ISSUE OWNERSHIP AND WHY LATINO CANDIDATES MIGHT OWN IMMIGRATION

Existing research has made clear that the public often perceives differences between political parties (and their members) when it comes to the ability to address specific issues, handle specific problems, or embody specific traits (Ansolabehere and Iyengar Reference Ansolabehere and Iyengar1994; Belanger Reference Belanger2003; Budge and Farlie Reference Budge, Farlie, Daalder and Mair1983; Doherty Reference Doherty2008; Dulio and Trumbore Reference Dulio and Trumbore2009; Goble and Holm Reference Goble and Holm2009; Hayes Reference Hayes2005; Holian Reference Holian2004; Petrocik Reference Petrocik1996; Petrocik, Benoit, and Hansen Reference Petrocik, Benoit and Hansen2003; Pope and Woon Reference Pope and Woon2009; Stubager and Slothuus Reference Stubager and Slothuus2013; Walgrave, Lefevere, and Nuytemans Reference Walgrave, Lefevere and Nuytemans2009; Walgrave, Lefevere, and Tresch Reference Walgrave, Lefevere and Tresch2012). This concept is referred to in the literature as issue ownership. As Petrocik (Reference Petrocik1996, 826) describes in his groundbreaking analysis of issues and presidential elections, issue ownership is built upon a “history of attention, initiative, and innovation” on the part of one party when compared with another. Based upon existing issue ownership, strategic candidates for office take note of where their party's strengths lie and attempt to maximize on said strengths. Issue ownership provides ammunition for candidates from the day they become their party's nominee; it gives candidates the “perception of credibility” (Damore Reference Damore2004, 392) and links them with “superior policies and programs” (Petrocik, Benoit, and Hansen Reference Petrocik, Benoit and Hansen2003, 601) in the eyes of many individuals.

While most research on issue ownership has been limited to which parties own which issues, a smaller body has noted how different types of candidates, regardless of party affiliation, might be able to own issues to their electoral advantage. Kaufmann (Reference Kaufmann2004), in her examination of the issue ownership concept, notes that candidates are more easily able to make use of issues on which they have an existing record of performance. More broadly than just issue ownership, Sellers (Reference Sellers1998) addresses the concept of candidate backgrounds being a useful tool on the campaign trail. Related literature (some of which directly uses the concept of issue ownership) has also moved toward candidate demographic traits (such as gender) and the strategic use of issues to build support (Alexander and Andersen Reference Alexander and Andersen1993; Fridkin and Kenney Reference Fridkin and Kenney2009; Herrnson, Lay, and Stokes Reference Herrnson, Lay and Stokes2003; Huddy and Terkildsen Reference Huddy and Terkildsen1993; Israel-Trummel Reference Israel-Trummeln.d.; Kahn Reference Kahn1992; Koch Reference Koch1999; Lawless Reference Lawless2004). Herrnson, Lay, and Stokes (Reference Herrnson, Lay and Stokes2003), for example, have shown that female candidates benefit from highlighting “compassion” issues (i.e., education, poverty alleviation, income redistribution). Gender can serve as an asset, especially for female candidates who can successfully leverage issues “owned” by women. Despite these findings on gender, we still know very little concerning the extent to which issue ownership might affect candidates of other demographic subgroups, including Latino candidates.

