Factors affecting 'Alphabetical Voting' in the Scottish Local Government Elections in 2012

Full text

Factors affecting ‘Alphabetical
Voting’ in the Scottish Local
Government Elections in 2012
James Gilmour
Electoral Reform Society Scotland
EPOP 2014

Scottish Local Government Elections
‘Alphabetical’ voting – list order effect
Candidate with the more First Preference Votes
in Wards where Parties nominated 2 Candidates
Election 2007 2012
Higher placed 277 331
Lower placed 50 82
Probability of such
result by chance 7.7 * 10-36 3.0 * 10-34
Data sources: 2007 Bochel & Denver 2012 Bochel, Denver & Steven

STV
Ballot
Paper
2012
3-members
189 wards
4-members
164 wards
2,496
candidates
23%women
1,223
councillors:
24% women

Scottish Local Government Elections 2012
Data Sources
•Electoral Reform Society Scotland
Juliet Swann & Malcolm Harvy
•Candidates’ Names and Sex
Meryl Kenny –University of Leicester
Fiona Mackay –University of Edinburgh
•Distribution of population surnames
Neil Bowie & G W L Jackson
GROS Occasional Paper no.9
Data from 1999/2000/2001

Comparisons of Party Pairs
•421 Party Pairs
one pair two pairs three pairs
134 wards 133 wards 7 wards
•Statistical analysis excluded
−6 pairs from 4 parties with only 1 or 2 pairs
−9 pairs that included an “incumbent” transferred
from another ward
giving 406 robust pair comparisons

Paired Party Candidates' Shares of
Party’s First Preference Votes in Ward
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
121 41 61 81 101 121 141 161 181 201 221 241 261 281 301 321 341 361 381 401 421
421 Pair Comparisons (sorted by shares of first preference votes)
Upper Candidate's Share
of Party Vote in Ward
Lower Candidate's Share
of Party Vote in Ward

PAIRS: Position within Party Pair
mean shares of party first preference votes in ward
Position Upper Lower N
Overall 64% 36% 406
Primary metric:
candidate’s share of party’s first preference votes in ward
−removes effects of differences in electorate, turnout
and party support
−for statistical analysis candidates’ shares of party
first preference votes were transformed to logits

PAIRS: Party
mean shares of party first preference votes in ward
Party Upper Lower N
Cons 64% 36% 26
Lab 62% 38% 152
LD 53% 48% 15
SNP 62% 38% 213

PAIRS: Sex
mean shares of party first preference votes in ward
Sex Upper NLower N
Female 61% 111 37% 96
Male 62% 295 39% 310

PAIRS: Party & Sex
mean shares of party first preference votes in ward
Party Sex Upper NLower N
Cons F 69% 528% 5
M 62% 21 39% 21
Lab F 56% 43 32% 39
M 64% 109 40% 113
LD F 68% 550% 7
M 45% 10 45% 8
SNP F 63% 58 40% 45
M 62% 155 38% 168

PAIRS: Incumbent
mean shares of party first preference votes in ward
Incumbent
Upper NLower N
Yes 67% 243 46% 197
No 54% 163 31% 209

PAIRS: Incumbent & Party
mean shares of party first preference votes in ward
Incumbent
Party
Upper NLower N
Yes
Cons
68% 19 46% 13
Lab 67% 98 44% 77
LD 55% 12 55% 10
SNP 68% 114 46% 97
No
Cons
51% 727% 13
Lab 53% 54 32% 75
LD 43% 333% 5
SNP 55% 732% 116

PAIRS: Incumbent & Sex
mean shares of party first preference votes in ward
Incumbent
Sex Upper NLower N
Yes F69% 58 44% 45
M66% 185 46% 152
No F53% 53 30% 51
M55% 110 32% 158

PAIRS: Elected
mean shares of party first preference votes in ward
Elected Upper NLower N
Yes 64% 364 43% 246
No 39% 42 31% 160

PAIRS: Elected & Incumbent
mean shares of party first preference votes in ward
Elected
Incumbent
Upper NLower N
Yes Yes 68% 99 40% 63
No 65% 265 44% 183
No Yes 42% 12 30% 33
No 39% 30 31% 127

