Multicandidate Campaigns

1
Multicandidate Campaigns

• Menu-Dependent Choice Behavior
• Renan Levine
• Duke University

2
Did Nader help Gore?

• Political Science No
• Violates Independent of Irrelevant Alternatives.
• Behavioral Decision Theory (Consumer behavior)
  YES!
• Simonson (1989), Sen (1997), Pettibone and Wedell
  (2000)
• Certain items are more desirable when compared to
  certain other items.
• Provides reason for choice
• Affects perceptions, difficulty of choice, risk
  perceptions, evaluative anxiety and facilitates
  comparisons

3
Can Some Candidates Appear More Desirable in
Comparison with Certain Other Candidates?
1948
H. Wallace
Truman
Dewey
Thurmond
4
Why Three Sizes of Popcorn?
Econ Product differentiation / Market
segmentation Add Large Popcorn, sales of Regular
should go down. Instead, sales of Regular
increases! Large takes sales away from Regular,
but Regular takes sales away from Small.
5
Why do sales of Medium increase?

• Informational effect
• Medium now appears moderate sized and priced.
• Small Too Small? Large Too Expensive?
• Non-informational/emotional effect
• Medium is less-difficult choice / easiest to
  justify.
• Compromise Effect
• Attraction effect compensates for substitution
  effect (Simonson 1989, Pettibone and Wedell 2000)

6
Menu-Dependent Preferences

• Occurs when decision involves one or multiple
  dimensions (Sen 1997)
• Common when attributes are hard to evaluate or
  hard to map onto utility.
• Decision is difficult
• Good options
• Lesser of evils
• Ideal is poorly defined or conditional
• May be especially likely in politics

7
What makes my study different

• If Voters ? Consumers
• If Candidates ? Popcorn
• Nader helped Gore win votes away from Bush!!
• Conventional Wisdom (Abramson, Aldrich and Rohde
  2002)
• ¾ of Nader Voters in FL said Gore was their
  second choice
• If they had voted for Gore, Gore would be
  President.
• Cannot test directly

8
Experiment

• People on street / community groups in
  Triangle, North Carolina
• Randomly assigned to conditions
• Control condition
• Subjects read profile of two local candidates
• Candidates differ on growth issue.
• Salient in a high growth area.
• Three Treatments
• Three candidates
• Extreme, Compromise, Off-Dimension

9
Experiment Results
Pro-Growth -gt Anti-Growth
A
B
Extreme Candidate C" wins more than 20 but share
of B hardly drops!
42
58
( C" in paper )
10
B won votes away from A
Control
Treatment
More Likely to Vote B
11
B won Republicans
Vote Distribution Republicans Only
12
Why? Choice Difficulty

• People who found choice to be difficult were more
  likely to vote for B
• If choice was difficult, odds of voting is 18
  times larger.
• In 3rd Treatment, people who found choice to be
  difficult were more likely to vote for another
  compromise candidate.
• Works in addition to strategic considerations
  (odds of voting are 8 times larger).

13
Why? Perceptual

• ½ point shift in average placement of Candidate B
  in presence of C
• On 5-pt scale, Bs average placement 1.9
• In presence of C, Bs mean 2.56
• Shift helps B win votes of subjects who place
  themselves between A and B.
• More respondents thought that B was within
  one-point of their ideal
• More thought that B was closer than A to ideal.
• Simulated move in placement of B in control and
  found shift tipped likelihood of voting for B
  for respondents 2 points away from B.

14
Implications for Poli Sci

• Choice is result of comparative process.
• Menu MATTERS (Sen 1997, Simonson 1989)
• Rational Choice axioms need to be revisited and
  revised.
• Completeness (Sen, Machina and others)
• Take into account difficulty of choice
• Huber, Payne and Puto 1982 Pettibone and Wedell
  2000 Luce, Payne and Bettman 2001
• But dont throw out spatial models of
  competition.

15
Politics Policy Implications

• Potential heresthetic tool.
• Electoral
• Get someone loony to run on your flank.
• Triangulate using other politicians.
• Policy initiatives and processes
• Initial proposals in a committee are not neutral.
  They could anchor perceptions of later options or
  make certain options more likely to be choice.
• Might help if there is someone or another group
  advocating something extreme or radical.
• Choice difficulty may help explain status quo
  bias.