Much as Walgrave, Lefevere, and Tresch (Reference Walgrave, Lefevere and Tresch2012) find an “associative dimension” of issue ownership (wherein a spontaneous link exists) between political parties and issues in the minds of citizens, we argue a similar link exists between Latino candidates and the issue of immigration. This link springs forth from the experiential and statistical associations between Latinos and immigration built up over time in the minds of Americans. Latino immigrants currently represent the largest group of immigrants in the United States (U.S. Census 2010). Likewise, Latinos are more likely to know or have been related to recent immigrants to the United States, and are more likely than other ethnic groups to identify as immigrants themselves (Barreto Reference Barreto2007a; Reference Barreto, Espino, Leal and Meier2007b; Citrin et al. Reference Citrin, Green, Muste and Wong1997; Pew 2009). As a result, there is a strong tendency for Americans to organize their thoughts about Latinos alongside the issue of immigration, with recent evidence suggesting that Americans sometimes view immigrants and Latinos as synonymous groups (Masuoka and Junn Reference Masuoka and Junn2013; Ramakrishnan et al. Reference Ramakrishnan, Esterling and Neblo2009). As a result of these findings, an existing relationship with the issue of immigration becomes, as Sigelman et al. (Reference Sigelman, Sigelman, Walkosz and Nitz1995) might say, an “assumed characteristic” of Latino candidates. Latino candidates who discuss immigration should be able to “activate existing predispositions” (Mendelberg Reference Mendelberg2001, 121) about the link between this ethnic group and this issue. Talking about immigration primes these existing associations in the minds of citizens for the benefit of Latinos on the campaign trail.

Existing theories of cue-based processing provide additional useful frameworks to support our argument, as they predict a positive relationship between who a candidate is, his or her skill set, and the likelihood that an individual expresses favorable opinions about that candidate. Candidates strongly and positively associated with an issue in the minds of the public should be able to persuade individuals to accept their abilities on the issue (i.e., adopting favorable opinions of a candidate), potentially generating a state of issue ownership over time, whereas non-credible “issue trespassing” candidates are unlikely to be perceived as well in this regard (Iyengar and Valentino Reference Iyengar, Valentino, Lupia, McCubbins and Popkin2000). This argument is compatible with past studies, which have shown that across many contexts, messenger characteristics (including those linked with race) tend to be more persuasive in swaying individuals than the content of the political messages they espouse (Arceneaux and Kolodny Reference Arceneaux and Kolodny2007; Iyengar and Valentino Reference Iyengar, Valentino, Lupia, McCubbins and Popkin2000; McDermott Reference McDermott1997; Reference McDermott1998; Nelson and Garst Reference Nelson and Garst2005; Rahn Reference Rahn1993). The source of each message delivered during the campaign, or “who said it,” has great potential to explain public opinion in the United States since Americans rate low in political knowledge, display lackluster information processing tendencies, lack the motivation to process the overwhelming information environment of modern day campaigns, and at times struggle to determine how to evaluate candidates (Converse Reference Converse and Apter1964; Reference Converse, Ferejohn and Kuklinski1990; Delli Carpini and Keeter Reference Delli Carpini and Keeter1996; Gilens Reference Gilens2001; Lau and Redlawsk Reference Lau and Redlawsk2001; Zaller and Feldman Reference Zaller and Feldman1992). As such, Americans often rely upon source cues or traits (such as a messenger's ethnicity, race, or gender) because they lack the ability to comprehensively analyze the costs and benefits associated with each candidate to come to a fully informed decision (Converse Reference Converse and Apter1964; Hayes Reference Hayes2010; Hovland et al. Reference Hovland, Janis and Kelley1953; Koch Reference Koch2000; Kuklinski and Hurley Reference Kuklinski and Hurley1994; Lau and Redlawsk Reference Lau and Redlawsk2001; Lodge, McGraw, and Stroh Reference Lodge, McGraw and Stroh1989; Rahn Reference Rahn1993). Source cues such as the name of a candidate might be especially important in low-information electoral or decision-making contexts (Butler and Broockman Reference Butler and Broockman2011; Matson and Fine Reference Matson and Fine2006; McDermott Reference McDermott1997; Reference McDermott1998).

