Debates: Voter and Politician Response to Political Communication

Debates: Voter and Politician Response to
Political Communication in Sierra Leone
Kelly Bidwell
November 5, 2015
Presentation Outline
•
•
•
•
Introduction
Institutional context: Sierra Leone
Intervention details and Data
Results
– Voter response
– Candidate response
– Disentangling effects: Persona vs. Policy
– Accountability
• Conclusions
Introduction
Motivation
• How do we deliver political information in a way that is
credible, engaging and accessible to voters who are
geographically remote, with low literacy levels and low media
penetration?
Motivation
•
•
Candidate debates have a rich history and are a key part of many campaign
strategies
However, there has been no definitive evidence of whether they affect actual
voting behaviour
– Many studies limited to panel opinion polls and impact is inconclusive
•
In countries with limited media penetration, limited reliable political information
and low literacy – debates could be especially powerful
– Effects on voters could be significant and be linked to electoral outcomes
– Candidates can stand on equal footing and discuss key policy questions – could facilitate
election of better candidates
– Publicizing campaign promises could enhance accountability
– Impact in this context could be more pronounced and persistent
Insights from Related Literature
•
Providing information about incumbent performance and candidate qualifications
can have large effects on voting:
– Brazil: publicly released audits before election
– India: report cards on incumbent legislator performance
However…
– Drawbacks to increasing transparency along one dimension (Liessem and Gersbach,
2003; Prat, 2005)
– Exposing high levels of corruption could demotivate voters (Chong et al, 2014)
– Politicians may seek to undermine the validity of 3rd party information (Humphreys and
Weinstein, 2012)
•
Working with politicians to engage voters can have positive effects
– Benin: public deliberation of policy options in town hall meetings between party
representative and constituents
Institutional Context: Sierra Leone
Context: Political System
Context: Ethnicity and Geography
Context: Voters
Election Day (Nov 2012) exit polls from villages that didn’t receive the intervention
reveal:
• Only 28% of voters could name the two Parliamentary candidates
• 64% couldn’t name a single MP job responsibility
• 3% knew amount of the MPs’ constituency facilitation fund
• 35% knew the proposed quota for women’s representation in government and
17% knew the candidates’ positions on bill
Intervention Details and Data
The Intervention
•
•
•
Search for Common Ground, hosted
and disseminated debates via a
mobile cinema “road show”
Debates were between MP
candidates from the three largest
parties – filmed and moderated by
SFCG
Debates were followed a
standardised format:
–
–
–
–
Moderator introduction
Get to know you questions
5 same national policy questions
2 local policy questions
The Intervention
•
Parties
– Received scorecards on 54 MPs & 105 LCs based on surveys with constituents- over
115 copies distributed
– Encouraging them to nominate better qualified candidates
• Voters
•
– Debates between candidates in 14 constituencies
– 197 screenings, over 200 additional communities invited
– Estimated to reach over 37,000 voters
– Received information on candidate policies, personal characteristics and qualifications
Candidates
– Encouraging them to invest more time and money in their constituencies
The Intervention
Bridges….
Roads and breakdowns…
The Timeline
•
•
•
Debates
– Pilot: January – February 2012
– Filming of Debates: October 2012
– Screening: October – November
2012
– Election and Exit Poll: November
2012
Report Cards
– Data Collection: January – April 2012
– Presentations: May – August 2012
Follow up
– Through 2013
The Data
•
•
•
Representative survey
– Randomly selected
households/individuals
– spread across 14 constituencies
(highlighted in yellow)
Before and After Debate Survey
– Knowledge of politics and
candidates, policy
– stances, voting intentions
– 5000 surveys in 10 districts
Exit Poll Survey
– Same people, policy questions, who
they voted for and why
– Over 8,000 surveys in 10 districts
– Collected over 6+ days
Results: Voter Response
Voters learned about constituency
resources and candidate
qualifications
Voters learned about policy positions
Debates changed how people voted
Crossing party lines
Results: Pre-Election Period Local Actions
Pre-election Campaign Period
• Increases in voter reports of having received more gifts, the
reported monetary value of gifts, and the number of
reported visits to their community by candidates
– This campaign response is stronger for the party that generally had
less chance of winning seats
– Also stronger where the “outsider” candidate won the debate
Results: Persona vs. Policy
Debates Combined Persona and
Policy Information
Only Debates Changed Voting
Results: Accountability
Impact on MP Behavior
•Compared the 14 MPs from constituencies with debates to the 14 not
chosen for debates
•No difference in activity in parliament or consistency of stated priority
issue over time
•Increase in constituency engagement
•
More visits to constituency (4.2 vs 2.9), more meetings held, more likely to be
thought of by clinic staff as doing a good job
•Increase in % of Constituency Facilitation Fund expenditure verified on
the ground
•
•
•
Comparison MPs, 38%
Treatment MPs, 94% (can be more than 100%)
Translates to $6,000 more spending per constituency
CFF Verification: Treatment vs. Comparison
Conclusions
Conclusions
• Voters acquired significant political knowledge from watching debates.
• This knowledge persisted over weeks and influenced voting behaviour
on Election Day.
• Debates can make political contests more competitive, which is
consistent with the observed increase in campaign effort
• Information conveyed by debates is comprehensive
• Debates publicity enhances accountability pressure over elected
politicians, increasing constituency engagement and development
spending
Policy Implications
•Debates are logistically feasible to host and disseminate:
•
•
Candidates are willing to participate
Voters exhibit high demand for political information
•Accountability effects cover the costs of video production
•
Increasing scale of dissemination would likely enhance observed accountability
effects and implicit cost recovery via development spending
•Dissemination via mobile cinema is relatively resource intensive
•
Alternative mass dissemination mechanisms with lower marginal costs include
television and radio
•Commissioners in Sierra Leone’s National Electoral Commission are
interested in taking debates to scale