This is an open access article distributed under the terms of the Creative Commons
Attribution License (“I’ve been talking about negatives, and you’re up on him!” said an astounded [Republican pollster] Frank Luntz. “That’s the story of Trump’s poll numbers.” (From Guo, 2015)
Controversial United States President Donald Trump presents a bit of a puzzle. Since the beginning of his candidacy during the election season, Trump’s poll numbers suggested that a lot of people did not like him, and yet in actual election results he has garnered surprising levels of support that belie those negatives (Guo, 2015).
This juxtaposition can be seen in Trump’s often alarming use of non-normatively negative language. From his comments about Mexicans being “murderers” and “rapists” (Guo, 2015), to questioning POW John McCain as a war hero (Noble, 2015), to his insult of former GOP rival Carly Fiorina: "Can you imagine that, the face of our next president?... are we serious?" (Cohen, 2017), Trump has shown that he does not conform to typical political language. Yet despite all of his often bracingly insensitive language, Trump won both the Republican Party nomination for president and ultimately the General Election.
One explanation for Trump’s success involves a socio-historical discussion that is essentially exclusive to conservative American politics (see, e.g., Oliver & Rahn, 2016). For example, evidence suggests that increased authoritarianism, increased social dominance, and lower cognitive abilities were all predictive of Trump support during the election (Choma & Hanoch, 2017). Although these dimensions are important in our understanding of Trump support, they may also lead to a dismissal of Trump’s electoral success as a function solely of traits associated with ideologically extreme conservatism in the U.S. And while it is true that his supporters are largely ideologically conservative (e.g., Burnett, 2016; Pew Research Center, 2016; Thompson, 2016), that is not the whole story. In the present article, we expand upon existing social psychological theories of deviance to argue that part of his support came from a source that is not directly ideological in nature: Namely, that the surprising levels of support for Trump were the result of the salience of restrictive communication norms. In order to establish the connection between psychological theory about cultural norms and Donald Trump, we first discuss a theoretical framework on what causes deviance from cultural norms in the first place.
The Backfiring of Cultural Norms
It is practically an axiom in social psychology that cultural norms influence behavior. If there is a cultural norm to drive on the right side of the road, then most people drive on the right side. If there is a cultural norm to drive on the left side, then most people become left-side drivers. And when we shift cultural contexts, we frequently learn which side of the cultural road to drive on – and drive accordingly. However, it is clear that people do not always follow cultural norms, and an emerging, theory-driven literature has sought to more clearly illuminate those conditions under which cultural norms do and do not induce compliance.
One theoretical perspective focuses on the sometimes ironic consequences of top-down, heavy-handed norms. This communication perspective suggests that while top-down cultural norms may often produce short-term compliance, the communication of heavy-handed norms sometimes simultaneously produces a sense of forced artificiality that causes them to ultimately backfire and instead produce long-term deviance (see, e.g., Conway & Schaller, 2005; Conway & Schaller, 2007; Conway et al., 2009). Thus, the forced consensus that a cultural norm produces can be like a pane of glass with a small, barely perceptible crack: As long as all else remains equal, it will remain as a single piece of glass – but a little change in context or pressure, and the glass pane will break (see Conway & Schaller, 2005; Conway et al., 2009).
Why Do Cultural Norms Backfire? Reactance and Informational Contamination
What specifically causes the cracks on the pane of cultural glass? Prior research and theory suggests that this can be for two separate reasons. (1) First is emotional reactance (e.g., Brehm, 1966; Crawford, McConnell, Lewis, & Sherman, 2002; Fuegen & Brehm, 2004). Freedom of choice is a valued psychological commodity, and so we often will deviate from others’ expectations—both in belief and in action—in order to re-assert our right to choose. This feeling may be suppressed in the short-term to avoid embarrassment or public sanction; but in the long-term, any reason (such as a surprisingly-popular candidate who is not politically correct) to express it may draw it out. (2) Second is informational contamination (Conway & Schaller, 2005; Conway et al., 2009; see also Fein et al., 1990; Harkins & Petty, 1987; Maio & Esses, 1998). A heavy-handed norm doesn’t just produce a strong emotional response – it also cognitively undermines the informational value of any emergent consensus by causing observers to attribute the consensus to the coercion induced by the norm instead of the potential information contained in the observed behavior.
