8A_2_A containment-first search algorithm for higher-order analysis of urban ...
8A_1_To vote or not to vote
1. 1 To vote or not to vote? Investigating changes in the predicted probability of voter turnout when re-siting polling stations Scott Orford WISERD Cardiff University GISRUK 2010
2. 2 Structure Introduction Micro-geographical factors that affect voter turnout Brent case study Binomial ML model Predicting changes in turnout Conclusions
3. 3 Introduction Concern over low turnout and the ‘democratic defict’ Turnout gap in GB largest of any Western liberal democracy (25 – 40 percentage points) Factors influencing turnout at elections well known Research tends to be election specific and not systematic Little still known about importance of spatial and micro-geographical factors in a UK context
4. 4 People usually have to vote in person at a designated polling station Polling district boundaries and stations are determined by the council – administrative function Not accountable to the boundary commission Accessibility: “if possible, it needs to be close to where voters live and be fully accessible” A review of polling districts and polling stations must take place at least once every four years
5. 5 Possible factors when siting polling stations affect turnout Distance Morphology (compactness) Voter density (compactness & distance) Terrain Ease of parking etc Opportunities How do these vary in different elections? Rural/ suburban/inner-city differences (US research says there are)
11. Geographical factorsLocal campaigning Marginality of seat (closeness of contest) Population stability Social composition “People who talk together vote together” (Pattie and Johnston) – clear evidence that conversation and context can influence voting behaviour
30. 25 Voter dispersion (density) measures (combined measure of compactness and distance) Euclidean distance measures (metres) Percentage of postcodes in PD less than X metres from polling station Where X is 100, 200, 300, 400, 500, 600, 700, 750, 1000, 1250, and 1500 Road network distance measures (metres) Percentage of postcodes in PD less than X metres from polling station Where X is 100, 200, 300, 400, 500, 600, 700, 800, 900, 1000, 1250, 1500, 2000, and 2500
32. 27 Model Specification i = 1, …, 115 polling districts; j = 1, …, 31 wards; k = 1, …, 3 constituencies; Dependent variable is the proportion of turnout at the polling district with postal voters removed Model specification is binomial with a logit link Estimated using second order predictive quasi-likelihood (PQL) in MLwiN 2.10
35. 30 Voter density estimates (network distance) Election Maximum significance B-value T-stat European: Density ND < 500m 0.040 3.08 Local: Density ND < 600m 0.070 3.07 E.g. European and (Local) elections If 50% of voters in a PD live within 500m (600m) of polling station, turnout increases by 2% (3.5%) If 100% of voters in a PD live within 500m (600m) of polling station, turnout increases by 4% (7%)
42. 37 Differences in the predicted probabilities of turnout by constituency and election at the locations of maximum, minimum and average voter densities
43. 38 Percentage differences in the predicted probability of turnout at ward level when re-siting polling stations in the European election
44. 39 Percentage differences in the predicted probability of turnout at ward level when re-siting polling stations in the local election
45. 40 Percentage differences in the predicted probability of turnout at polling district level when re-siting polling stations in the European election
46. 41 Percentage differences in the predicted probability of turnout at polling district level when re-siting polling stations in the local election
47. 42 Percentage differences in the predicted probability of turnout at polling district level when re-siting polling stations at the maximum and minimum voter density locations for European and local elections
48. 43 Conclusions Supports idea of second order elections and rational choice theory of voting Geographical factors are influential in lower salience elections EEA 4 year review – perhaps examine polling station location with regards to accessibility and voter densities Target certain polling districts and re-site polling station Problem – trade-off between existing polling station building and portable polling stations (cost effectiveness) New voting technologies may decrease numbers of polling stations and therefore increase accessibility and decrease turnout