The Nuanced Role of Religious Rhetoric in Presidential Elections
We know that religious rhetoric plays an important role in defining the boundaries of the political community and in setting the tone for political discourse. But we generally do not think of candidates as centrally preoccupied with trailblazing new standards of religious discourse or codifying a particular vision of the American creed. Candidates are, instead, “single minded seekers of reelection” (Mayhew 1974, 5), principally concerned with crafting speech in a way that will garner electoral favor (Jacobs and Shapiro 1994).
Understood this way, the evidence strongly supports the conclusion that emotive and identity-laden cues are persistent elements in religious rhetoric largely because they are effective means to gain electoral support among religious voters. Let us begin by addressing the identity question, examining subgroup, civil religion, and culture wars identities in separate multilevel models. These models allow us to examine how the relationship between candidate evaluation and religiosity changes as a function of religious rhetoric, while controlling for other factors typically thought to play a role in candidate evaluation.2 2 Figure 5.2 summarizes the results, providing evidence that the language of identity influences change in candidate evaluation in complex and varied ways, activating different dimensions of religiosity under different circumstances.
Panel A of figure 5.2 illustrates the interaction between religious commitment in the electorate and civil religion identity rhetoric during a campaign. As the graph illustrates, at high levels of civil religion rhetoric (held at the 75th percentile), religiously committed voters substantially improved their evaluations of candidates. When candidates failed to invoke this rhetorical style (low levels of civil religion rhetoric, held at the 25th percentile), the opposite effect holds—committed voters actually lowered their evaluation of the candidate. This can be thought of as a sort of “sin of omission.” When candidates failed to invoke the language of civil religion identity—the norm in campaign religious rhetoric—religiously committed voters essentially punished them at the polls. Consistent with our expectations, the opposite effect holds when we consider the relationship between religious orthodoxy in the electorate and the language of civil religion identity (panel C). All else being equal, orthodox individuals responded to civil religion rhetoric by actually lowering their opinion of the candidate. When candidates refrained from references to American civil religion, however, the orthodox tended to improve their overall evaluation of them. It is plausible that, whereas nonsectarian religious language appeals to the religiously committed in a very general way, these watered-down nonsectarian versions of religious faith are ultimately threatening to the most orthodox individuals. This is consistent with many theoretical accounts of religious orthodoxy as an orientation toward religious exclusivity. Civil religion rhetoric ultimately exists to paint American spirituality with a broad brush, distilling commonalities across religious worldviews. Accordingly, this rhetorical construction may ultimately threaten identities that define themselves through contrasts with secular society and other religions.
The opposite pattern holds for Christian subgroup rhetoric. All else being equal, the religiously committed tended to have less favorable opinions of candidates who used Christian subgroup rhetoric, and their opinions improved substantially when candidates refrained from subgroup references (panel B). This finding makes sense, especially if candidates tend to have some sort of bias in the particular subgroup that they rhetorically privilege. For example, in his 1984 campaign, Mondale repeatedly made reference to his Methodist faith and upbringing. Although he certainly never suggested that Methodism was the “one true faith” in any sense, this rhetorical strategy may have reminded many committed individuals of their differences with the candidate rather than establishing a point of shared identity.
The interaction between Christian subgroup rhetoric and orthodoxy displayed in panel D is not statistically significant, although for orthodox individuals, the picture does tend to improve slightly when candidates regularly referenced Christian subgroups. These results fit with the
FIGURE 5.2 Religious rhetoric and change in candidate evaluation
Relationship between type of religious rhetoric and postelection feeling thermometer scores, by level of religious commitment (panels A and B) and level of religiosity or orthodoxy (panels C and D). All models control for preelection feeling thermometer scores and a host of control variables. Civil religion rhetoric significantly interacts with religious commitment (panel A) and orthodoxy (panel C) (p < .01). Christian subgroup rhetoric interacts significantly with religious commitment at p < .1 (panel B). The Christian subgroup rhetoric interaction with orthodoxy is not significant (panel D). For full model details, see the online appendix, http://facstaff.uww. edu/chappc/.
expectation that subgroup appeals work very differently than civil religion rhetoric. Unlike rhetoric linking religion to country, references to specific Christian denominations (e.g., Catholicism and Protestantism) and to Christianity tended to lower the most religiously committed individuals’ opinions of candidates. This makes sense. Because religious commitment transcends faith traditions, language privileging a particular tradition may seem threatening to many committed individuals. And, while the candidate evaluations and orthodoxy do not vary significantly as a function of Christian subgroup rhetoric, it is at least possible to conclude that orthodox Christians are not turned off by sectarian rhetoric in the same way the religiously committed are.
