Love the Science, Hate the Scientists: Conservative Identity Protects Belief in Science and Undermines Trust in Scientists

Abstract

Introduction

The sociology of science has long recognized that extra-scientific social dynamics and agendas have profound influences on scientific production (Crane 1972; Latour and Woolgar 1979) and that the boundaries of what is considered legitimate science are continually redrawn, contested, and subsequently defended by invested parties (Gieryn 1983, 1999). These are critical sociological insights as science is among the most powerful epistemological currencies in the world, with those claiming its support enjoying significant advantage in a range of debates spanning culture, morality, and politics (Brown 2009; Epstein 1998). This is particularly clear in the United States, where the general population has been divided along political, religious, and cultural lines by disagreement on scientific issues of collective concern such as climate change (Farrell 2016a, 2016b; Fisher, Waggle, and Leifeld 2013; McCright and Dunlap 2010), safety of vaccines (Blume 2006; Nyhan et al. 2014; Omer et al. 2009; Salmon et al. 2005), and scientific education (Binder 2002; Roos 2017), to name some examples. But no cultural divide on scientific issues touches on as many subjects or has been as institutionalized as the one between politically liberal and conservative partisans (Farrell 2016a, 2016b; Fuller 2014a; Gauchat 2012; Gross, Medvetz, and Russell 2011; Medvetz 2012).¹ Improving understanding of this scientific-political divide remains an important line of inquiry not only for social scientists, but for policymakers attempting to appeal to audiences in areas where scientific communication is hamstrung by divergent scientific attitudes and perceptions.

Perhaps the most important recent findings on this issue have been that political polarization on scientific issues is characterized by a sharp decline in trust in the scientific community among self-identifying political conservatives (Gauchat 2012) and their enduring suspicion of the relationship between science and government (Evans and Feng 2013; Gauchat 2015; Moore 2008).² The flipside of these findings is that trust in scientists remains high worldwide compared to other institutions (Krause et al. 2019) and that these negative attitudes do not extend to science itself as a method or epistemology. Indeed, it is not well understood why other scientific attitudes have remained highly and nearly universally positive among both liberals and conservatives in the United States. For instance, about 91 percent of liberal, conservative, and moderates agree that science and technology provide more opportunities for the next generation.³ Moreover, this pattern has remained stable over the past 10 years. A similar trend emerges when looking at those who believe the harms of scientific research outweigh the benefits. Only around 8 percent agree with this notion and this attitude has remained relatively stable. Understanding this mismatch between attitudes towards scientists and attitudes toward scientific research itself, as well as the role of conservative identity in the subjective organization of these attitudes, is the main goal of this study.

Although self-identifying political conservatives in the United States show high levels of distrust toward the scientific community (Gauchat 2012), they are far from abandoning science as a valid epistemology and a field in which crucial cultural contests might be won. This insight—that audiences are able to partition scientific beliefs and attitudes according to cultural preferences—has been most fully appreciated in the context of conservative Protestants. Scientifically knowledgeable religious conservatives have been able to effectively partition their knowledge (Roos 2014, 2017) and attitudes (O’Brien and Noy 2015) in ways that maintain a broad recognition of the legitimacy of scientific endeavor while selectively rejecting the science and, more importantly the scientists, that contradict particular religious (e.g., creationist) or political (e.g., climate science) identities and worldviews (Ecklund et al. 2017); impinge on areas perceived as outside their purview, like public policy or morality (Evans 2018; Evans and Feng 2013); or, in the case of scientists specifically, are perceived as personally hostile toward religion (Ecklund 2012; Ecklund and Scheitle 2017). These studies demonstrate how faith in science is maintained among conservative religious practitioners in the face of growing distrust toward the scientific community. Yet, while these studies ask important questions about political and religious identities in relation to scientific perceptions, none examine the differences between attitudes toward scientific research itself and attitudes toward scientists among self-identifying political conservatives, more generally.

Drawing on research on conservative social movements, political partisanship, and trust in scientific institutions, we address this gap by developing hypotheses on the effect of conservative identity on public perceptions of science. We argue that a stable conservative identity should be associated with both distrust of scientists and belief in scientific research as a benefit to society as both are indicators of familiarity with, and investment in, scientific cultural contests. Furthermore, we posit that these relationships are enabled by the historical development of conservative scientific institutions that provide conservatives with the cognitive and rhetorical tools to maintain engagement in scientific debates (Gross 2013; Gross et al. 2011; Medvetz 2012). Support for these arguments would provide much needed context for both scholars of conservative perceptions of science and the general public as they indicate the continued cultural differentiation of scientific authority rather than abandonment of the scientific field among those critical of the scientific community. They would also constitute evidence that maintaining belief in scientific research as a benefit to a society is at least partially predicated on having stable cultural associations that are orienting and enabling of scientific participation in political contexts.

