Selective online exposure and political polarization during Swedish election campaignsa longitudinal analysis using four waves of panel data
Peter M. Dahlgren
Adam Shehata
Jesper Strömbäck
@peterdalle
peterdahlgren.com
Selective online exposure
Background
• Internet is a high-choice media environment
• Personal motivations increasingly important in news selection
• Prefer attitude-consistent news, but don’t avoid attitude-inconsistent
• Political ideologies more salient during elections
Purpose
Examine how political ideology and selective exposure mutuallyreinforce over time, and whether they become more extreme
Media Use
Belief/attitude/
behavior
Media Use
Belief/attitude/
behavior
Media Use
Belief/attitude/
behavior
Reinforcing spirals model
Slater (2007, p. 284)
Time 1 Time 2 Time 3
Media Use
Belief/attitude/
behavior
Media Use
Belief/attitude/
behavior
Media Use
Belief/attitude/
behavior
Reinforcing spirals model
Slater (2007, p. 284)
Time 1 Time 2 Time 3
Media Use
Belief/attitude/
behavior
Media Use
Belief/attitude/
behavior
Media Use
Belief/attitude/
behavior
Reinforcing spirals model
Slater (2007, p. 284)
Time 1 Time 2 Time 3
Hypotheses
More selective exposure more ideologically extreme
Less selective exposure less ideologically extreme
Research questions
Selective exposure over time?
Role of political interest?
RQ1
RQ2
H1
H2
Method
• Panel survey during five months in the 2014 EU and Swedish national election: ”super election year”
• Random sample
• 2,281 (33%) completed all waves
Wave 1 Wave 2 Wave 3 Wave 4
May 22–25 September 14May 26–June 4April 11–22 August 1–13 September 15–24
Cross-lagged panel model
Political
Ideologyw1
Political
Ideologyw2
Selective
Exposurew2
Political
Ideologyw3
Political
Ideologyw4
Selective
Exposurew4
Political
interest
Income
.7
.31 1*** .91***
1*** .96***
.03*
.23***
.09***
–.05***
.08***
–.07***
.03**
Selective
Exposurew1
Selective
Exposurew3
Sex
.05**
.04**
Education
Standardized path coefficients.
χ2(6, N=3,557) = 3.67
p = .721
RMSEA < .001
CFI = 1.0
R2 = .107
*p<.05 **p<.01 ***p<.001
Results: mutual reinforcement
Cross-lagged panel model
Political
Ideologyw1
Political
Ideologyw2
Selective
Exposurew2
Political
Ideologyw3
Political
Ideologyw4
Selective
Exposurew4
Political
interest
Income
.7
.31 1*** .91***
1*** .96***
.03*
.23***
.09***
–.05***
.08***
–.07***
.03**
Selective
Exposurew1
Selective
Exposurew3
Sex
.05**
.04**
Education
Standardized path coefficients.
χ2(6, N=3,557) = 3.67
p = .721
RMSEA < .001
CFI = 1.0
R2 = .107
*p<.05 **p<.01 ***p<.001
Results: mutual reinforcement
Cross-lagged panel model
Political
Ideologyw1
Political
Ideologyw2
Selective
Exposurew2
Political
Ideologyw3
Political
Ideologyw4
Selective
Exposurew4
Political
interest
Income
.7
.31 1*** .91***
1*** .96***
χ2(6, N=3,557) = 3.67
p = .721
RMSEA < .001
CFI = 1.0
R2 = .107
*p<.05 **p<.01 ***p<.001
.03*
.23***
.09***
–.05***
.08***
–.07***
.03**
Selective
Exposurew1
Selective
Exposurew3
Sex
.05**
.04**
Education
Standardized path coefficients.
Results: mutual reinforcement
Cross-lagged panel model
Political
Ideologyw1
Political
Ideologyw2
Selective
Exposurew2
Political
Ideologyw3
Political
Ideologyw4
Selective
Exposurew4
Political
interest
Income
.7
.31 1*** .91***
1*** .96***
.03*
.23***
.09***
–.05***
.08***
–.07***
.03**
Selective
Exposurew1
Selective
Exposurew3
Sex
.05**
.04**
Education
Standardized path coefficients.
χ2(6, N=3,557) = 3.67
p = .721
RMSEA < .001
CFI = 1.0
R2 = .107
*p<.05 **p<.01 ***p<.001
Results: mutual reinforcement
Cross-lagged panel model
Political
Ideologyw1
Political
Ideologyw2
Selective
Exposurew2
Political
Ideologyw3
Political
Ideologyw4
Selective
Exposurew4
Political
interest
Income
.7
.31 1*** .91***
1*** .96***
.03*
.23***
.09***
–.05***
.08***
–.07***
.03**
Selective
Exposurew1
Selective
Exposurew3
Sex
.05**
.04**
Education
Standardized path coefficients.
χ2(6, N=3,557) = 3.67
p = .721
RMSEA < .001
CFI = 1.0
R2 = .107
*p<.05 **p<.01 ***p<.001
Results: mutual reinforcement
Conclusions
No support for political ideology and selective exposure mutually reinforcing and becoming more extreme over time.
Individuals exposed to attitude-inconsistent news get slightlyless extreme (but mostly moving toward right).
High political interest: more extreme political ideology, and more exposure to attitude-inconsistent news.
RQ1
RQ2
H1
H2
Peter M. Dahlgrenselective exposure and media effects
@peterdalle
peterdahlgren.com