Thus, a heavy-handed norm may induce short term compliance, but it simultaneously sows the seeds of its own later undoing: It makes people feel bad about the norm emotionally, and (independent of that) it cognitively contaminates the informational value that the norm otherwise might have produced.
Restrictive Communication Norms and Donald Trump
Viewed from this norm communication perspective, Donald Trump is not the cause of cultural deviance – rather, support for him is (in part) the product of the salience of restrictive communication norms. To illustrate, we discuss the specific set of communication norms in question and Donald Trump’s overlap with those norms.
The Backfiring of Positively-Aimed Communication Norms
An extremely powerful set of norms in North American society, often called “political correctness” (or “PC”) norms, explicitly attempts to remove negative group-relevant language (see Conway et al., 2009). As a result, in situations where the norms are in evidence, they create particularly strong public pressure to restrict one’s communication. Although sometimes derided, few academics would disagree that the practical goal of the political correctness movement is well-aimed.
One of the distinguishing features of modern theories of norm backfiring is that they do not require a norm to have a negative aim itself in order to ultimately show negative effects. Indeed, some research suggests that over-salience of PC norms may actually undermine its positive goal and produce more negative communication in the long-term. In one set of research studies involving fictitious scenarios, the stated presence of a heavy-handed PC norm caused participants to later report that they would communicate more negatively about a stereotyped fraternity to a fictional “friend” in the scenarios (Conway et al., 2009; see also Conway & Schaller, 2005).
Donald Trump as Cultural Revolt Against Restrictive Communication Norms
“It’s not just that Trump is willing to be provocative – he’s exciting to many people because he says things they feel they can’t say.” (From Guo, 2015)
So there is reason to suspect that consistent salience of PC norms might cause a crack on the cultural pane of glass. Is there reason to suspect that Donald Trump might be a product of that crack? Yes. As the above quote suggests, it may be in part because of his politically incorrect rhetoric that he garnered support. At a general level, evidence suggests that Trump’s grandiose rhetorical style was one of the reasons he won the Republican primary (Ahmadian, Azarshahi, & Paulhus, 2017). Further and more specifically, polls from the election cycle suggested that people liked his provocative language (Guo, 2015; Thompson, 2016) and that feeling voiceless better predicted Trump support than multiple other variables, some of which include age, race, and attitudes towards Muslims, illegal immigrants, and Hispanics (Thompson, 2016). And indeed, Trump himself has publically recognized the value of this anti-communication norm stance. As he said at the Republican Primary debate in August 2015: "I think the big problem this country has is being politically correct. I’ve been challenged by so many people and I don’t, frankly, have time for total political correctness” (quoted in Guo, 2015).
The Present Research
In summary, to date evidence exists that (1) restrictive communication norms can sometimes backfire in fictitious scenarios (e.g. Conway et al., 2009), and (2) some people report they like Trump’s willingness to violate cultural communication norms (Guo, 2015; Thompson, 2016).
However, this leaves many gaps to fill. (1) First, prior scientific work on the backfiring of PC norms has manipulated norm salience only in hypothetical role-playing contexts where participants imagined their behavior in fictitious scenarios using fictitious target groups. (2) Further, although polling and survey data suggest reasons to believe that Trump is an iconic representative of anti-communication norms, no scientific test has directly tied support for Trump to political correctness norms in a controlled experimental environment. (3) There is an implication in polling data and public discourse to focus on support for Trump as a conservative ideological issue (e.g., Burnett, 2016; Pew Research Center, 2016), and yet some academic analyses also suggest that Trump’s supporters may not be true ideological conservatives (Noel, 2016). Thus, it is important to parse out the effects of restrictive communication norms from that of political ideology.