To test this view, we use two different approaches. First, we use cross-sectional data from the General Social Survey (GSS) to explore the relationship between political identity and two measures of scientific perceptions: one that characterizes science as research with potential benefits to society and another that characterizes it as a community of scientists. We then test whether change in political identity is differentially associated with change in these scientific attitudes. We find evidence that those with more stable conservative identities are more likely to reject mainstream scientists than those who switch their conservative identity at some point. Stable conservatives, however, are also more likely than those who switch political identities to believe scientific research is a benefit to society. These results suggest that negative attitudes toward the scientific community go hand in hand with a belief in scientific research itself among conservatives.

Decline of Trust in Mainstream Science

Concerned with the general cultural fragmentation of scientific attitudes and knowledge in the twentieth century, Moore (2008) points toward the issue of government-scientific cooperation, observing that the post-WWII union between mainstream scientific institutions and the military, as well as regulatory bodies, led to backlash from many elite scientists and the general public. This resulted in a widespread crisis for institutional scientific authority across party lines among those critical of science’s supportive role in specific political agendas. The resulting cultural differentiation of scientific perceptions has been studied by social scientists (see e.g., Fuller, Stenmark, and Zackarisson 2014b), but the most dramatic trends, and therefore the most attention, have been focused on political conservatives.

In his examination of long-term declines in trust in scientists in the United States, Gauchat (2012) finds evidence that this decline is most dramatic among more educated conservatives and particularly with regard to the perceived role of scientific activity in the implementation of government regulations. This suggests an important relationship between political and scientific sophistication and animosity toward scientific institutions. More recently, Gauchat (2015) demonstrates that the social-psychological mechanisms underlying this conservative decline are multiplex, being driven by both cognitive and emotional tendencies housed under the conservative identity. Gross (2013) extends this work to academia in general, arguing that the political liberalization of the academy (due mainly to self-selection of liberals into academic careers) lead to the broad perception of the academy as a liberally biased institution and helped provide the impetus for the establishment of alternative conservative sources of scientific and academic authority. The latter trend contrasts sharply with the notion that conservatives are abandoning science as a valid epistemology and indeed a literature is emerging on the contemporary rise of this kind of culturally oriented knowledge production.

Conservative Scientific Institutions

We define “conservative scientific institutions” broadly as organizations and individuals that leverage the institutional forms, language, and legitimation practices traditionally associated with academic science to counter findings perceived as liberally biased and to help advance conservative political goals. This can mean anything from organizations like conservative think tanks that produce policy relevant scientific research and Evangelical universities with research and teaching missions meant to provide a fundamentalist counterpoint to secular universities to online content creators aiming to undermine scientific arguments that enjoy broad consensus among experts.⁴ This definition is in line with Gross et al. (2011, 333) and Blumenthal (1986) and their characterization of the rise of the “conservative counter-establishment.” Yet, whereas they view the conservative counter-establishment as the institutionalization of a general intellectual movement, the term “conservative scientific institutions” has the benefit of indicating conservative organizations and people concerned specifically with challenging scientific consensus in scientific terms.

In recent years, research has been exploring the ascendance of conservative scientific authority that is designed to counter the cultural authority of mainstream science. Medvetz (2012) shows how the rise of think tanks in the latter half of the twentieth century allowed conservatively partisan organizations—often straddling the boundaries of science, politics, and media—to bypass traditional academic gatekeepers and communicate research and findings amenable to their causes directly to their audiences. On the issue of climate change, evidence accumulates on the central role of oil companies and special interests in creating an ecosystem of anti-climate change science that enjoys a wide lay audience of conservatives (Dunlap and Jacques 2013; Farrell 2016a, 2016b; Jacques, Dunlap, and Freeman 2008; McCright and Dunlap 2000, 2003, 2010). Additionally, historians have been documenting the rise of conservative institutions dedicated to providing a counterpoint to what they see as liberally biased science and education (Blumenthal 1986; Martin 2013). Together, these studies show a conservative unwillingness to cede the scientific field or to adopt another authoritative epistemology (e.g., religion) in its place. Rather, significant resources and time have been dedicated toward building a scientific movement (Frickel and Gross 2005) meant to counteract mainstream science by using the same language, methods, and credentials to establish expertise.