PAIRS: Elected & Sex
mean shares of party first preference votes in ward
Elected Sex Upper NLower N
Yes F63% 99 40% 63
M65% 265 44% 183
No F42% 12 30% 33
M39% 30 31% 127

PAIRS: Number of Party Pairs in Ward
mean shares of party first preference votes in ward
Number Upper Lower N
160% 40% 131
262% 38% 258
366% 34% 17

PAIRS: Number of Candidates in Ward
mean shares of party first preference votes in ward
Candidates Upper Lower N
4 73% 27% 10
5 60% 40% 45
6 59% 41% 86
7 63% 37% 106
8 62% 38% 72
9 63% 37% 38
10 59% 41% 20
11 65% 35% 17

PAIRS: Position on Ballot Paper
mean shares of party first preference votes in ward
Position Upper NLower N
1 63% 107 *0
2 63% 101 39% 21
3 59% 90 36% 51
4 62% 49 40% 62
5 59% 29 39% 81
6 67% 16 41% 65
7 57% 12 35% 57
8 * 037% 38

PAIRS: Top or Bottom on Ballot Paper
mean shares of party first preference votes in ward
Position Upper NLower N
Top 63% 107 **
not Top
not Bottom
61% 299 39% 296
Bottom * *37% 110

PAIRS: Pair Separation on Ballot Paper
mean shares of party first preference votes in ward
Separation Upper Lower N
1 61% 39% 136
2 60% 40% 87
3 63% 37% 67
4 63% 37% 46
5 62% 38% 25
6 64% 36% 27
7 64% 36% 6
8 53% 47% 6

PAIRS: Optimal Regression Model
Main Effects of 7 Factors
logit share of party first preference votes in ward
term added % variance
accounted for
residual
mean square
constant - 0.7128
Upper v Lower
37.9% 0.4428
Incumbent 51.2% 0.3479
Ballot paper
position 51.8% 0.3439
Party 51.8% 0.3435

PAIRS: Optimal Regression Model
Main Effects and First Order Interactions
logit share of party first preference votes in ward
term added % variance
accounted for
residual
mean square
4
m
ain effects
51.8% 0.3435
Party x
Upper v Lower
52.8% 0.3366
Party x
Incumbent 52.9% 0.3357
B.P
.
position x
Incumbent 53.0% 0.3353

PAIRS: Local Government Area
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
211 12 10 16 12 15 711 33 10 20 617 17 12 36 719 735 4 5 8 7 28 12 14 8 5
30 LGA Comparisons (labelled by number of pair comparisons in LGA)
Mean Upper Candidate's
Share of Party Vote in Ward
Mean Lower Candidate's
Share of Party Vote in Ward

Three Party Candidates' Shares of
Party’s First Preference Votes in Ward
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
1 5 9 13 17 21 25 29 33 37 41 45
45 Comparisons (sorted by High Candidate's share of first preference votes)
Low Candidate's Share
of Party Vote in Ward
High Candidate's Share
of Party Vote in Ward
Middle Candidate's Share

Alphabetical Distribution of Surnames
0%
2%
4%
6%
8%
10%
12%
A B C D E F G H I J K L Mac Mc M N O P Q R S T U V W X Y Z
Population

Alphabetical Distribution of Surnames
0%
2%
4%
6%
8%
10%
12%
A B C D E F G H I J K L Mac Mc M N O P Q R S T U V W X Y Z
Population Candidates 2012

Alphabetical Distribution of Surnames
0%
2%
4%
6%
8%
10%
12%
A B C D E F G H I J K L Mac Mc M N O P Q R S T U V W X Y Z
Population Candidates 2012 Councillors 2012

Future Ordering of the Ballot Paper ?
•Robson rotation
–superceded technology
•One randomisation for each ward
–recommended in Gould Report (2007 elections)
–would remove ‘alphabetical’ effect but not
the list order effect
•Full randomisation of every ballot paper
–confusion? − disability discrimination?