Given the link between Latinos and immigration in the minds of Americans, we expect to see a positive and significant effect of a candidate's Latino ethnicity upon trait scores—trustworthiness on immigration, strength of leadership on immigration, qualification to discuss immigration, and experience on immigration—as compared with an Anglo candidate, ceteris paribus. Furthermore, given the aforementioned literature on messengers trumping messages, we inquire into whether this effect may exist for multiple types of potential messages. We expect Latino candidates voicing “pro-immigration” stances to be rated higher than Anglo candidates who voice “pro-immigration” stances, but might Latino candidates expressing “anti-immigration” stances be rated higher than Anglo candidates who voice the same “anti-immigration” stances as well? In other words, is the reach of issue ownership by Latino candidates strong enough to encompass both sides (“pro” or “anti”) of the issue itself?

Existing findings are inconsistent on this question of whether issue ownership holds across issue position. Research in race and politics suggests that “pro-immigration” messages might provide a level of issue congruence with “assumed characteristics” of Latino candidates, which would play favorably among the public, in a way that “anti-immigration” messages might not (Sigelman et al. Reference Sigelman, Sigelman, Walkosz and Nitz1995). This would mean we should expect Latino candidates voicing “pro-immigration” messages to receive the most positive evaluations on their relationship with the issue of immigration. The evidence from the related gender and politics research, however, concludes that candidate characteristics hold sway regardless of issue position. Since the evidence here is mixed, we expect overall that immigration is an issue that the public associates favorably with Latino candidates, much like compassion issues are ones that the public associates favorably with female candidates. However, we formulate no expectations about whether both directions of the Latino candidate's message (“pro” or “anti”) should reflect the advantage gained by the candidate's ethnicity. Our research design will allow us to test these expectations as well as answer these questions.

EXPERIMENTAL PROCEDURE

In order to test whether candidate ethnicity serves as a cue that affects attitudes in this issue area, we conducted an original survey experiment. The subjects were 350 students in introductory-level American Government and Politics courses and upper-level Political Science courses at a southern university in late 2012.Footnote ¹ Students were told they were participating in an exercise in which they would evaluate congressional candidates appearing at a public forum. After signing consent forms, students began by filling out a short pre-treatment questionnaire (the details of which can be found in Appendix A).Footnote ²

Following completion of the pre-treatment questionnaire, participants read a simulated Associated Press news story on a fictional state senator who was running as the Democratic nominee for the U.S. House of Representatives to represent Missouri's 6th congressional district. Examples of these stories can be found in Appendix B. The general premise of the story (across all treatments) was that this candidate was appearing at a public forum at the University of Missouri-Kansas City to talk about important issues, but the focus of the story was the candidate's opinions on undocumented immigrants.Footnote ^{3 ,} Footnote ⁴

The story participants received, however, varied from one participant to the next in two distinct ways: the name of the candidate and the message from the candidate. Some individuals received a story in which the name of the candidate was Luis Ramírez, while others received a story in which the name of the candidate was Robert Rooney. Such names were chosen to attempt to send participants clear signals that the former candidate was Latino while the latter was Anglo.Footnote ⁵ Such choices are in keeping with a breadth of research employing similar experimental designs (using a name to signal ethnic identity) to examine the impact of minimal cues on evaluations (Abrajano, Nagler, and Alvarez Reference Abrajano, Nagler and Alvarez2005; Bertrand and Mullainathan Reference Bertrand and Mullainathan2004; Butler and Broockman Reference Butler and Broockman2011; Carlsson and Rooth Reference Carlsson and Rooth2007; Fryer and Levitt Reference Fryer and Levitt2004; Holbrook, Fessler, and Navarrete Reference Holbrook, Fessler and Navarrete2016; Oreopoulos Reference Oreopoulos2011; Riach and Rich Reference Riach and Rich2002). The control group received a more general story in which the candidate had no name.