The rise of explicitly conservative intellectual authority has been more fully appreciated in the context of political news media. Similar to the scientific trends described above, the liberal leanings of the majority of professional journalists has alienated conservative audiences and has led to conservative elites’ establishment of alternative sources of information (Baum and Groeling 2008; Berry and Sobieraj 2013; Ladd 2012). Much of this research has been concerned with the purported rise of “echo chambers” online, among both liberals and conservative, where the increased agency of individuals to pick their news sources, the proliferation of more partisan news sources, and the tendency of advanced algorithms to channel people to evermore personally catered content have all conspired to reinforce existing beliefs and biases among partisans (Bakshy, Messing, and Adamic 2015; Bennett and Iyengar 2008; Faris et al. 2017; Stroud 2010; Sunstein 2001, 2002). These studies document the flourishing of culturally specific knowledge online and the increasing ability of conservative elites to bypass traditional gatekeepers in media (D’Alessio and Allen 2000; Shoemaker, Vos, and Reese 2009) and science (Dunlap and Jacques 2013; Farrell 2016a, 2016b; Medvetz 2012) to deliver their message to conservatives audiences. Though rarely concerned with scientific issues specifically, it stands to reason that not only should some portion of these partisan messages be scientific in nature, but that we should expect similar effects from the decrease in conservative trust toward the scientific community detailed above.

Conservative Scientific Repertoire

To understand the role of conservative identity in orienting scientific perceptions among conservatives in a culturally differentiated field, we build on Perrin et al.’s (2014) theory of a performative model of conservatism as a collective identity. This theory recognizes conservatism neither as a singular ideology that ties beliefs together (see Martin 2002) nor as a coalition of disparate groups (Miller and Schofield 2003, 2008). Rather, Perrin et al. argue that identification with conservatism, “endows individuals with access to additional repertoires which, in turn, may change their political positions …”, where they “… learn conservative modes of thought by common participation in conservative identity…” and which “… provides members with political knowledge and cultural maps for deploying and culturally locating and warranting conservative ideas” (Perrin et al. 2014:289). We argue that one such repertoire is the conservative scientific repertoire, an accessible cultural schema that is able to reconcile various and sometimes conflicting conservative values including traditionalism, authoritarianism, anti-intellectualism, and religious fundamentalism with scientific norms, language, and action that enables a coherent presentation of scientific arguments across the political spectrum.

Structurally, conservative knowledge institutions—like the ones elaborated above that are run by degree-holding scientists, intellectuals, media elites and others—play an important role in the creation, dissemination, and activation of the content of a conservative scientific repertoire, with elites in various fields offering distinct but related components. For instance, leaders in conservative political media are well placed to criticize and cast doubt on the scientific community, while those writing books and running think tanks are better suited to provide the scientific scaffolding for arguments against things like anthropogenic climate change and evolution by natural selection. Placing Gauchat (2015) and Perrin et al. (2014) in conversation clarifies how conservatives can have such vastly different attitudes toward science as a social practice and science as a source of knowledge. By adopting an orientation that sees culturally distinct scientific institutions as variably legitimate or illegitimate, while valuing science itself as an important societal cornerstone, conservatives can preserve their faith in science while showing less trust in the mainstream scientific communities that are promoting research inimical to conservative causes. In the context of the aforementioned rise in conservative scientific institutions and an increasingly pluralistic scientific landscape, this view sees these alternative sources of elite scientific authority as providing conservatives with the conceptual and rhetorical tools (Swidler 1986, 2001) to participate in scientific debates—in scientific terms—that have bearing on political outcomes they view as important.⁵ This is how specific scientific communities might be distrusted and challenged while scientific research itself, a useful tool in political debates, maintains its status as a widely revered societal benefit. This leads to our first set of hypotheses:

• H₁: There will be an increasingly negative relationship between conservative identity and trust in the scientific community over time.

• H₂: There will be no change in relationship between conservative identity and agreement that scientific research is a general benefit to society.

Political Identity Switching

In addition to predictions about general trends, this perspective also sets up expectations about how conservative identity will affect perceptions of science at the individual level. To start, ideologues are better able than their politically moderate or ambivalent counterparts to efficiently interpret situations according to ideologically informed schemas (Goren 2003; Martin and Desmond 2010), more likely to consume partisan media (Prior 2013), and are generally more sophisticated and knowledgeable about political issues (Cohen 2003; Kahan 2012; Sherman and Cohen 2006). Keele and Wolak (2006) examine differences between ideologues and moderates in terms of “partisan volatility” or “party switching” and find that switching arises in cases of low information (the political sophistication argument) but also when individual values contradict those of partisan elites. These findings distinguish the ideologically stable from unstable by political knowledge and familiarity with elite characterizations of partisan divides, both of which align with a performative theory of conservative identity and familiarity with conservative institutions and arguments. In other words, stable conservatives are an ideal group to examine a performative theory of conservative identity not only because their identities have been shown to be robust to external political change, but because this robustness is drawn from an awareness and understanding of elite conservative influence.

