Selling Favors in the Lab: Experiments on Campaign Finance Reform Daniel Houser Thomas Stratmann George Mason University October, 2005.

Selling Favors in the Lab: Experiments on Campaign Finance

Reform

Daniel HouserThomas Stratmann

George Mason University

October, 2005

Purpose• Use experiments with human

subjects to investigate the political-economic implications of various campaign finance policies.

• Characterize how voter decisions are affected by advertising in different campaign finance regimes.

• Shed light on validity of various campaign finance models, particularly Coate (2004).

Experiment Design• Incorporates the salient features of Coate

(2004)• Power hungry candidates of high or low

quality, candidates always want to be elected.

• Only swing voters included, and they don’t know candidate quality.

• Ads are truthful and costly. Ads can be financed by special interests or the public.

• Electing a candidate who advertised using special interest money is costly to voters.

Screen Candidates See During Campaign

Screen Voters See During Campaign

Voting Screen

Data• We report data from eight sessions,

each including 24-26 unique subjects.• Each session included between 8 and

15 campaigns with two unique, randomly selected candidates for each campaign.

• Total of 2,509 observations on 194 subjects.

• We draw inferences using panel data procedures that account for both session and individual effects.

Voting Equation • Probability that a randomly selected voter will vote for the

R candidate is:1

1

1

1

1

1

1

Pr( ( ) | )Pr( ( ,0) ( )) Pr( ( ,0) | )Pr( ( ,1) ( )) Pr( ( ,1) | )Pr( ( , 2) ( )) Pr( ( , 2) | )Pr( ( , 2) ( )) Pr( ( , 2) | )Pr( ( ,0) ( )) Pr( ( ,0) | )Pr( ( ,1) ( )) Pr( ( ,1) | )Pr( ( , 2)

i j R ni R j R i R ni R j R i R ni R j R i R ni R j R i R ni L j R i L ni L j R i L ni L

1

1

( )) Pr( ( , 2) | )Pr( ( , 2) ( )) Pr( ( , 2) | )

j R i L ni L j R i L n

Probability of Voting for Own Party Striped Candidate by Number of Striped Ads Received and Treatment

0.00

0.20

0.40

0.60

0.80

1.00

0 1 2 3+

Number of Ads Received

Prob

abili

ty Blue

Red

Cap

Matching

Probability of Voting for Other Party Striped Candidate by Number of Striped Ads Received and Treatment

0.00

0.20

0.40

0.60

0.80

1.00

0 1 2 3+

Number of Ads Received

Prob

abili

ty Blue

Red

Cap

Matching

Results

Expected and Actual Margin of Victory by Number of Ads

-10

-5

0

5

10

15

20

0 10 20 30 40 50

Number of Ads

Mar

gin

of V

icto

ry

Red Expected

Blue Expected

Blue Actual

Red Actual

Results

Expected and Actual Aggregate Switch Votes Blue Campaigns

0123456789

10

0 10 20 30 40 50Number of Ads

Num

ber o

f Sw

itch

Vote

sExpected SwitchStripedActual Switch Striped

Expected Switch Solid

Actual Switch Solid

Expected and Actual Aggregate Switch Votes Red Campaigns

0123456789

10

0 10 20 30 40 50Number of Ads

Num

ber o

f Sw

itch

Vote

s

Expected SwitchStripedActual Switch Striped

Expected Switch Solid

Actual Switch Solid

Results

Margin of Victory and Net Aggregate Benefit of Electing Striped CandidateBlue and Red Campaigns

-200.00

-150.00

-100.00

-50.00

0.00

50.00

100.00

-10 -5 0 5 10 15 20 25

Margin of Victory

Aggr

eage

Val

ue S

tripe

d -

Aggr

egat

e Va

lue

Solid

Blue Campaigns

Red Campaigns

Results

Probability of Voting for Own Party Striped Candidate by Number of Striped Ads Received and Treatment

0.00

0.20

0.40

0.60

0.80

1.00

0 1 2 3+

Number of Ads Received

Prob

abili

ty Blue

Red

Cap

Matching

Probability of Voting for Other Party Striped Candidate by Number of Striped Ads Received and Treatment

0.00

0.20

0.40

0.60

0.80

1.00

0 1 2 3+

Number of Ads Received

Prob

abili

ty Blue

Red

Cap

Matching

Results

Margin of Victory and Net Aggregate Benefit of Electing Striped Candidate

Blue and Blue Cap

-200.00

-150.00

-100.00

-50.00

0.00

50.00

100.00

-10 -5 0 5 10 15

Margin of Victory

Aggr

eage

Val

ue S

tripe

d -

Aggr

egat

e Va

lue

Solid

Blue Campaigns

Blue Cap Campaigns

Results

Expected and Actual Margin of Victory by Number of Ads

Blue and Matching Campaigns

-10

-5

0

5

10

0 5 10 15 20

Number of Ads

Mar

gin

of V

icto

ry

Matching Expected

Blue Expected

Blue Actual

Matching Actual

Results

Margin of Victory and Net Aggregate Benefit of Electing Striped Candidate

Blue and Matching

-200.00

-150.00

-100.00

-50.00

0.00

50.00

100.00

-10 -5 0 5 10 15 20

Margin of Victory

Aggr

eage

Val

ue S

tripe

d -

Aggr

egat

e Va

lue

Solid

Blue Campaigns

Matching Campaigns

Results

Conclusion

• Voters are sophisticated: they “update” (respond to advertising) differently in different campaign finance environments, and in a way that is consistent with theory.

• The results inform efficient design of campaign finance policy.