As for the candidate's message, some individuals participating received a story in which their candidate espoused the positive impact that undocumented immigrants have on American society, while others taking part received a story in which their candidate spoke of the negative impact of undocumented immigrants. Those receiving a story on the positive impact of undocumented immigrants read about how their candidate believed more visas should be granted, that undocumented immigrants take jobs Americans don't want, and that current policies block paths to citizenship and higher education. Those receiving a story on the negative impact of undocumented immigrants read about how their candidate believed that the immigration status of workers should be checked, that undocumented immigrants were taking jobs Missourians wanted, and that the children of undocumented immigrants caused problems related to tuition rates.Footnote ⁶ Participants in the control group received no positive or negative information about undocumented immigration and were just told that the nameless candidate addressed immigration policy at the forum. Each story mentions three times (once in each of the story's three paragraphs) that the candidate is the Democratic candidate/nominee in this congressional race; by holding party affiliation constant across all treatments, respondents could focus on the candidate's name and message on immigration. The manipulation of the candidate's ethnicity and message leaves us with five treatments to which participants were randomly assigned: Latino/Pro-Immigration, Latino/Anti-Immigration, Anglo/Pro-Immigration, Anglo/Anti-Immigration, and Control.

After reading the story, participants answered a post-treatment questionnaire asking for assessments of the candidate about whom they just read. This questionnaire asked respondents to evaluate general traits of the candidate in the story. The questionnaire then asked respondents to assess the candidate's strengths and weaknesses specifically related to the issue of immigration. We use four trait measures from the post-treatment questionnaire as the main dependent variables for our analysis: a candidate's experience on immigration, qualification to discuss immigration, trustworthiness on the issue of immigration, and strength to lead on immigration legislation.Footnote ⁷ Respondents rated the candidate in their treatment on these traits using a five-point Likert scale. The questions from this post-treatment questionnaire that comprise our dependent variables can be found in Appendix C.

FINDINGS

Our analyses of the four immigration-related traits just discussed are presented in two empirical sections, which ultimately will demonstrate: (1) whether and to what degree candidate evaluations differ across treatments that vary by candidate ethnicity and message content, and (2) what the main effect is (as well as its size) driving candidate assessments once we control for alternative explanatory variables. In the latter portion of our analysis, we also present evidence that supports the claim that Americans, even Anglo Americans who should prefer Anglo candidates according to psychological theories (Allport Reference Allport1954; Brewer Reference Brewer1999; Conover Reference Conover1984; Tajfel Reference Tajfel, Turner and Giles1981), view Latino candidates as owners of the immigration issue. More generally, the findings indicate that, in the context of a campaign, both messenger ethnicity and message content have the power to shape respondent assessments of candidates on the issue of immigration.

A comparison of favorable evaluations by candidate ethnicity in Figure 1 first shows a consistent trend, where experimental participants rated Latino candidates more favorably than Anglo candidates on the issue of immigration. This occurred regardless of which issue trait we analyzed. For example, 67% of participants in the Latino condition rated the candidate favorably on strength to lead, compared with just 53% of participants in the Anglo condition. The trend of assigning favorable evaluations to Latino candidates repeats for the measure about candidate qualification, which yielded a 74% versus 51% gap between Latino and Anglo treatment groups. The Latino/Anglo evaluation gap persisted on the experience measure (69% versus 36%) as well as for the trustworthy evaluation, where 67% versus 46% expressed favorable candidate evaluations. Figure 1’s findings overall indicate a clear trend of favorable evaluations for Latino candidates over Anglo candidates on the issue of immigration.Footnote ⁸ To further investigate the significance of ethnic cues and message content, we conducted analysis-of-variance (ANOVA) comparisons, which corroborate the statistically significant relationship between the candidate ethnic cue and respondent evaluations.Footnote ⁹

Notes: Proportion of favorable candidate evaluations with 95% confidence intervals across traits related to handling of immigration. Participants offer higher evaluations for the Latino candidate across all traits.