The relationship between strong or stable political identities and familiarity with partisan arguments has two important implications for conservative perceptions of science. First, this literature would indicate that those with stable conservative identities should be more familiar with political criticisms of the scientific community. This leads to our third hypothesis:

• H₃: Those who have stable conservative identities will show less trust in the mainstream scientific community compared to those who report unstable conservative identities.

Second, stable conservatives are also more likely to be familiar with the scientific arguments generated from conservatively partisan institutions described above, providing them with the scientific competency to participate in scientific debates. This leads to our fourth hypothesis that:

• H₄: Those who have stable conservative identities will be more likely to agree that scientific research benefits society compared to those who report unstable conservative identities.

Together, these hypotheses outline the scientifically orienting role of conservative identity that we described above: an orientation that is protective of the belief of the benefits of scientific research in general while allowing a sharp distrust of the mainstream scientific community.

Methods

Data

This study uses two types of GSS data to test this perspective. The GSS is a nationally representative face-to-face survey of the noninstitutionalized, adult population in the United States and has been widely used to study perceptions of science (e.g., Gauchat 2012; O’Brien and Noy 2015; Roos 2014). The GSS has collected a cross-sectional sample of the US population from 1972 to 2016 and, since 1973, has collected information about respondents’ confidence toward the scientific community. Since 2006, the GSS cross-section has captured additional data about scientific perceptions among respondents. We use this information to uncover whether the relationship between political identity and scientific perceptions has changed among the US population. After accounting for missing data, our analytical sample from the cross-sectional GSS includes 35,266 individuals from 1973 to 2016 and 5,204 individuals from 2006 to 2016.

Starting in 2006, the GSS collected a series of three-wave panels, and as of 2014, they have completed three full three-wave panels. For example, of the 4,510 respondents surveyed in 2006, the GSS randomly selected 2,000 to re-interview in 2008 and 2010. Overall, 64 percent of the respondents empaneled in 2006 were captured in all three waves (N = 1,276). In 2008, the GSS empaneled 2,023 new respondents, and among these individuals, 64 percent (N = 1,295) completed all three waves. Finally, the GSS interviewed an additional 2,044 new respondents for their 2010 panel and 64 percent (N = 1,304) of those empaneled in 2010 completed all three waves. We combine these panels into a single overlapping panel dataset.

We follow similar procedures for the 2008 panel (which becomes the 2008–2010 and 2010–2012 mini panels) and the 2010 panel (which becomes the 2010–2012 and 2012–2014 mini panels). We then combine these six two-wave panels into a single dataset for our analyses. Because of this transformation, some individuals may occur twice in our analytical dataset and to account for this we use panel robust standard errors.

Dependent Variables

We focus on confidence in the scientific community and attitudes toward the general benefits of scientific research for society as our outcomes. The GSS captures information about respondents’ confidence in “the people running” several different social institutions, and while the role of general institutional trust might be intriguing in this context, we focus our primary analysis on the scientific institution.⁸ The GSS asks “As far as the people running [the scientific community] are concerned, would you say you have ‘a great deal of confidence’ (1), ‘only some confidence’ (2), or ‘hardly any confidence at all’ (3) in them?” Following Gauchat (2012), we focus on those expressing a great deal of confidence in the people running the scientific community compared to those who do not. We recode this outcome into an indicator for those expressing a “great deal of confidence” in the scientific community (coded 1).⁹

To model attitudes toward scientific research’s overall benefit to society, we use the GSS’s scale on the benefits of scientific research.¹⁰ The GSS presents respondents with the following statement: “People have frequently noted that scientific research has produced benefits and harmful results. Would you say that, on balance, the benefits of scientific research have outweighed the harmful results, or have the harmful results of scientific research been greater than its benefits?” The potential answers to this question are “the benefits are greater” (1), “the benefits are about equal” (2), and “harm is greater” (3). For those respondents who believe the benefits are greater, the GSS asks the follow-up questions: “Would you say that the balance has been strongly in favor of the benefits (2), or only slightly (1)?” For respondent who believe the harms are greater, the GSS asks: “Would you say that the balance has been strongly in favor of the harmful results (2), or only slightly (1)?” With all this information, we have a five-point scale that runs from strongly agreeing that scientific research is harmful to society to strongly agreeing that scientific research benefits society. Again, we want to understand differences between those expressing a great deal of confidence in scientists and scientific research and so recode this information into an indicator for those who strongly agree that scientific research benefits society (coded 1).

