyad
model theory
political science
public opinion
time series
voter

Keywords

  • algorithms
  • Stimson dyad ratios algorithm
  • survey data
  • response theory model

Cite this

@article{8aec53d7feb74b2ab52c122cdd5d0727,
title = "Estimating the political center from aggregate data: an item response theory alternative to the Stimson dyad ratios algorithm",
abstract = "This article provides an algorithm to produce a time series estimate of the political center (or median voter) from aggregate survey data, even when the same questions are not asked in most years. This is compared to the existing Stimson dyad ratios approach, which has been applied to various questions in political science. Unlike the dyad ratios approach, the model developed here is derived from an explicit model of individual behavior – the widely used item response theory model. I compare the results of both techniques using the data on public opinion from the United Kingdom from 1947-2005 from Bartle, Dellepiane-Avellaneda and Stimson (2011a). Measures of overall model fit are provided, as well as techniques for testing model’s assumptions and the fit of individual items. Full code is provided for estimation with free software WinBUGS and JAGS.",
keywords = "algorithms, Stimson dyad ratios algorithm, survey data, response theory model",
author = "McGann, {Anthony J.}",
year = "2014",
doi = "10.1093/pan/mpt022",
language = "English",
volume = "22",
pages = "115--129",
journal = "Political Analysis",
issn = "1047-1987",
number = "1",

}

TY - JOUR

T1 - Estimating the political center from aggregate data

T2 - Political Analysis

AU - McGann, Anthony J.

PY - 2014

Y1 - 2014

N2 - This article provides an algorithm to produce a time series estimate of the political center (or median voter) from aggregate survey data, even when the same questions are not asked in most years. This is compared to the existing Stimson dyad ratios approach, which has been applied to various questions in political science. Unlike the dyad ratios approach, the model developed here is derived from an explicit model of individual behavior – the widely used item response theory model. I compare the results of both techniques using the data on public opinion from the United Kingdom from 1947-2005 from Bartle, Dellepiane-Avellaneda and Stimson (2011a). Measures of overall model fit are provided, as well as techniques for testing model’s assumptions and the fit of individual items. Full code is provided for estimation with free software WinBUGS and JAGS.

AB - This article provides an algorithm to produce a time series estimate of the political center (or median voter) from aggregate survey data, even when the same questions are not asked in most years. This is compared to the existing Stimson dyad ratios approach, which has been applied to various questions in political science. Unlike the dyad ratios approach, the model developed here is derived from an explicit model of individual behavior – the widely used item response theory model. I compare the results of both techniques using the data on public opinion from the United Kingdom from 1947-2005 from Bartle, Dellepiane-Avellaneda and Stimson (2011a). Measures of overall model fit are provided, as well as techniques for testing model’s assumptions and the fit of individual items. Full code is provided for estimation with free software WinBUGS and JAGS.

KW - algorithms

KW - Stimson dyad ratios algorithm

KW - survey data

KW - response theory model

UR - http://pan.oxfordjournals.org/

U2 - 10.1093/pan/mpt022

DO - 10.1093/pan/mpt022

M3 - Article

VL - 22

SP - 115

EP - 129

JO - Political Analysis

JF - Political Analysis

SN - 1047-1987

IS - 1

ER -