To accomplish these goals, in the midst of the 2016 U.S. Presidential election, we manipulated the salience of PC norms by randomly assigning some participants to a restrictive communication prime where they read and responded to a brief description of the benefits of PC norms. We then measured (a) support for Trump and Clinton, (b) chronic concern with restrictive communication norms, and (c) political conservatism. We did this on a slightly left-leaning (M = 4.36 on a 1-9 conservatism scale) sample of non-college adults (M age = 34.38).
We expected that (1) participants’ chronic feelings of concern with restrictive communication norms would predict support for Trump, and (2) priming restrictive communication norms would increase support for Trump. We further expected that these effects would either not be in evidence – or be reversed – for Hillary Clinton.
Method
Participants and Disclosures
Three hundred and twelve participants were recruited through Mechanical Turk. We chose Mechanical Turk in part because it is been particularly validated for use as a representative sample for research related to politics and political ideology (see, e.g., Clifford, Jewell, & Waggoner, 2015) and generally shows similar results as other samples (for an example, see Houck, Conway, & Repke, 2014). Participation occurred in early-to-mid September, a period where the race between Clinton and Trump had tightened, but Clinton still maintained a lead in the polls (see http://www.realclearpolitics.com/epolls/2016/president/us/). We excluded seven participants who failed to answer an attention-check question accurately, leaving 305 for final analyses. No data collection occurred after data analysis, and all relevant manipulations, measures, and exclusions are disclosed.
Restrictive Communication Prime Manipulation
Prime Condition
Some participants (n = 95) were randomly assigned to the Restrictive Communication Prime Condition. Participants in this condition read the following introduction:
“First, we would like to get your opinions on societal norms. In our modern society, we have norms that dictate that we refrain from saying negative things – especially those things deemed as politically incorrect to say. These norms state that it is better to have rules that constrain us from anything that might sound too-negative or might be offensive to members of particular groups. These social norms that discourage too-negative conversation have many good benefits, and we first want to get your opinion on these norms before moving forward.”
Participants were then asked to respond to four questions expressing either support for or opposition to those norms. These questions (on a scale anchored by 1 and 7) asked participants the degree they were in favor of such norms, that such norms have value, that they oppose societal pressures to restrict communication, and that they believe in the value of norms governing communication.
Control Conditions
The present study used two different control conditions. First, some participants (n = 91) were randomly assigned to receive a description parallel to Restrictive Communication Prime participants; however, their description contained only positively-framed suggestions rather than negative restrictions (and made no reference to political correctness). Participants in this condition also answered four questions in support of norms that suggest we should be positive, sensitive, and respectful in our communication.
A second group of participants (n = 119) was randomly assigned to receive no introduction about communication or complete any questions relevant to communication norms.
These two control conditions showed largely identical patterns in the present work; for simplicity, we combined them into one Control Condition.i
Dependent Measures: Support for Trump and Clinton
Voting Intent
In two separate items, participants were first asked to report on a 7-point rating scale how likely they were to vote for Trump and Clinton, respectively.
Trait Positivity Ratings
Participants were then asked to rate each candidate on a series of parallel trait rating scales drawn from prior work (e.g., Conway et al., 2012). We organized these by semantic meaning: Five questions pertained to honesty (honest, liar [reverse], speaks mind, goes against norms, truthful; alphas = .92 and .87), four pertained to adaptability (flexible, thoughtful, complex, smart; alphas = .87 and .90), three pertained to strength (strong leader, tough, consistent; alphas = .85 and .88), two to general likability (like, respect; alphas = .93 and .91), and one each to kindness and attractiveness.
In order to produce an overall trait positivity rating that was unbiased by this semantic organization, we converted all sixteen trait rating items to z-scores and averaged them into a single trait positivity score for each candidate (alphas = .96).