Key Independent Variables

The GSS collects information about respondent’s political identity and this information has been used to capture political switching in the United States (Hout and Fischer 2014). Around 22 percent of individuals change political ideology across survey waves. The GSS presents respondents with the following scale “extremely liberal” (1), “liberal” (2), “slightly liberal” (3), “moderate” (4), “slightly conservative” (5), “conservative” (6), and “extremely conservative” (7) and ask them: “Where would you place yourself on this scale?” We recode the first three responses categories into an indicator for political liberals and the last three into an indicator for political conservatives, with political moderates as the comparison group.¹¹

Our main interest for these political identity indicators is not current identity, but whether the respondents changed their affiliation indicating a less stable identity. To capture this, we create a series of change indicators for each of our political identity categories. For example, we create a binary variable for those who were always conservative |$({\mathrm{Conservative}}_{t1}=1\,\&\,\mathrm{Conservative}{l}_{t2}=1)$|⁠, those who became conservative between waves |$({\mathrm{Conservative}}_{t1}=0\,\&\,{\mathrm{Conservative}}_{t2}=1)$|⁠, and for those who left the conservative category between our two panel waves |$({\mathrm{Conservative}}_{t1}=1\,\&\,{\mathrm{Conservative}}_{t2}=0)$|⁠. Here, the comparison group is those who were never conservative |$({\mathrm{Conservative}}_{t1}=0\,\&\,{\mathrm{Conservative}}_{t2}=0)$|⁠. We follow the same procedures for the political liberals, and these six change indicators form our measures of political stability for our panel models.

Control Variables

While our panel outcome measures are from wave 2 of our mini-panels, we control for several factors from wave 1 that have been shown to affect scientific perceptions. There are larger literatures on the effects of religion on attitudes towards science (Baker 2012; Evans and Evans 2008; O’Brien and Noy 2015; Roos 2014) and individual politics (Brooks and Manza 2004; Hout and Fischer 2002, 2014; Manza and Brooks 1997; Regnerus, Sikkink, and Smith 1999; Wald and Calhoun-Brown 2014). To control for religious affiliation, we use a modified version of the RELTRAD classification scheme (Steensland et al. 2000; Woodberry et al. 2012) that decomposes black protestants into mainline and evangelical protestants to avoid collinearity with our race controls (Uecker and Ellison 2012; Wilcox and Wolfinger 2007) and places Jewish individuals into the other religious traditions category due to the small number of Jewish respondents (Schleifer and Chaves 2017). With these adjustments, we have four indicators for Mainline Protestants, Catholics, nonreligiously affiliated, and other religious traditions, with the evangelical Protestants as the comparison group. We also control for frequency of religious service attendance. The GSS asks respondents: “How often do you attend religious services?” with the response categories: “‘never’ (0), ‘less than once a year’ (1), ‘once a year’ (2), ‘several times a year’ (3), ‘once a month’ (4), ‘two or three times a month’ (5), ‘nearly every week’ (6), ‘once a week’ (7), and ‘more than once a week’ (8).” We recode this measure into an indicator for those who attend once a week or more (coded 1).

Drawing from the aforementioned literatures on scientific perceptions that indicate greater knowledge is associated with more partisan views on science (Bauer, Allum, and Miller 2007; Kahan et al. 2012; Sturgis and Allum 2004), we include indicators for those who have completed either a bachelor’s or an advanced degree (coded 1), with those with less than a bachelor’s degree as the comparison group.¹² We control for socioeconomic differences by including a variable for the natural log of equivalized family income.¹³ To control for family differences, we include binary dummies for those currently married as well as those who have ever had a child. Race is captured by two binary indicators for black and other race individuals, with white individuals as the reference. We include a dichotomous indicator for female respondents, and to account for age differences in scientific perceptions, we include a continuous measure of age that runs from 18 to 98 years or more.

Finally, in a supplementary analysis, we explore where stable conservatives get their information about science. The GSS asks: “We are interested in how people get information about science and technology. Where do you get most of your information about science and technology?” Possible response categories include newspapers, magazines, the Internet, books or other printed materials, TV, radio, government agencies, family, friends, colleagues, or some other source. Each of these categories is mutually exclusive, and because we are primarily interested in those who get scientific information from the Internet—building off the aforementioned research that suggests the Internet and social media has been integral in the balkanization of partisan knowledge ecosystems—we create a binary indicator for those who report getting most of their scientific information from the Internet compared to those who do not. This question was asked of a small subsample of the GSSP from 2006 to 2014 and therefore not included in our main models. We are able, however, to use a descriptive analysis to explore how this suggestive mechanism may be influencing our findings.

