Transcript Predictors

Social Networks, Campaigns, and Reasoned Candidate Preferences
Introduction
Disagreements with close others shape political decisionmaking, promoting tolerance of opposing attitudes, reducing
political interest and participation, and sometimes inducing
attitude change. (Huckfeldt Mendez, & Osborn, 2004; Mutz,
2006). Work to date has focused primarily on the impact of
social networks on outcomes of such decisions. The present study
investigates the influence of social networks on the predictors of
such outcomes. Does disagreement promote antipathy and
defensive biases—or might it encourage a more thoughtful
approach to candidate choice?
Because people value their relationships with friends and
family, because they are often exposed to close others’ opinions,
and because political disagreements often inspire animated
disputes, we predicted that individuals who disagreed with close
others about politics would be willing and able to critically
evaluate their own candidate preferences.
Hypotheses
If political disagreement promotes more effortful political
decision-making, then disagreement should moderate the
determinants of candidate choice.
Hypothesis 1: Policy preferences will predict changes in
candidate preferences most strongly among individuals who
experience more intense and/or frequent disagreement with
close others.
Hypothesis 2: Party identification will predict changes in
candidate preferences most strongly among individuals who
experience less intense and/or frequent disagreement with
close others.
Results
Effects of Key Predictors on Obama FT Preference (MN Panel)
(t)
b (SE)
Policy Preferences (Conservative)
Party Identification (Republican)
Network Agreement
Policy Preferences X Agreement
Party Identification X Agreement
Lagged DV (Wave 1)
Constant
R2
N
Controls included demographic characteristics (i.e., age, sex, education, income, white vs nonwhite racial
identification) and other potential confounding variables (i.e., political interest, extremity of party
identification, extremity of policy preferences, political knowledge).
Effects of Key Predictorson Obama Vote Probability (ANES Panel)
Method
Policy Preferences X Agreement
2012 Minnesota Panel (Chen et al., 2014)
N = 927
Predictors
• Frequency of network disagreement (3-item scale)
• Policy preferences (4-item government spending composite)
• Party identification (7-point branching scale)
• Feeling thermometer ratings of candidates (October 2012)
Outcome
• Feeling thermometer ratings of candidates (November 2012)
• The determinants of individuals’ candidate preferences
and vote choice vary depending on their social context.
Figure depicts first differences across the full range of network agreement. The Y-axis indicates the marginal
effect of predictors on our feeling thermometer difference score. All controls are held at their means, except
gender, race, and education, which are held at their medians.
Party Identification (Republican)
Network Agreement
Party Identification X Agreement
Vote Intention (October 2008)
Marginal effects of Policy Preferences and
Party ID (with 95% Confidence Intervals)
See below for a list of control variables
Constant
AIC
Percent Correct Predictions
N
 H1: Policy preferences predicted changes in candidate
preferences most strongly among individuals who
experienced relatively frequent disagreement.
Implications
Note. Table entries are OLS coefficients, with t-statistics in parentheses. The outcome is the difference
between respondents’ feeling thermometer ratings of the two candidates (Obama – Romney). All continuous
variables besides age were rescaled to run from 0 to 1, while age is naturally coded in years.
* p<0.05; ** p<0.01
Policy Preferences (Conservative)
Predictions were supported.
 H3: Disagreement also moderated the effects of policy
preferences and party identification on vote choice;
indeed, the pattern of effects is virtually identical.
0.773**
(34.39)
0.330**
(9.02)
0.88
927
(t)
b (SE)
-5.569*
(2.57)
2.642
(1.56)
2.411
(1.55)
6.671*
(2.08)
-9.875**
(3.74)
5.68**
(11.85)
Discussion
 H2: Party identification predicted changes in candidate
preferences most strongly among individuals who
experienced relatively infrequent disagreement
See below for a list of control variables
Hypothesis 3: Disagreement will also moderate the effects of
policy preferences and party identification on vote choice
(controlling for earlier vote intention).
2008-2009 ANES Panel
N = 1,542
Predictors
• Mean disagreement with political discussants (up to 5 items)
• Policy preferences (10-item left-right composite)
• Party identification (7-point branching scale)
• Vote intention (self-reported in October 2008)
Outcome
• Vote choice (self-reported in November 2008)
-0.338**
(5.30)
0.041
(1.14)
-0.011
(0.29)
0.328**
(3.43)
-0.276**
(4.26)
Marginal effects of Policy Preferences and
Party ID (with 95% Confidence Intervals)
Pierce Ekstrom
Brianna Smith
Allison Williams
Hannah Kim
-2.229
(1.53)
350.83
82.62
1,542
Note. Table entries are logistic regression coefficients, with z statistics in parentheses. The outcome is selfreported vote choice in November 2008, dichotomized such that a vote for Obama yields a value of 1 and a
vote for McCain yields a value of 0. All continuous variables besides age were rescaled to run from 0 to 1,
while age is naturally coded in years.
Figure depicts first differences across the full range of network agreement. The Y-axis indicates the
marginal effect of predictors on respondents’ probability of voting for Obama rather than McCain. All
controls are held at their means, except gender, race, and education, which are held at their medians.
Controls included demographic characteristics (i.e., age, sex, education, income, white vs nonwhite racial
identification) and other potential confounding variables (i.e., political interest, extremity of party
identification, extremity of policy preferences, political knowledge).
• Political disagreement may promote more effortful—and
perhaps more normatively desirable—political decisionmaking in presidential elections.
Remaining Questions
Does disagreement cause increased reliance on policy preferences relative
to party identification?
Panel data rules out “reverse cause” explanations for our
effects, but cannot account for all unmeasured variables.
Perhaps the people who select into relatively lowdisagreement networks are also more likely to rely on party
identification over policy. However, we have controlled for
many such “third variables” (e.g., extremity of partisanship,
political knowledge).
Is disagreement a “good thing?”
We contend that policy preferences are a more normatively
desirable basis for political decision-making than party
identification. Thus, disagreement appears to have some
positive consequences.
That said, disagreement also has less desirable effects (e.g.,
negative affect, damage to interpersonal relationships,
reduced political engagement; Mutz, 2002).
It is therefore unclear whether the potential benefits of
disagreement outweigh its potential costs.
References
Chen, P.G., Appleby, J., Borgida, E., Callaghan, T. H., Ekstrom, P. D.,
Farhart, C. E., Housholder, E., Kim, H., Ksiazkiewicz, A., Lavine, H.,
Luttig, M. D., Mohanty, R., Rosenthal, A., Sheagley, G., Smith, B. A.,
Vitriol, J. A., & A. Williams. (2014). The Minnesota Multi-Investigator
2012 Presidential Election Panel Study. Analyses of Social Issues and
Public Policy.
Huckfeldt, R., Mendez, J. M., & Osborn, T. (2004b). Disagreement,
ambivalence, and engagement: The political consequences of
heterogeneous networks. Political Psychology, 25, 65-95.
Mutz, D.C. (2002). The consequences of cross-cutting networks for
political participation. American Journal of Political Science, 46, 838-855.
Mutz, D.C. (2006). Hearing the other side. New York: Cambridge University
Press.