Figure 5.2 provides evidence that religious ingroup appeals interact with religious identity in nuanced ways. Interestingly, culture wars rhetoric does not have a significant relationship to any of the measures of individual religious identity under scrutiny. There are several interpretations possible of this nonfinding. At one level, the null finding could simply be a product of the methodological approach adopted in this chapter. I examined only presidential stump speeches, and evidence indicates that under many circumstances presidential candidates avoid going “on the attack”; they relinquish this duty to campaign surrogates (Sigelman and Buell 2003). Thus, the campaigns might have activated religious predispositions through culture wars rhetoric in a way that the present study simply does not observe.
Despite the fact that campaign surrogates might be primarily responsible for waging culture wars, there is still considerable variation in the extent to which presidential candidates used religious language to identify outgroups. It is thus important to grapple with why the data do not show the public reacting to this variation in any significant way. Social psychology provides one good response to this question. Although many have assumed that the construction of social ingroups necessitates the definition of outgroups (Sumner 1906), numerous studies have demonstrated that this is not in fact the case—ingroup identification need not go hand in hand with outgroup differentiation (see Brewer 1999 for a review). Although the case of culture wars rhetoric is theoretically consistent with many of the conditions that tend to increase the likelihood of negative reciprocity, there are also reasons to suspect that culture wars rhetoric will not actively differentiate outgroups in the ways its practitioners hope (Brewer 1999). Specifically, Brewer (1999) argues that “concentric loyalties” can mitigate the effects of negative outgroup feelings. As we have already seen, American religiosity is nothing if not a rich tapestry of crosscutting identities. It is certainly possible that the effects of culture wars rhetoric are greatly diminished by the activation of a superordinate identification with American civil religion. In this way, the potentially divisive elements of culture wars rhetoric could be mitigated by the effects of other religious genres.
As with the identity models, I examined the effects of each of my emotion variables separately. The cross-level interaction between individual religiosity and a specific emotive rhetoric indicates the extent to which that rhetorical mode induced an emotional state consistent with a reliance on religious predispositions. The direct effects of an emotive style on the intercept are, in contrast, more consistent with the idea of mood congruent judgment. In other words, if emotive style is influencing the relationship between religiosity and evaluation, this suggests that emotive religious rhetoric is influencing the religious basis of candidate evaluation. If the emotive style impacts individuals regardless of their level of religious commitment, it suggests that religious rhetoric is probably best seen as a vehicle used to express emotions in a more general sense.
Let us begin by exploring the effects of two negative emotions, anxiety and anger. I examined the effects of these rhetorical elements in conjunction with one other feature—references to the candidate’s opponent. Negative emotive rhetoric can be directed at a specific target, or negative emotive rhetoric can be characteristic of a candidate’s style more generally. Angry rhetoric can be directed at a candidate’s opposition, or it can be a more general lamentation about the state of the country. It is important to make a distinction between these rhetorical choices, controlling for emotive rhetoric directed at an opponent from a more general negative emotive style. The anxiety and anger models thus contain control variables for opponent references.
Several results of this analysis stand out. First, religious opponent references are very important to how candidates are evaluated; however, the effects are not necessarily straightforward. Figures 5.3 and 5.4 display the results of the anger and anxiety analyses, isolating the impact of references to the opponent. When we compare the pairs of lines in each model, we can see that religious references to the opponent are generally a useful rhetorical strategy. The lines that correspond to high levels of opponent rhetoric are consistently (and significantly) higher than the low opponent lines, indicating that candidates who “called out” their opponents on religious grounds were, on balance, more favorably evaluated. Interestingly, religious references to the opponent do not appear to play differently for religiously committed versus uncommitted voters, as the small slopes for the solid and dashed lines illustrate. When religious rhetoric about the opponent was kept to a minimum, religiously committed voters tended to evaluate the candidate more favorably than did the nonreligious voters in the same rhetorical context. In contrast, when candidates went on the attack using religious language, there was little difference in how this message was heard by religious versus nonreligious voters. This unexpected finding suggests that, in general, candidates can help their own standing by characterizing their opponent in negative religious terms—and that the this religious appeal will be effective with voters regardless of their religious predispositions. Candidates who refrain from using religious language as an attack device, however, do not pay as steep a price among religious voters as among nonreligious voters. To explain this finding, we may speculate refraining from opponent-centered religious language may be consistent with a “righteous” religious image—an image that might resonate more with religious voters.