Analytical Strategy

Results

Table 2 explores how attitudes toward the scientific community and scientific research as a social benefit have changed over time |$({H}_1\ \mathrm{and}\ {H}_2)$|⁠. Model 1 on “confidence in scientific community” captures the general pattern across our covariates. The coefficient for liberals shows that relative to political moderates, liberals report a higher degree of confidence in the scientific community. When averaged over this time-series, around 47 percent of liberals are predicted to have a great deal of confidence in the scientific community, while only around 42 percent of conservatives are predicted to show this level of confidence.¹⁷ Moreover, the significant coefficient for year suggests that overall, all Americans have become less confident in these scientific institutions over time. In 1973, the probability that any given American had a great deal of confidence in the scientific community was around 46 percent and by 2016 this number had changed to around 41 percent.

For Model 2, we decompose trends over time by political ideology. We can see that political moderates (here the main effect of year) and political liberals show little to no change in their confidence in scientific institutions over time. Political conservatives, however, show a large and meaningful decrease in their confidence in the scientific community over time. To makes this pattern clearer, we plot the predicted probabilities for each group over time in figure 1. Here, we can see that in 1973 political conservatives showed a higher predicted probability in their confidence in scientists compared to both political liberals and political moderates. At this time, around 49 percent of conservatives showed a great deal of confidence in the scientific community. While at this point conservatives and liberals (46 percent predicted to be confident in the scientific community) were not statistically distinguishable, by 2016 the confidence in the scientific community among conservatives had decreased 10 percentage points to 37 percent.

Figure 1.

Trends in attitudes towards science (Source: General Social Survey, 1973–2016).

Open in new tabDownload slide

This dramatic trend can be juxtaposed to the fairly stable attitudes toward scientific research as a benefit to society. While the time frame for this outcome is shorter, we see a positive, if insignificant, trend for conservatives. This cross-sectional snapshot provides support for both Hypothesis 1 and Hypothesis 2. Attitudes toward scientists among conservatives are becoming more negative, while attitudes toward scientific research as a general benefit are remaining largely stable. Moreover, the slight upward trend in the “scientific research as a benefit” category among conservatives, though insignificant, might be indicative of the successful efforts of conservative scientific institutions to communicate findings and counterarguments to conservative audiences.

Our panel models formally test whether stability of political identity plays a role in shaping scientific perceptions. Table 3 presents the results from a series of lagged logistic regression models with change indicators. From the confidence in the scientific community models, we can see that relative to those who are never politically liberal, stable liberals show a significant coefficient. The probabilities from this model predict that 38 percent of “never liberals” are expected to show a great deal of confidence in the scientific community, while stable liberals show a predicted probability of around 47 percent.

While the significant coefficients in the table are in comparison to those who are never liberal, we ran a series of additional tests of coefficient equality to determine whether the coefficients we capture here show any differences from one another.¹⁸ Here, we find a marginal difference between stable liberals and those who become liberal (a 42 percent predicted probability of agreement for this latter group). Moreover, when comparing those who leave the liberal category to stable liberals, we find a significant difference between these groups with stable liberals showing significantly higher log odds of confidence in scientists compared to those who leave the liberal identity.

Among conservatives, we see a similar pattern. Relative to those who are never conservative (41 percent predicted probability), stable conservatives show a significantly lower probability of possessing confidence in the scientific community (36 percent predicted probability). The highest probability across those who are ever conservative is among “conservative leavers” (44 percent predicted probability), and while these individuals are not distinguishable from the “never conservative” group |$(p=0.15)$|⁠, they are significantly different from the stable conservatives. These results support previous research on conservative declines in confidence in the scientific community and provide evidence that these declines are most pronounced among those with stable conservative identities (H₃).

The additional models from Table 3 focus on those who strongly agree that scientific research benefits society. Among our liberal indicators, the stable liberals show significantly higher log-odds of agreement on this outcome. The predicted probabilities from these models show that around 50 percent of the “never liberals” are expected to agree compared to 58 percent of the stable liberals, 48 percent of the “become liberals,” and 55 percent of the “liberal leavers”. For the conservatives, the probabilities show that around 52 percent of the “never conservatives” are expected to agree compared to 58 percent of the “stable conservatives”, 47 percent of the “become conservatives”, and 48 percent of the “conservative leavers”. The difference on this outcome between stable identifiers and switchers seems even stronger for conservatives (H₄), demonstrating that a stable conservative identity is associated with higher confidence in scientific research despite its strong association with distrust in the scientific community.

Discussion and Conclusion

Confidence in the scientific community has declined among political conservatives in recent years but attitudes toward scientific research as a benefit to society have remained stable. Meanwhile, conservative social movements have established their own conservatively oriented scientific institutions (e.g., see Dunlap, Riley and McCright 2016; Dunlap and Jacques, 2013; Jacques, Dunlap, and Freeman, 2008; McCright & Dunlap, 2000, 2003, 2010; Gross et al., 2011) and the dawn of the Internet and social media has made it easier than ever for conservative audiences to access conservative knowledge. The preceding analysis aimed to show how these developments intersect by demonstrating that stable conservative partisans are more likely than their switching counterparts to distrust the scientific community and to believe that scientific research is a benefit to society. These findings support arguments that conservative efforts to communicate alternative scientific knowledge have been successful insofar as stable conservatives maintain trust in science while rejecting the authority of mainstream scientists. The implications of these developments are numerous.