Figure 1. Favorable trait evaluations, by candidate ethnicity

The ANOVA results show that when exposed to treatments in which a Latino candidate delivered a pro-immigration message, respondents rated the candidate significantly more positively than those who received a treatment in which an Anglo candidate delivered the very same pro-immigration message. For the trait of strength of leadership on immigration, there was a mean score of 3.75 out of 5 assigned to the candidate by respondents in the Latino candidate group compared with a 3.36 assigned by respondents of the Anglo candidate group (F _{(1, 145)} = 6.57, p = .001).Footnote ¹⁰ This significant effect was similar for respondent ratings of candidate qualifications to discuss the issue of immigration, which resulted in a mean score of 3.85 for the Latino candidate group, compared with the 3.51 score assigned to the candidate by respondents in the Anglo candidate group (F _{(1, 145)} = 5.95, p = .016). On the trait measure for candidate experience on immigration, respondents in the Latino treatment group rated the candidate as 3.74 out of a possible 5, which was significantly higher than the 3.16 that respondents of the Anglo candidate group (F _{(1, 145)} = 13.75, p = .0003). Slightly less clear are attitudes related to trustworthiness on immigration, with a mean score of 3.64 on this trait for those in the Latino candidate group, versus 3.36 for those in the Anglo treatment group (F _{(1, 145)} = 3.27, p = .073). While these differences vary in size, with some appearing smaller than others, the comparisons from the ANOVA analyses ultimately result in high levels of statistical significance, with one case producing near-significant results.

Substantively, these differences illustrate that respondents receiving pro-immigration messages from an Anglo candidate were likely to offer mid-range, or neutral, evaluations for all the traits being examined. This outcome seems reasonable, since respondents knew very little about the candidate's immigration stance other than what can be gleaned from a short story. We would expect respondents to assign neither overly positive nor negative evaluations about this fictional candidate's ability to handle the issue of immigration based on this information. Respondents that were exposed to our Latino candidate treatment with the same exact pro-immigration message, however, offered evaluations that were distinctly more positive about the candidate's relationship with the issue of immigration.

In comparing Latino candidates and Anglo candidates delivering anti-immigration messages, we do at least find one significant difference in respondent assessments of traits. That difference lies in the extent to which the respondents found the candidates qualified to discuss immigration. The Latino candidate delivering an anti-immigration message received an average score of 3.31 out of 5 when it comes to qualification to discuss immigration, while the Anglo candidate delivering the exact same message received an average of 2.99 (F _{(1, 145)} = 3.99, p = .048). The remaining results of the anti-immigration comparisons continue to demonstrate consistently higher mean scores for Latino candidates over Anglo candidates, despite their statistical insignificance. For example, evaluations resulted in mean scores of 3.44 for the Latino candidate versus 3.22 for the Anglo candidate on the trait of strength of leadership (F _{(1, 143)} = 2.02, p = .157), 3.13 for the Latino candidate versus 3.07 for the Anglo candidate on the trait of trustworthiness on immigration (F _{(1, 143)} = .14, p = .70), and 3.06 for the Latino candidate versus 2.86 for the Anglo candidate on the trait of experience on immigration (F _{(1, 143)} = 1.31, p = .25).

We can infer from the significant findings (given that the only difference across pro-immigration messages was the name of the candidate) that the ethnic cue, as it turns out, was important information in the formation of these evaluations. However, unlike the typical scenario where that ethnic information is used to the detriment of minority candidates, in this case, the ethnic minority candidate benefits from this information. When it comes to the population at large, Latino candidates likely derive benefits from their ethnicity when delivering pro-immigration messages. In this context, there is ownership of this issue based on name alone. All of the evidence from the pro-immigration ANOVA tests, considered together, ultimately suggests the existence of a Latino issue ownership effect.

