Modelling Campaign Effects in the 2005 British Election Study Harold Clarke, David Sanders, Marianne...

21
Modelling Campaign Effects in the 2005 British Election Study Harold Clarke, David Sanders, Marianne Stewart and Paul Whiteley The 2005 British Election Study www.essex.ac.uk/bes BES 2005
  • date post

    21-Dec-2015
  • Category

    Documents

  • view

    216
  • download

    1

Transcript of Modelling Campaign Effects in the 2005 British Election Study Harold Clarke, David Sanders, Marianne...

Page 1: Modelling Campaign Effects in the 2005 British Election Study Harold Clarke, David Sanders, Marianne Stewart and Paul Whiteley The 2005 British Election.

Modelling Campaign Effects in the 2005 British Election

Study

Harold Clarke, David Sanders, Marianne Stewart and Paul Whiteley

The 2005 British Election Studywww.essex.ac.uk/bes

BES 2005

Page 2: Modelling Campaign Effects in the 2005 British Election Study Harold Clarke, David Sanders, Marianne Stewart and Paul Whiteley The 2005 British Election.

The Official Campaign Study The official campaign started on April 6th 2005 after

the election date was announced It continued until polling day on May 5th 2005 The BES conducted daily internet surveys averaging

approximately 210 respondents per day during this period.

These surveys represented the second wave of a panel (overall N=7794) – the first wave was conducted before the election was announced

Page 3: Modelling Campaign Effects in the 2005 British Election Study Harold Clarke, David Sanders, Marianne Stewart and Paul Whiteley The 2005 British Election.

The Probability of Voting Scale(Please think of a scale that runs from 0 to 10, where 0 means very unlikely and 10

means very likely, how likely is it that you will vote in the general election?)

Page 4: Modelling Campaign Effects in the 2005 British Election Study Harold Clarke, David Sanders, Marianne Stewart and Paul Whiteley The 2005 British Election.

Trends in the Probability of Voting in the Official Campaign

Probability of Voting

7.6

7.8

8

8.2

8.4

8.6

8.8

9

9.2

Probability of Voting

Page 5: Modelling Campaign Effects in the 2005 British Election Study Harold Clarke, David Sanders, Marianne Stewart and Paul Whiteley The 2005 British Election.

Campaign Exchanges, 21st April ‘Crime has gone down by 30 per cent since 1997 and

concern about loutish behaviour has begun to fall’ (Tony Blair, Labour’s Press conference, 21st April)

‘Violent crime has almost doubled, and there are now a million violent crimes committed every year. I know: my wife Sandra and my daughter Larissa have both been mugged’ (Conservative Leader, Michael Howard’s speech, 21st April)

‘Mother Stabbed on walk with Young Son’ (Guardian Headline, April 22 nd)

Page 6: Modelling Campaign Effects in the 2005 British Election Study Harold Clarke, David Sanders, Marianne Stewart and Paul Whiteley The 2005 British Election.

Campaign Exchanges, 22nd April ‘I said at the manifesto launch that I would deal with the issue of asylum

and immigration during this campaign… It is precisely because we have been working hard at it that over the past few years asylum claims have fallen in Britain faster than anywhere else in Europe’ (Tony Blair’s speech April 22nd)

‘If you’re unhappy about high taxes, uncontrolled immigration, rising crime and dirty hospitals, then why reward Mr Blair’s Labour party with your vote? Quite frankly, Tony Blair’s lost the plot. He’s talked a lot and failed to deliver. He’s told lies to win elections. And he’s only taken a stand on one thing in the last eight years – taking Britain to war. And he couldn’t even tell the truth about that’ (Michael Howard’s speech April 23rd)

Page 7: Modelling Campaign Effects in the 2005 British Election Study Harold Clarke, David Sanders, Marianne Stewart and Paul Whiteley The 2005 British Election.

Voter Reactions to the Leaders TV Debate on April 28th

(seen by 28 per cent of respondents)

0

5

10

15

20

25

30

35

40

45

Tony Blair MichaelHoward

CharlesKennedy

Most EffectiveLeast Effective

Page 8: Modelling Campaign Effects in the 2005 British Election Study Harold Clarke, David Sanders, Marianne Stewart and Paul Whiteley The 2005 British Election.

The voting turnout model

Vi = pi (B) - Ci + Di

Where: Vi is the probability of respondent i voting pi is respondent i’s perceptions of their own pivotality B are the collective benefits obtained from voting Ci are the costs of voting Di is i’s sense of civic duty to vote (see Riker and Ordeshook, 1968)

Page 9: Modelling Campaign Effects in the 2005 British Election Study Harold Clarke, David Sanders, Marianne Stewart and Paul Whiteley The 2005 British Election.

Perceptions of Pivotality On a scale from 0 to 10, where 10 means very unlikely and 0 means very likely,

how likely is it that Labour will win the election in your local constituency?

Likelihood Labour Win R's Constituency

Very likely

9.00

8.00

7.00

6.00

5.00

4.00

3.00

2.00

1.00

Very unlikely

Pe

rce

nt

20

10

0

Page 10: Modelling Campaign Effects in the 2005 British Election Study Harold Clarke, David Sanders, Marianne Stewart and Paul Whiteley The 2005 British Election.

Collective Benefits On a scale from 0 to 10, where 0 means strongly dislike and 10 means strongly like, how do

you feel about the Labour party?

Feeling Thermometer-Labour Party

Strongly like

9.00

8.00

7.00

6.00

5.00

4.00

3.00

2.00

1.00

Strongly dislike

Pe

rce

nt

20

10

0

Page 11: Modelling Campaign Effects in the 2005 British Election Study Harold Clarke, David Sanders, Marianne Stewart and Paul Whiteley The 2005 British Election.