First, this study replicates the findings of Gauchat (2012) and helps confirm one of the most dramatic trends in scientific perceptions in the last fifty years. Second, we build on previous work (O’Brien and Noy 2015; Roos 2017) that shows how rejections of mainstream scientific knowledge often signal specific cultural perceptions as opposed to deficits in scientific knowledge itself (although see Allum, Sturgis, Tabourazi, & Brunton-smith, 2008; Sturgis & Allum, 2004). We contribute to this work by studying political identity and scientific attitudes and finding that rejections of scientists need not be driven by a broader rejection of scientific research itself. This is further evidence that cultural communities viewed as being anti-science maintain a complex arrangement of scientific perceptions that can include high levels of scientific knowledge and positive views of scientific research. Furthermore, consistent identification in such a community can be indicative of positive scientific attitudes.

We are not the first to examine how membership in a cultural community affects perceptions of science. Moscivici (1961/2008) coined the concept of “social representations” by studying how the advent of psychoanalysis was received and communicated among three different moral communities in France—urban-liberals, Catholics, and Communists—and observing how new scientific ideas were refracted through the organizational and cultural lenses of these social milieus. This study extends this long line of research on cultural membership and scientific perceptions by examining the issue of consistency in political identity and attitudes toward scientists and scientific research, as opposed to interpretations of a distinct scientific discipline or the relationship between scientific knowledge and attitudes.

More specifically, this research applies Perrin et al.’s (2014) performative theory of conservative identity and extends their work by examining it in the context of identity stability. Identity stability is important for a performative theory of political identity because it reflects enduring familiarity with and acceptance of elite characterizations of political identity. In other words, if conservatives learn to be conservative (or if any partisan learns to be partisan), identity stability is a direct reflection of a period in which this learning can occur and the resilience of this identity through national political change. We find that consistent identification predicts having learned that it is scientists, and not science itself, that produce findings counter to conservative political goals. Furthermore, learning implies teaching and we have also argued that the pattern of attitudes shown here is indicative of successful social movement efforts to establish alternative and conservatively oriented institutions of knowledge (Gross et al. 2011). In this respect, we join other scholars in identifying the construction of politically partisan knowledge institutions as an important social movement outcome that has been under-studied among social movement scholars (Frickel and Gross 2005; Gross et al. 2011) and especially by those interested in framing processes (Benford and Snow 2000; Snow et al. 1986).

Several limitations to our empirical analysis warrant discussion. Most importantly, these data were not ideal for examining the mechanisms of engagement with conservative science explicitly. Computational researchers are well positioned to more accurately measure exposure to, and consumption of, conservative scientific information online. This type of work is well underway in the context of political news media (Barberá et al. 2015; Conover, Ratkiewicz, and Francisco 2011; Etling, Roberts, and Faris 2014; Faris et al. 2017; Guess, Nyhan, and Reifler 2018), but very little explores the impact of conservative scientific institutions. Think tanks like the Heritage Foundation and Discovery Institute, while unique in their missions and ideologies, offer politically conservative and religiously fundamentalist scientific resources to their audiences respectively, while partisan content creators like “Prager University” provide conservative information to subscribers with the veneer of an academic approach. But the effect of increased exposure to these kinds of partisan scientific resources—whose main point of public contact is through the Internet and social media—remains unclear.

In this article, we were not able to directly measure the consumption of conservative scientific information on the Internet, but we can offer some suggestive evidence that getting scientific information from the Internet makes a difference for stable and unstable conservative attitudes. Using a question that asks, “Where do you get most of your information about science and technology,” we can examine how using the Internet to consume scientific information affects differences between stable and unstable conservatives on our two dependent variables over time. Figure 2 shows these descriptive trends from 2006 to 2014 using fractional polynomial best fit trend lines with 95% confidence intervals. It is most important to note that stable conservatives that get their scientific information from the Internet are among the least likely to trust scientists over this timespan and the most likely, by a good margin, to see scientific research as a benefit. They are the group with the largest gap between their trust in scientists and belief in the benefits of scientific research.

Figure 2.

Descriptive trends in attitudes towards science among stable and unstable conservatives by science media outlet. Fractional polynomial best fit trend lines with 95% confidence interval in gray shading (Source: General Social Survey Panel, 2006–2014).