FIGURE 5.3 Angry and opponent-directed religious cues
Relationship between religious commitment and postelection candidate evaluation at different levels of angry and opponent-directed rhetoric. The graph displays effects with anger and opponent-directed rhetoric held at the 25th and 75th percentiles. The interaction of opponent- directed rhetoric and religious commitment is significant (p < .01), whereas the interaction with anger is not. For full model details, see the online appendix, http://facstaff.uww.edu/chappc/.
FIGURE 5.4 Anxious and opponent-directed religious cues
Relationship between religious commitment and postelection candidate evaluation at different levels of anxiety and opponent-directed rhetoric. The graph displays effects with anxiety and opponent-directed rhetoric held at the 25th and 75th percentiles. The interaction of opponent- directed rhetoric and religious commitment is significant (p < .01), whereas the interaction with anxiety is not. For full model details, see the online appendix, http://facstaff.uww.edu/chappc/.
Angry appeals, illustrated in figure 5.3, do little to win the hearts of voters, and in fact, there is little difference in how religiously committed and uncommitted voters responded to angry religious messages (the differences in slopes of the lines in figure 5.3 are not statistically significant). Interestingly, candidates who conveyed anger through religious rhetoric actually tended to be evaluated less favorably by the electorate as a whole as a campaign progressed. This is consistent with the concept of affect transfer, whereby the negativity expressed by a candidate is adopted by the message recipients (with both religious and secular voters responding to anger in a similar manner). All this suggests that religious rhetoric is more of a vehicle to convey emotion to a general audience, not a means to reach out to a religious constituency (or possibly that candidates who are losing ground tend to adopt an angrier rhetorical style).
The anxiety-inducing character of religious rhetoric does not correspond with any significant differences in the slopes of the lines graphing candidate evaluation against religious commitment or with the intercepts in the model (figure 5.4). In the case of the slope, this null finding was ex?pected. Anxiety was not expected to activate religious predispositions as an evaluative criterion because it is typically associated with an increased information search, not a reliance on existing predispositions. Nevertheless, it is interesting to note that, whereas an angry mood tends to lower a candidate’s overall standing, anxiety does not have a similar effect.23
Finally, I examined whether enthusiastic religious rhetoric activates religious commitment as an evaluative criteria. The results, displayed in figure 5.5, indicate that enthusiastic religious rhetoric tended to activate religious commitment as an evaluative criteria. This is consistent with the idea that emotive rhetoric can indirectly influence the basis of political evaluation rather than transferring to all message recipients. Rather sedate expressions of religious identity did little to fire up religious voters. In contrast, and consistent with the work of Marcus, Neuman, and MacKuen (2000), religious voters responded to enthusiastic expressions of religious identity by drawing on this identity in the process of candidate evaluation.
FIGURE 5.5 Enthusiastic religious cues
Relationship between religious commitment and postelection candidate evaluation at different levels of enthusiasm. The graph displays the effects with enthusiasm held at the 25th, 50th, and 75th percentiles. Note that, although the intercepts run counter to expectations, this difference is not significant. In contrast, the cross-level interaction between enthusiasm and religious commitment is significant at p < .05. For full model details, see the online appendix, http://facstaff.uww.edu/chappc/.
These results indicate that emotive forms of religious expression have non-negligible consequences on the electorate but that these effects are quite nuanced. Religious rhetoric appears to operate as a vehicle to convey anger, although religiously committed and uncommitted voters respond to this style in similar ways. Enthusiastic religious language, in contrast, appeals to religious predispositions, activating religious commitment as an evaluative criterion, which is consistent with evidence that enthusiasm can activate group-based evaluations.