Collective Benefits On a scale from 0 to 10, where 0 means strongly dislike and 10 means strongly like, how do

you feel about the Conservative party?

Feeling Thermometer-Conservative Party

Strongly like

9.00

8.00

7.00

6.00

5.00

4.00

3.00

2.00

1.00

Strongly dislike

Pe

rce

nt

30

20

10

0

Page 12: Modelling Campaign Effects in the 2005 British Election Study Harold Clarke, David Sanders, Marianne Stewart and Paul Whiteley The 2005 British Election.

Collective Benefits On a scale from 0 to 10, where 0 means strongly dislike and 10 means strongly like, how do

you feel about the Liberal Democrats?

Feeling Thermometer-Liberal Democrat Party

Strongly like

9.00

8.00

7.00

6.00

5.00

4.00

3.00

2.00

1.00

Strongly dislike

Pe

rce

nt

30

20

10

0

Page 13: Modelling Campaign Effects in the 2005 British Election Study Harold Clarke, David Sanders, Marianne Stewart and Paul Whiteley The 2005 British Election.

Collective Benefits Scale – difference between parties

Collective Benefits Scale

great dealfair amountsomelittle

Pe

rce

nt

40

30

20

10

0

Page 14: Modelling Campaign Effects in the 2005 British Election Study Harold Clarke, David Sanders, Marianne Stewart and Paul Whiteley The 2005 British Election.

Duty Indicator I‘I feel a sense of satisfaction when I vote’

Sense of Satisfaction From Voting

Don’t know

Strongly disagree

Disagree

Neither agree nor di

Agree

Strongly agree

Pe

rce

nt

40

30

20

10

0

Page 15: Modelling Campaign Effects in the 2005 British Election Study Harold Clarke, David Sanders, Marianne Stewart and Paul Whiteley The 2005 British Election.

Duty Indicator II‘I would be seriously neglecting my duty as a citizen if I didn’t vote’

Serious Neglect Citizen Duty If Not Vote

Don’t know

Strongly disagree

Disagree

Neither agree nor di

Agree

Strongly agree

Pe

rce

nt

50

40

30

20

10

0

Page 16: Modelling Campaign Effects in the 2005 British Election Study Harold Clarke, David Sanders, Marianne Stewart and Paul Whiteley The 2005 British Election.

Turnout Model – pooled data

Predictor Variables  

Pivotality * Collective Benefits

0.005*** (8.5)

Perceptions of Duty 0.82*** (39.6)

Male 0.35*** (5.3)

Age 0.02*** (10.9)

Ethnic minority status -1.26*** (4.5)

Educational Attainment 0.07*** (3.5)

Household Income 0.07*** (6.7)

R square 0.42

   

Page 17: Modelling Campaign Effects in the 2005 British Election Study Harold Clarke, David Sanders, Marianne Stewart and Paul Whiteley The 2005 British Election.

Multi-level Model of Campaigning Setup Individual Level Model

Yij = β0j + β1j X1ij + β2j X2ij + …. + βkj Xkij + r ij

Where: Yij is the dependent variable βij are the individual level model coefficients X1ij are the individual level predictor variables r ij is an individual level error term

Aggregate Level Model

βij = γq0 + γq1 W1j + γq2 W2j + γq3 W3j …. + γq4 Wsj + uqj

Where: β0j is the intercept from the individual level model β1j is the slope coefficient of variable X1 in the individual level model γqj are the coefficients of the aggregate level covariates Wij are the aggregate level covariates uqj is an aggregate level error term

Page 18: Modelling Campaign Effects in the 2005 British Election Study Harold Clarke, David Sanders, Marianne Stewart and Paul Whiteley The 2005 British Election.

Trends in the Attention to Politics Variable

5.2

5.4

5.6

5.8

6

6.2

6.4

6.6

6th

April 8th

10th

12th

14th

16th

18th

20th

22nd

24th

26th

28th

30th

2nd

4th

Page 19: Modelling Campaign Effects in the 2005 British Election Study Harold Clarke, David Sanders, Marianne Stewart and Paul Whiteley The 2005 British Election.

Trends in the Perceptions of Duty Variable

7.17.27.3

7.47.57.67.7

7.87.9

8

Page 20: Modelling Campaign Effects in the 2005 British Election Study Harold Clarke, David Sanders, Marianne Stewart and Paul Whiteley The 2005 British Election.

Multi-Level Random Intercept Turnout Model with Aggregate Covariates

Multi-level Model

Mean Pivotality * Collective Benefits -0.01 (0.7)

---

Mean Perceptions of Duty -0.46* (1.8)

-0.57** (2.0)

Mean Attention to Politics   0.38* (1.7)

0.39* (1.8)

Time Trend 0.002

(1.0)

---

Pivotality * Collective Benefits

0.005*** (8.5)

0.005** (8.4)

Perceptions of Duty 0.82*** (39.4)

0.82*** (39.5)

Male 0.35*** (5.3)

0.35*** (5.4)

Age 0.02*** (10.9)

0.02*** (10.9)

Ethnic minority status -1.26*** (4.5)

-1.30*** (4.5)

Educational Attainment 0.07*** (3.5)

0.07** (3.6)

Household Income 0.07*** (6.6)

0.07*** (6.7)

R square 0.42 0.42

   

Page 21: Modelling Campaign Effects in the 2005 British Election Study Harold Clarke, David Sanders, Marianne Stewart and Paul Whiteley The 2005 British Election.

Conclusions The individual’s sense of duty to vote dominates a

rational choice model of turnout Contextual variables relating to the campaign

influence individual turnout in this model An increase in people’s attention to politics arising

from the election campaign stimulates their turnout But a contextual increase in the duty to vote variable

tends to inhibit individual turnout