Cumulative Trump and Clinton Support Measures
To capture the overall degree across all measures that participants supported Trump and Clinton, we converted each candidate’s voter preference and trait positivity scores to z-scores and averaged them into cumulative Trump Support (alpha = .92) and Clinton Support (alpha = .88) measurements.ii
Restrictive Communication Concern
We measured two constructs related to restrictive communication concern by adapting items from prior studies (Conway & Schaller, 2005; Conway et al., 2009). Three items measured reactance (alpha = .55): “I often feel pressured by society to keep my own negative political opinions to myself – and when I do, it makes me feel like I want to say my opinions anyway so no one can tell me how to think or talk,” “I often feel like political pundits are trying to make me fit their own narrow views and it makes me upset,” and “Political correctness norms generally aggravate me.”
Six items measured informational contamination (alpha = .71): “I believe that a lot of what politicians say is just to avoid saying things that might offend particular groups,” “I distrust a lot of what politicians say because I assume it is reflective of some political agenda,” “I believe that a lot of what politicians say is just to avoid saying things that might offend particular groups,” “I believe that politicians generally keep their real opinions to themselves because if they didn’t, they would be in trouble with powerful groups in society,” “I believe that politicians do not respond to social norms telling them what to say and, therefore, what they say is what they generally really believe” (reverse-scored), and “I distrust most of what members of the Democratic party say because I assume they are just manipulating people,” and “I distrust most of what members of the Republican party say because I assume they are just manipulating people.” Reactance and informational contamination were modestly correlated (r = .43).
Political Ideology
Participants completed two standard ideology items anchored by democrat/republican and liberal/conservative on 1-9 rating scales (see, e.g., Conway, Gornick, et al., 2016; Conway, Houck, Gornick, & Repke, 2016). We averaged them into a single measurement of political conservatism (alpha = .92, M = 4.36; SD = 2.30).
Results
Chronic Concern With Restrictive Communication Predicting Support for Trump and Clinton
We correlated our two primary measurements of restrictive communication concern (reactance, informational contamination) with measurements of support for Trump and Clinton. Table 1 presents the outcome of these analyses both for zero-order predictions as well as those which control for political ideology.
Table 1
Chronic Restrictive Communication Concern Predicting Support for Trump and Clinton
| Reactance
|
Informational Contamination
|
|||
|---|---|---|---|---|
| Zero-Order | Controlling for Ideology | Zero-Order | Controlling for Ideology | |
| Trump Trait Ratings | ||||
| Honest | .40*** | .20*** | .09 | -.10 |
| Strong | .37*** | .18*** | .12* | -.02 |
| Kind | .37*** | .18** | .04 | -.13* |
| Adaptable | .32*** | .12* | .11 | -.04 |
| Attractive | .19*** | .04 | -.07 | -.18*** |
| General Liking | .43*** | .25*** | .10 | -.07 |
| Trump Total Trait Positivity | .39*** | .18*** | .09 | -.09 |
| Trump Voting Likelihood | .35*** | .10 | .08 | -.14* |
| Total Trump Support | .38*** | .16** | .09 | -.12* |
| Clinton Trait Ratings | ||||
| Honest | -.33*** | -.20*** | -.33*** | -.26*** |
| Strong | -.37*** | -.22*** | -.18** | -.08 |
| Kind | -.36*** | -.22*** | -.30*** | -.23*** |
| Adaptable | -.30*** | -.16** | -.11 | -.01 |
| Attractive | -.30*** | -.19*** | -.28*** | -.22*** |
| General Liking | -.38*** | -.21*** | -.26*** | -.17** |
| Clinton Total Trait Positivity | -.37*** | -.23*** | -.26*** | -.17** |
| Clinton Voting Likelihood | -.39*** | -.20*** | -.23*** | -.12* |
| Total Clinton Support | -.40*** | -.24*** | -.26*** | -.16** |
Note. N = 305.