Open in new tabDownload slide

This aligns with our overall analyses—in that ostensibly greater access to partisan scientific authority exaggerates this gap for conservatives—but it remains a suggestive finding for future research to adjudicate more thoroughly. For instance, are these patterns really the result of better access to partisan science or is there something qualitatively different about online scientific content that exaggerates perceptions of scientists as over-stepping their authority (Evans 2018)? And in what ways are the populations getting their scientific information online different from others? Work in this vein could help answer important descriptive questions about conservative scientific sources, including how pervasive and heterogenous they are, and what associations exist between the sources themselves in terms of shared staff, audiences, and even content. A comprehensive study on public-facing scientific sources online could help map the cultural heterogeneity of scientific communication itself beyond the politically binary analysis provided here and provide a welcome point of comparison by suggesting other cultural scientific repertoires that are orienting and enabling of participation in scientific debates.

Future research should also include qualitative examinations of the conservative scientific repertoire. Differences and similarities in how stable liberals and conservatives, both groups that report high levels of belief in scientific research as a benefit to society, talk about and understand scientific issues is not well understood. Just as Swidler (2001) examined how people brought the universally valued concept of love to bear on their particular circumstances, future researchers can examine how political partisans selectively deploy “science” and its related concepts in their daily lives. This includes further examination into how attitudes toward scientists and scientific research are partitioned and how this disassociation is expressed or reconciled in the context of in-depth interviews. Scholars of religion and science (see e.g., Ecklund 2012; Ecklund and Scheitle 2017; Evans 2018) have been hard at work on questions like these and have set the stage for similar work on political partisans, including in non-US contexts.

These findings also raise questions about how cultural groups navigate moments of institutional trust and their relationships with other communities that may not support their worldview. The title of this article is a play on the (conservative) Christian saying, “Love the Sinner, Hate the Sin,”—a call to separate the actor (the sinner who might accept God’s forgiveness) and the action (the sin, which is against God’s will) in terms of one’s attitude toward a social performance (the sinner committing the sin). For our case, the process is inverted, with the political conservative showing low approval for the actor (the scientist) while maintaining a high approval for the action/process (the method of science). In both cases, individuals have the cognitive ability to separate actor and action in their evaluations, an ostensibly counter-intuitive process, and so the need for a snappy turn of phrase. Testing when and under what conditions people make striking actor/action distinctions in their evaluations is beyond the scope of this article. However, we demonstrate the integral role of identity and cultural membership in these processes, suggesting future research that might examine variation in actor/action evaluations among different cultural groups.

For example, we show how attitudes toward individual elites (scientists) are hurt, while attitudes toward the institutional practice (scientific research) are protected for stable political conservatives. But how this distinction between actors and action extends to other cultural groups and institutions depends on a variety of factors. Parallels might be found in how stable political liberals view capitalist institutions, where economic elites might be viewed unfavorably while belief in capitalism itself as an overall benefit to society remains stable. Other movements distrust elites and seek the abolition of entire institutions (e.g., anti-religious atheists), while still others distrust institutions while preserving positive attitudes toward individuals within them, as when political reactionary movements like the Tea Party or the Democratic Socialists successfully place leaders in elected political roles. This line of thinking suggests that actor/action distinctions are not indicative of conservativism itself or any kind of specifically conservative mentality (Mannheim 1993). We argue that one mechanism guiding the organization of these attitudes is whether an institution is politically useful (i.e., whether scientific appeals might help conservatives make political arguments) but further comparative studies can elucidate how different contexts shape attitudes toward individual elites and the institutions of which they are a part.

Finally, these results carry implications for science communication policy experts and strategists. Those conservatives most skeptical of man-made climate change and the scientists promoting it are also the most likely to believe that scientific research is a general benefit to society. Therefore, promoting policy that promotes the idea of science as a valid epistemology in order to increase belief in anthropogenic climate change seems misguided. Rather, outreach efforts might be more effective if geared toward humanizing the scientific community and correcting misperceptions of scientists themselves. By improving public agreement on where legitimate and trustworthy science is being accomplished, future debates at the intersections of science and politics can begin to focus more on what problems to prioritize instead of what the problems are.

About the Authors

Marcus Mann is an assistant professor of sociology at Purdue University. His research examines the political differentiation of knowledge authority in science and political media and is interested in political polarization and communication more generally. His research has appeared in Social Problems, Proceedings of the National Academy of Science, American Sociological Review, and Socius among others.

Cyrus Schleifer is an assistant professor of sociology at the University of Oklahoma. His research examines different statistical approaches to modeling religious change and within-occupational inequality. Some of his research has appeared in the American Sociological Review, Journal of Sociological Methods & Research, Social Science Research, Journal for the Scientific Study of Religion, and the Sociology of Religion.

Footnotes

References