Next, we employ multivariate empirics to investigate the Latino issue ownership effect controlling for a host of relevant covariates. We further incorporate tests of the claims that pro-immigration and anti-immigration messages could be linked to higher evaluations of candidates. While we have some reason to suspect one type of message will lead to positive evaluations, the literature supports the association of certain messages with candidate evaluations (i.e., Druckman Reference Druckman2004; Fridkin and Kenney Reference Fridkin and Kenney2009), which is why we now turn our focus towards a test of these hypothetical claims. In addition, we include a more thorough examination of whether Anglo respondents differ substantially from the full sample of respondents (Abosch, Barreto, and Woods Reference Abosch, Barreto, Woods and Henderson2007; Barreto Reference Barreto2007a; Reference Barreto, Espino, Leal and Meier2007b; Reference Barreto2010; Barreto, Segura, and Woods Reference Barreto, Segura and Woods2004; Dawson Reference Dawson1994; Tate Reference Tate1993). We focus on the same four traits of strength of leadership, qualification to discuss, trustworthiness, and experience on immigration. We also include individuals who received the control treatment into the analysis. Due to the categorical nature of the dependent variables, we employed ordered probit models. Finally, for ease of interpreting the coefficients, predicted probabilities were generated to get a better picture of the size of the effects in respondent evaluations when moving from the Anglo candidate to the Latino candidate.

The multivariate analysis corroborates the findings of the bivariate analyses in terms of candidate ethnicity effects—Latinos continue to “own” the issue of immigration, for all traits. Table 1 illustrates that the Latino treatment is a highly significant predictor of positive candidate evaluations on immigration, holding all else equal. The Latino treatment retains significance when controlling for respondent partisanship, sex, income, identification as Latino, and prior immigration attitudes regarding deportation versus applying for legal status. This is found across all candidate trait assessments when the full sample of respondents is analyzed. The results also hold when we examine the responses of only the Anglo respondents involved in our experiment. This means that Anglo respondents respond just as positively to the Latino candidate as the entire respondent pool. Interestingly, the pro-immigration treatment also results in a significant finding, suggesting that exposure to statements in support of more lenient positions on immigration led respondents to rate candidates more positively.

Table 1. Multivariate analysis of candidate ethnicity and message on immigration-related traits

Note: resp. = respondents. An ordered probit analysis of respondent assessment of the candidate traits specifically as they are related to the candidate on the issue of immigration: “strong leader,” “qualified to discuss,” “trusted,” and “experienced,” on a five-point scale with 1 as low scores and 5 as high scores on each trait. The first column under each trait presents the full sample of respondents, while the second column for each trait limits the analysis to Anglo respondents only.

*p < .05.

+p < .10.

How might we explain the significant effects of pro-immigration messages on positive evaluations? Since our models controlled for partisanship and prior immigration attitudes, respondents most likely are not using their own personal biases to project their political or personal preferences into the attitude formation process, meaning the message and evaluation are not linked because respondents approve of the particular message being given. We also know that this effect is not explained by whether the respondent is female, has a certain income, or identifies as Latino, since our models also control for all of these measures. This puzzle led us to take a deeper look into the data to determine if there were any more specific treatment group effects driving the significant findings of these message effects. We therefore examined the predicted probabilities based on the ordered probit analysis in Table 1, to determine the size of each factor's effect, and also to examine whether these factors’ ability to affect candidate evaluations were conditional upon each other. Looking at the predicted probabilities can tell us whether a candidate needs to be both Latino and hold pro-immigration messages to benefit from the issue of immigration, or whether each factor on its own can result in a net advantage when candidates discuss the issue of immigration.

The predicted probabilities presented in Figure 2 suggest that both candidate ethnicity and message content have a strong independent influence on a candidate's probability of receiving positive evaluations on the issue of immigration. These probabilities are based on the likelihood that a typical respondent of our survey will voice an evaluation that is more positive than negative or neutral, while holding all control variables at their mean values. Figure 2 compares the varying probability that a typical respondent of our survey voices a more positive evaluation when exposed to the four different treatments. Probabilities range from 0, meaning there is no chance that the respondent will voice a positive evaluation, to 1, meaning there is complete certainty that the respondent will voice a positive evaluation. A probability of .50 means that the respondent is equally as likely to voice a positive evaluation as a negative evaluation; the probability of a positive evaluation is therefore indistinguishable from a random event. The predicted probabilities demonstrate overall that a Latino candidate can benefit from the association individuals make with Latinos and the issue of immigration, resulting in more positive evaluations, and also that Latino candidates and Anglo candidates both stand to benefit if they simply voice pro-immigration stances.