*p ≤ .05. **p ≤ .01. **p ≤ .001.
As can be seen there, a pattern emerged for Reactance consistent with expectations: Persons who felt chronic reactance at restrictive communication norms were significantly more likely to support Trump (overall support r = .38) and significantly less likely to support Clinton (overall support r = -.40) – and both of those effects remained significant even when controlling for political ideology (p’s < .01).
The pattern for informational contamination was less consistent. Informational contamination weakly but positively predicted support for Trump, but this pattern was reversed when controlling for ideology. Support for Clinton was, consistent with expectations, significantly negatively predicted by informational contamination (and this remained significant when controlling for ideology).iii
Overall, these data suggest qualified support for the perspective outlined here: It is clear that support for Trump (and opposition to Clinton) is especially likely amongst people who feel emotional reactance to restrictive communication norms – and importantly, this effect goes beyond political ideology. To a lesser degree, persons who express concern about informational contamination conformed to the expected pattern for zero-order correlations; but while a significant effect remained for Clinton in the expected direction when controlling for ideology, the effect for Trump actually reversed (albeit weakly). We return to what these differences might mean in the Discussion.
Priming of Restrictive Communication Norms
A series of One-Way ANOVAs compared the Restrictive Communication Prime versus Control Conditions. Table 2 presents these results in full. As can be seen there, these results provided clear support for the norm backfiring perspective outlined in the introduction: Participants made to think about restrictive communication norms showed more support for Trump than controls (cumulative Trump Support F[1,304] = 5.39, p =.02). Also consistent with expectations, this effect was exclusive to Trump (cumulative Clinton Support F[1,303] = 0.05, p =.82).iv
Table 2
Effects of Restrictive Communication Prime on Support for Trump and Clinton via One-Way ANOVAS
| Control Conditions | Restrictive Prime | Effect Size r | Difference p-value | |
|---|---|---|---|---|
| Trump Trait Ratings | ||||
| Honest | -.09 | .19 | .15** | .009* |
| Strong | -.05 | .11 | .09 | .137 |
| Kind | -.09 | .20 | .13* | .020* |
| Adaptable | -.05 | .10 | .08 | .157 |
| Attractive | -.03 | .07 | .05 | .398 |
| General Liking | -.08 | .16 | .11* | .047* |
| Trump Total Trait Positivity | -.06 | .14 | .12* | .039* |
| Trump Voting Likelihood | -.09 | .20 | .13* | .019* |
| Total Trump Support | -.09 | .19 | .13* | .021* |
| Clinton Trait Ratings | ||||
| Honest | -.03 | .06 | .05 | .418 |
| Strong | .01 | -.01 | -.01 | .852 |
| Kind | .02 | -.04 | -.03 | .667 |
| Adaptable | -.02 | .05 | .04 | .560 |
| Attractive | -.01 | .01 | .01 | .872 |
| General Liking | -.00 | .01 | .01 | .920 |
| Clinton Total Trait Positivity | -.01 | .02 | .02 | .760 |
| Clinton Voting Likelihood | .02 | -.05 | -.03 | .556 |
| Total Clinton Support | .01 | -.01 | -.01 | .823 |
Note. N = 305.
*p ≤ .05. **p ≤ .01. **p ≤ .001.
Moderation and Mediation of Priming Effect
We expected that the effect of the prime would likely be especially in operation for persons who had higher levels of chronic concern with restrictive norms. To test this, we performed simultaneous regression analyses (using the SPSS macro developed by Hayes, 2013) entering (a) the dummy-coded prime, (b) each moderator, and (c) the interaction of each moderator with the prime (for exemplars, see e.g., Conway & Schaller, 2005; Conway et al., 2009). For brevity, we focus here on analyses predicting the cumulative support measures which incorporate all the support-based measurements.
