Dialing the Wrong Number? Robocalls, Messaging, and Minority Republican Turnout

• In Event: Explaining Minority Voter Turnout

Abstract

The relationship between the Republican party and minority voters is fraught, to say the least. Recent work in Latino politics has shown how Donald Trump angered and mobilized Latinos in general (Gutierrez et al. 2019) and Mexican-Americans in particular (García Ríos, Pedraza and Wilcox Archuleta 2018). While Republican support among minorities is very low, there are still prominent examples of minorities who support the Republican Party who might be crucial to winning swing states with increasing minority populations.

In this study we employ a field experiment embedded in robocalls to identified Republican voters in the months preceding the 2016 Presidential Election. Moving beyond the baseline finding that a generic observability message increases voter turnout above and beyond the effect of a GOTV call without the observability language for the aggregate sample, we focus on how this message impacts turnout among Black and Latino voters.

In the experiment we distinguish between increasing the salience of observability – telling
subjects that whether or not one votes can be observed – and descriptive norms – encouraging a certain behavior, in this case voting, by emphasizing its prevalence. These forms of social pressure have been shown to affect voter participation (Gerber et al. 2008).

Our sample involves likely Republican voters who were registered to vote prior to October 15, 2012 in one of 14 states: Arizona, Colorado, Connecticut, Illinois, Iowa, Kansas, Maine, Massachusetts, New Jersey, New Mexico, North Carolina, Ohio, Oklahoma, and Utah. These states were selected to capture a diverse range of political environments. There were
1,759,586 total subjects in 776,803 households in our subject pool.

We used a 2x2 design. The scripts for Treatment P included only the base political messages that were used for all four treatment groups. The scripts for Treatment PN added “descriptive norm” language to the base political messages. The norm language included the phrases, “join with millions of Republicans who are voting,” “Republicans all across the country are turning out to vote,” and “Join your friends and neighbors.” The scripts for Treatment PO added
“observability” language. The observability language included the phrases “While who you vote for is private, whether you vote is public record and your friends and neighbors can learn whether or not you voted,” “Your friends and neighbors will want to know you voted in this election,” and “Make sure the record shows you voted.” The scripts for Treatment PON added both the norm and observability language. We used verified voting data to measure the effectiveness of our treatments by comparing live answerers in the placebo group to live answerers in the treatment groups.

The results for minority respondents differ strikingly from the results for non-Hispanic whites.
Hispanic subjects in our sample (n= 28,093) are about 5 percentage points less likely to vote than the rest of the sample. Using the political-only message decreases that gap to about 3 percentage points (i.e., a statistically significant 2 percentage point treatment effect). However, when we include the norm or observability language, the gap jumps back to 5 percentage points (i.e., no measurable treatment effect). For African American voters (n= 4,139), all the treatments depress turnout except for the observability condition (Treatment PO). Comparatively, the treatment effects for white voters are 0.3 and 0.4 percentage points for treatments PO and PON, respectively: small, but positive and statistically significant (n= 553,093). These differences suggest that there might be a negative effect to targeting minority candidates with the observability messages that otherwise resonate with the larger majority of white voters in the party. In other words, the use of recent GOTV innovations can backfire when a potential voter’s political preferences do not align with the modal preferences for their sociopolitical group, in this case race and ethnicity. This effect seems likely to be exacerbated during a campaign in which the rhetoric and policies raise existential concerns about race and belonging.

In this paper we show that robocalls with different norm messaging can have substantive on turnout. We also show, in ways related to growing literature on the implicit effect of group norms and self-regulating, that increasing the salience of social networks affects the impact that group norms have on voting behavior. We argue that poorly targeted GOTV campaigns might negatively impact the likelihood of activating even ideologically receptive voters when the values of the party do not align with their social group norms.

Authors

• Angel Saavedra Cisneros, St. Norbert College

• Daniel Kling, Belmont Abbey College