Notes: Probabilities were generated by holding all other variables at their mean values. The different column sizes illustrate the change in the probability that an average survey respondent would express a positive—as opposed to negative or neutral—candidate evaluation, when moving across the four different treatment groups. These probabilities were generated using the full sample of survey respondents.

Figure 2. Predicted probabilities of positive trait evaluations, by candidate ethnicity and message delivered

The figure illustrates these effects, first, by showing that all Latino treatments resulted in an increase in the probability of a positive score within a range of .10–.15, regardless of message content. A “positive score,” is defined as a respondent rating a candidate more positively than negatively (i.e., “strongly agree and agree” versus “disagree and strongly disagree” with a positive candidate evaluation). In the pro-immigration treatments, the probability of receiving a positive score goes from probabilities near .50 of receiving a positive score for the Anglo candidate, to a .60 probability of receiving a positive score if that candidate stays the same in every respect except for now being portrayed as a Latino candidate. The same sized increase in probability is uncovered when examining within the anti-immigration treatments. Moving from the Anglo candidate treatments to the Latino candidate treatments results in a corresponding shift in probabilities of a positive score from roughly .40 to .50. Again, these differences result in a consistent gain in probability of .10 just by replacing an Anglo candidate with a Latino candidate. The figure, however, does not solely indicate a Latino ownership effect. It also illustrates a similar advantage from pro-immigration message content, with a consistent positive difference in the probability of positive scores (an average increase >.10) when respondents are exposed to pro-immigration stances, regardless of candidate ethnicity. For Anglo candidate treatments, this increase starts with probabilities in the range of .37–.44 of a positive score when the anti-immigration stance is delivered and shifts up a range of .44–.55 when the message switches to a pro-immigration stance. For Latino candidate treatments, Figure 2 exposes a similar increase in probability levels, although starting at a higher probability range (.47–.52) of a positive score when anti-immigration stances were delivered, increasing to the highest probability range of the entire experiment, .57–.64, once a pro-immigration stance is presented. Just as changing the ethnicity of the candidate from Anglo to Latino resulted in a consistent advantage in receiving a positive score by a .10 probability, regardless of message content, pro-immigration messages resulted in over a .10 probability advantage over anti-immigration messages, regardless of ethnicity. These findings suggest that both Latino ethnicity and pro-immigration message content resulted in independent positive effects on respondent evaluations of the candidate.

When employed together, Latino ethnicity and pro-immigration stances therefore increased candidate evaluations on the issue of immigration by a probability of just under .20. This would suggest that both types of information are able to influence candidate evaluations in a significant and substantial way. When both are present, the probability of scoring high marks on the issue of immigration is dramatically higher than when both are absent from a candidacy. Importantly, these findings indicate that a candidate's ability to achieve higher evaluations on the issue of immigration is not conditional upon both factors being present. However, it certainly helps a candidate's chances of being perceived positively on the issue when both factors are present.

The results reveal that the Latino and pro-Immigration treatments outperformed all other explanations in an ordered probit analysis. Upon further analysis, it was also shown that these factors were independently exerting an influence on candidate assessments. Taken together, we can surmise that both the source of the message and the message content matter, when determining how a candidate will be evaluated on the issue of immigration. While a candidate can benefit from one factor on its own, the findings reveal that when the right mixture of messenger ethnicity (e.g., Latino candidates) and message content are present (e.g., pro-immigration messages), they can produce dramatically higher evaluations of candidates on the issue of immigration, the order of magnitude of which is an increase of a .30 probability of a candidate receiving a positive evaluation in this issue area.