These results revealed that reactance was a significant moderator for the Prime – Trump Support effect (interaction beta = .16, p = .03, lower CI = .01, upper CI = .32). Subsequent analyses (Hayes, 2013) revealed the expected pattern: When participants felt low levels of reactance against restrictive norms, the effect of the prime on Trump Support was small (beta = .13, p = .36); but at high levels, this effect was much larger and significant (beta = .57, p < .001). This same moderating pattern held for informational contamination on Trump Support; however, the interaction did not approach statistical significance (beta = .16, p =.184, LCI = -.08, UCI = .39).
A conceptually identical pattern emerged for Clinton Support, although which restrictive communication concern variable was the strongest moderator was reversed: Informational Contamination was a significant moderator (interaction beta = -.24, p = .034, lower CI = -.47, upper CI = -.02). Subsequent analyses revealed a pattern largely consistent with expectations: When participants felt low levels of informational contamination, the effect of the prime on Clinton Support was negative (beta = -.27); but at high levels, this effect was actually somewhat reversed (beta = .22). This same moderating pattern held for reactance on Clinton Support; however, the interaction did not approach statistical significance (beta = -.08, p = .33, LCI = -.23, LCI = .08).v,vi
To test the mediating effect of reactance and informational contamination, we tested their indirect effects in a series of A(Prime)→B(Mediator)→C(Candidate Support) analyses (Hayes, 2013). These analyses revealed weak and non-significant indirect effects of the two general concern measurements with restrictive communication norm measurements (all Sobel tests of indirect effect ps > .16)vii – indeed, the prime did not significantly affect either reactance or informational contamination, making the A→B→C path untenable. We return to what this might mean for interpretation in the discussion.viii
General Discussion
Taken in total, these results provide support for a connection between the salience of restrictive communication norms and support for Donald Trump. Participants primed to think about restrictive communication norms showed more support for Trump than participants who were not so primed. Further, participants who felt chronic reactance to communication norms were especially likely to express support for Trump (above and beyond their political ideology), and were further especially likely to respond to the prime by increasing Trump support. In the main, these results either did not hold – or were reversed – for supporting Clinton.
These results are not without their limitations and interpretational ambiguities, however. We first address those limitations and then discuss, in larger perspective, what we can learn from these results.
Interpretation Obstacles
Lack of Mediation by Chronic Concern Measurements
One of the interpretational ambiguities in the present results involves the fact that, although chronic concern measurements largely predicted support for Trump (and opposition to Clinton) across the whole sample, these measurements did not also mediate the relationship between the Restrictive Communication Prime and Trump Support. If the Prime operates by heightening the salience of restrictive communication norms, why would these measurements not partially account for the effect of the prime on candidate preference?
A few points suggest that this lack of mediation was likely the result of our reactance and informational contamination measurements being too stable to effectively serve as the mediator for the effect of a temporary prime. First, at face value, the questions were written at a very general level to apply to any political group, norm, or politician. Prior work that showed the mediation of emotional reactance and informational contamination on PC norm priming was written specifically for a very clearly identified scenario – thus the overall psychological distance between the prime, proposed mediator, and dependent measure was far less in that work than in the present study (see Conway et al., 2009). As an initial test of the possibility that this helps account for our results, we took the three questions from the trait positivity rating that were most related to scenario-specific concern with communication norms and combined them into a single measurement of scenario-specific concern with communication norms (all reverse-scored; alpha = .87). The three items were: “I appreciate that Donald Trump seems to say whatever is on his mind, even if it upsets people,” “For the most part, I trust Donald Trump to speak truthfully,” and “Donald Trump goes against cultural norms that restrict free speech, and that’s a good thing.” We then used this measurement as the mediator an A(Prime)→B(Mediator)→C(Outcome) analysis of indirect effects. (To avoid too much overlap between related measurements, we used only voting preference as the outcome measurement. Results are identical if we use the overall Trump Support measurement). Results suggested a significant indirect effect of the newly-constructed, Trump-specific communication norms measurement (beta = .50; lower CI = .04; upper CI = .96; Sobel test z = 2.16; p =.03). We recognize that this measurement has some method overlap (and that is one of the reasons why we did not use it as the primary measurement of the construct in the first place) and it should be interpreted cautiously. However, it is at least consistent with the notion that there was a more transient communication-relevant mediator of the present results.
