![Sample our Politics & International Relations journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5947/TOC-Subject-Sample-2021-Politics-International-Relations.jpg"})
Abstract
We develop a simple theory of information spillovers within political advertising: individuals use information conveyed by candidate-specific ads to evaluate other candidates seeking different offices. We test the theory by linking state court of last resort election returns with a complete set of gubernatorial, congressional, and state supreme court political advertisements in 2010. We show that up-ballot advertisements about crime have spillover effects on judicial race outcomes. Specifically, fearful references to crime in non-judicial campaign advertisements decrease judicial incumbent vote share, while enthusiastic references to crime increase incumbent vote share. We estimate that cumulative spillover effects from crime ads in gubernatorial and congressional advertising campaigns in 2010 decreased incumbent justices’ average expected vote share, affecting the outcome of at least one state supreme court election in 2010. We show that ads affect vote choice by providing information to voters, rather than by priming local crime rates. We also conduct two placebo tests to validate our results. Overall, these findings provide evidence of ads’ political effectiveness, while our research design proposes a conceptual framework for understanding advertising effects in context.
Notes
1 Among many others, see Shaw Citation1999, Goldstein and Freedman Citation2002, Huber and Arceneaux Citation2007, Gerber et al. Citation2011, Krasno and Green Citation2008, and Krupnikov Citation2014 for examples of empirical tests of political advertising’s effects.
2 In this article, we refer to ads run for all elections other than the judicial election of interest as “exogenous ads.” This is a convenient shorthand phrase for repeated use in the manuscript. We also argue in the manuscript that the ads are also substantially exogenous in the sense typically used in making causal claims. That said, given shared environments and geographic proximity, it is unlikely that there is absolute independence between ads run for one race and the politics of a judicial race in a similar area.
3 We also find no significant relationship between a DMA’s crime rate and the number of times negative crime ads were aired in that DMA (Pearson’s r = 0.05).
4 WMP provides pre-coded data at the gubernatorial, House, and Senate level. We also estimate the effect of just “crime” focused ads (excluding narcotics and capital punishment) in Appendix A2. Results are consistent across both specifications.
5 This type of coding has an unavoidable subjective element, introducing some inconsistency across coders. Recent data from the WMP (2020) indicates that emotion coding is more variable than other types of coding, though we attempt to mitigate this problem through using a variety of specifications, found in the manuscript and in the appendices.
6 A small number of ads included both enthusiastic and fearful appeals. We ultimately coded these as neutral ads, but our results are robust to a number of specifications for these ads.
7 We present such a model in Appendix A6.
8 Our results are robust to broader (all property and violent crimes) and narrower (just the murder rate, as in Hall Citation2001) definitions of crime rate; see Appendix A5 for details.
9 We find no evidence that crime matters differently for the Texas Supreme Court, which hears non-criminal cases, compared to the Texas Court of Criminal Appeals, which hears criminal cases.
10 We examine the racial dynamics of crime spillover effects in a separate paper.
11 All results are robust to using fractional logistic regression models with a rescaled dependent variable that is bounded by 0 and 1. We use OLS here to aid the immediate interpretability of regression coefficients.
12 Airing ads is expensive; large numbers of ad airings likely represent a better funded campaign, which in turn is likely correlated with a variety of other factors (including candidate quality and political talent) that impact vote share. Thus, it is best to see the total ads variables as generalized measures of electoral resources rather than the effectiveness of judicial candidates’ overall advertising efforts. Notably, this is not the case for our independent variable, since the ads are not purchased or broadcast by the candidates themselves.
13 For example, Charles Johnson defeated the sitting Chief Justice of the Washington Supreme Court, Keith Callow, in a 1990 nonpartisan election despite not even campaigning. At the time (and since), this has largely been seen as a result of low name recognition of the incumbent and (now Justice) Johnson’s name having a commonplace familiarity (London Citation1990).
14 The number of fearful ads broadcast per DMA ranges from zero to 1,606 airings, while enthusiastic crime ads range from 0 to 1,149. The 1606 fearful ad count occurs only in one media market (Chicago) and is approximately double the next highest in the country. This makes it a potential outlier. All model results in this paper are robust to the exclusion of Chicago’s media market. Figure 1 excludes these data points so that the graph better represents the actual distribution of the data.
15 State v. Veliz, 176 Wn.2d 849 (Wash. 2013).
16 State v. Eriksen, 172 Wn.2d 506 (Wash. 2011).
17 State v. Eriksen, 182 Wn.2d 191 (Wash. 2014).
18 Ockletree v. Franciscan Health Sys., 179 Wn.2d 769 (Wash. 2014).
19 Adam Bonica’s (Citation2014) Judicial CFscores rate Sanders a 0.57 and Wiggins a -1.30, with lower scores being more liberal, based on each candidate’s campaign finance history. For comparison, these values correspond closely to prominent national politicians like John McCain (0.68) and Hillary Clinton (−1.16).
20 This is most clearly seen in the case of the Native American police “hot pursuit” question, which had been decided differently as recently as 2010 with Justice Sanders, before being reversed on reconsideration in 2011 with Justice Wiggins casting the deciding vote. For more information, see Pulkkinen (Citation2010, Citation2011).