Motivated reasoning and democratic accountability

Abstract

<p>Standard political agency models assume voters form accurate beliefs about how politicians are performing. However, substantial evidence from political behavior research indicates that voters have ``directional motives'' beyond accuracy. These results are often taken as evidence that voters will not be able to hold politicians accountable for their actions. We probe this conclusion by formalizing a model of accountability where voters form beliefs with both directional and accuracy motives. We classify the effects of directional motives into (1) divergence, which causes people with different preferences to hold different beliefs about relevant facts, and (2) desensitization, which leads to a weaker relationship between incumbent performance and voter beliefs. We show that divergence has an ambiguous impact on politician behavior but desensitization uniformly decreases incentives to perform well in office. A key implication of our analysis is that motivated reasoning generally does harm democratic accountability, but this is not because partisans get more polarized, but because all voters become less sensitive to changes in incumbent performance. We also explore the implications for empirical work which examines the relationship between performance indicators and incumbent vote shares. While directional motivated reasoning always weakens the strength of this relationship, we can't necessarily infer that accountability is diminished.</p>

How to Distinguish Motivated Reasoning from Bayesian Updating

“…23 One question frequently studied experimentally is whether information provision affects voting behavior (e.g., Dunning et al, 2019), often with a focus on how partisan motivated reasoning could weaken this relationship (e.g., Adida et al, 2017). Trying to determine whether results from these experiments are driven by directional motives leads to a similar observational equivalence problem studied here (Little, Schnakenberg and Turner, 2020).…”

“…6 See Little, Schnakenberg and Turner (2020) for an application of this belief model to democratic accountability with some discussion of equivalence to standard models in this context. 7 Thaler (2020) builds on Mayraz (2019) with an experimental test, described in more detail in section 7.3.…”

“…Little, Schnakenberg and Turner (2020) build on the related approach inLittle (2019), focused in a setting where there is little loss to having voters pick out a single conclusion since they vote to re-elect an incumbent if and only if they think her competence is above a threshold. The formalization in this paper allows for actors to retain complete belief distributions, making it easier to port to other canonical models.…”

The promise and pitfalls of cross-partisan conversations for reducing affective polarization: Evidence from randomized experiments

“…Inspired by the worry that affective polarization would “weaken...willingness to punish one’s own party’s politicians” [( 5 ), p. 50] for taking incongruent positions, we asked participants which candidates they would vote for, and measured whether participants assigned to the Perfect Day condition would be more likely to select the candidate aligned with their issue views instead of their party. In Little et al ’s ( 45 ) framework, this measures both divergence and desensitization. All our remaining outcomes in this area were preregistered as secondary outcomes unless otherwise noted.…”

The short-term, circumscribed, and conditional effects of cross-partisan conversation

“…Participants' issue views were measured pre-treatment, before the conversation began.8 InLittle, Schnakenberg and Turner's (2020) framework, this measures both divergence and desensitization.…”