Second, the chronic measurements of emotional reactance and informational contamination did interface with the prime; however, they interfaced as one would expect a measurement that is more chronic and thus less subject to direct temporal fluctuations. Specifically, these measurements served as fairly consistent moderators of the prime→support relationship.
Third, these chronic measurements were predictive overall of support for Trump (and especially, opposition to Clinton) in ways that are generally consistent with predictions (with some caveats we return to later in the discussion). This is again consistent with their status as a more stable measurement that is relevant to the proposed outcomes at a larger, more stable level.
Finally, these results do not stand alone. Although they are unique in their application of political correctness priming to Trump support, other work shows the potential backfiring of communication norms and artificial consensus in other domains (Conway & Schaller, 2005; Conway et al., 2009). As a result, in the larger theoretical landscape, there is reason to believe that concern with restrictive communication norms may have an impact in the present work.
Taken together, this evidence suggests that while the lack of direct mediation by those measurements is a weakness in the present results, we feel there is ample reason to nonetheless trust that the larger picture emerging here is one best captured by a theoretical angle focusing on concern with restrictive communication.
Inconsistency of Chronic Concern With Communication Norms?
Another interpretational ambiguity lies in the fact that Trump Support was predicted by chronic emotional reactance but not chronic informational contamination. What are we to make of this inconsistency?
One possibility is that Trump simply evokes more emotional deviance than cognitive deviance. Reactance is more purely emotive (see Knowles & Linn, 2004); informational contamination is more purely cognitive (see Conway & Schaller, 2005; Conway et al., 2009). Recall that in prior research and theory these two types of deviance are independent predictors of cultural norms backfiring. Every theoretically sound predictor cannot predict every phenomenon – and in this case, it may be rather instructive that it is emotional reactance, and not informational contamination, that predicts Trump support. To the degree this is true, it suggests that it is more of a guttural emotive dislike of restriction than a cognitive erosion of information that is behind Donald Trump’s support.
Indeed, it is noteworthy that these results actually show an interesting set of consistencies across analyses that partially support this conclusion. (1) First, not only was emotional reactance a significant predictor of Trump support overall, it (and not informational contamination) was a significant moderator of the effect of the prime on Trump support. This suggests the prime is having its effect on people who have a chronic emotional dislike of restrictive communication norms, more so than people who are cognitively concerned with informational contamination. (2) Equally as interestingly, informational contamination had much more of an impact on support for Clinton (compared to support for Trump) in the opposite direction, both as a direct predictor and as a moderator of the prime→Clinton support relationship. While it is difficult to fully know how to interpret this and any interpretation must remain speculative, it is at least consistent with the notion that support for Clinton is more of a function of cognitive concerns than is support for Trump.
What Can We Learn?
At a minimum, these results provide evidence that (a) priming restrictive communication norms and (b) chronic emotional reactance against those norms both increased support for Donald Trump during the general election. These results help us in our theoretical understanding of culture in multiple ways. First, they provide for the first time to the authors’ knowledge an experimentally-manipulated example of a cultural norm’s ironic effect on a real-world political example. Indeed, a reasonable criticism of much of the past work on the backfiring effect of cultural norms is that it focuses on hypothetical scenarios that are removed from the real world (see, e.g., Conway et al., 2009). While this prior work has value, it is important to establish any set of effects in real contexts that truly matter to people, and the present study does just that.
Second, these results also contribute to the larger theoretical discussion about how cultures emerge, change, and have influence. It is noteworthy that, unlike in prior research on the backfiring of communication norms, the prime in this study bore little direct linguistic overlap with the outcome measurement (support for Donald Trump). As a result, it suggests the pervasive value of understanding the eventual emergence of controversial political figures through the lens of communication-based theories of cultural norms and cultural deviance.
Cultures emerge and change for many reasons. These can be considered in two basic categories: (1) Top-down ecological or political pressures – such as frontier topography or dictatorial political systems – may directly constrain how culture emerges (e.g., Conway et al., 2014; Kitayama, Conway, Pietromonaco, Park, & Plaut, 2010; Murray & Schaller, 2014; Van de Vliert, 2013). (2) However, often communication processes operate independent of these top-down pressures to shape the emergence and influence of culture (e.g., Conway, 2004; Conway & Schaller, 2007; Conway, Sexton, & Tweed, 2006; Schaller, Conway, & Tanchuk, 2002). The present study falls squarely at the intersection of these two categories. It suggests that the top-down pressures that often shape culture are themselves subject to communication and psychological processes that direct their influence. If the top-down communication norms are too heavy-handed, they may indeed create a forced consensus; but that consensus is psychologically fragile and prone to break. The present results thus complement prior work on hypothetical scenarios, providing a very important bellwether test for the degree that large-scale cultural norms might ironically rebound. They suggest that instead of producing the niceness that they are intended to produce, these norms instead might lead to more nasty political discourse in the long-term.
Third, these results contribute to the larger literature on the importance of the immediate context to political decisions. Consider, for example, work on framing that demonstrates the specific way a particular finding is discussed – or framed – can have a great deal of impact on political attitudes and behaviors (Bizer, Larsen, & Petty, 2011; Bizer & Petty, 2005; Brewer & Gross, 2005; de Vreese, Boomgaarden, & Semetko, 2011). Some of this works suggests that positive framing of news articles can increase the likelihood of endorsement of a specific opinion (e.g., de Vreese et al., 2011). Other work shows that framing an attitude as an “opposition” to something increases the subsequent longevity of the opinion and its strength (Bizer & Petty, 2005; Bizer et al., 2011).ix
Like this prior work on framing, our work also suggests a mechanism by which the immediate context can influence political attitudes. Specifically, our work suggests that the immediate salience of chronic communication norms may aid candidates who appear to violate those norms. Thus, communication norms can function psychologically in much the same way that frames can by focusing persons on a part of the context that ultimately shapes their attitudes and decisions.
Finally, these results also demonstrate the value of applying both existing social psychological theory and experimental methods to understand puzzles in the modern world. A descriptive illustration of this power can be seen in Figure 1. Looking at the raw mean scores for voting preference – the clearest marker in our study of the likelihood of potential cultural change – reveals a powerful pattern: In control conditions, Clinton was soundly preferred to Trump (Clinton M = 3.66, SD = 2.41; Trump M = 2.63, SD = 2.28), but when Political Correctness norms were made salient, this gap virtually disappeared (Clinton M = 3.48, SD = 2.51; Trump M = 3.32, SD = 2.51).
Figure 1
Voting likelihood for Trump and Clinton by restrictive communication prime.
Concluding Thoughts
Although Donald Trump presents an interesting paradox of sorts to modern political pundits, his emergence is precisely what a theory focusing on the backfiring of social norms would expect. It is a paradox, but a theoretically-expected one: As restrictive norms become ever more salient and heavy-handed, the more they will work in the short-term. But in the long-term, this salient heavy-handedness increases the likelihood that they will ultimately backfire. And this backfiring doesn’t just occur for norms that are genuinely repressive to political freedom – it also occurs for norms that have a clearly good and noble aim.
The present study suggests that communication norms that are designed to increase the amount of positive communication may ultimately backfire in political figures like Donald Trump – figures who do anything but increase the amount of positive communication. His emergence should serve as a lesson for students of cultural change and deviance. He is not the first and will certainly not be the last example of popular figures emerging in response to restrictive norms – and the present work illustrates specific psychological processes that help us better understand why that occurs and when it will occur.