Politicization Undermines Trust in Institutions, Even Among the Ideologically Aligned Public

doi:10.21203/rs.3.rs-3239561/v1

Six studies (five preregistered; total n = 5,925 U.S. adults), testing 40 institutions (e.g., journalists, the World Health Organization, police officers) and 30 academic disciplines (e.g., economists, psychologists, public health) found that perceived politicization—the extent to which political values impact an institution’s work—was associated with lower trust, lower willingness to defer to expertise, lower financial support, and greater skepticism. Institutions and disciplines perceived as the most politicized were also overwhelmingly the least trusted (Studies 1 & 4). Experimental evidence indicated that increasing politicization of a particular organization (e.g., Economics Professors of America) not only caused these negativeoutcomes, but also undermined trust toward entire broader professional groups (e.g., all economists in general; Studies 3 & 5). These negative relationships were observed among both participants who shared and opposed the institution’s ideological slant. In other words, both left- and right-leaning participants were less trusting of both left- and right-leaning institutions that appeared more politicized.Attempts to experimentally decrease perceived politicization mostly failed (Studies 2a, 2b & 5).Although institutions may have important and instrumental reasons for taking political stances, these data reveal that there are costs in trust and support among the entire ideological spectrum of the public.

Scientific community and society/Social sciences/Psychology/Human behaviour

Scientific community and society/Social sciences/Society

Scientific community and society/Social sciences/Decision making

Scientific community and society/Scientific community/Institutions

Scientific community and society/Social sciences/Interdisciplinary studies

Metascience

Trust

Politicization

Expertise

Organizational Behavior

Many institutions—from the World Health Organization and criminal justice system to journalists and scientists—depend on public trust to be successful. Yet public trust toward many essential institutions is near record lows. To explain these declines, some scholars have focused on ideological imbalances (e.g., journals and professors are left-leaning) and others have focused on the challenges of maintaining trust among the ideologically misaligned public (e.g., conservatives have lost trust in mainstream media and higher education). Here, we show the importance of public perceptions that political values impact an institution’s work, or “politicization” of institutions. Above and beyond ideological imbalances (which had inconsistent relations with trust), and ideological alignment between institutions and the public, we find that perceived politicization is a robust predictor and cause of lower trust in institutions across the ideological spectrum. Both the left- and right-leaning public are less trusting of both left- and right-leaning institutions that appear more politicized. The negative influence of perceived politicization further undermined public support for institutions, willingness to defer to their expertise, increased skepticism toward academic disciplines, and rippled out beyond specific organizations to undermine trust toward all members of a profession. These findings suggest that even in the absence of ideological diversity or ideological alignment with the public, minimizing perceptions of political interference in work might improve public trust toward essential institutions.

Many institutions depend on public trust to function properly¹. Groups providing public health recommendations (e.g. WHO, pharmaceutical companies) are less effective if the public disregards their recommendations^2-7, groups that inform (e.g. media organizations, universities, think tanks) have less impact when the public disbelieves the information they share^8-10, and institutions that protect public safety (e.g. police and the broader criminal justice system) may struggle to maintain law and order without public cooperation^11,12. Yet trust in institutions has declined in recent years^13,14. From the police to professors to public schools, trust has fallen to levels at or near all-time lows^15-20.

One possible explanation is that institutions have become politicized, or at least appear to have become politicized. A recent analysis found that physicians’ political ideology predicted their beliefs about the effectiveness of COVID-19 treatments²¹. Many scholars have contended that political values impact scientific decision-making^22,23, a concern shared by everyday people²⁴. Likewise, people often see the media as biased against their own side on political issues²⁵. Institutions are run by people after all, and political values can impact human judgment and decision-making^26-28. The public may believe—correctly or not—that political values also impact institutional decision-making and sometimes interfere with institutions’ stated objectives, such as “report the news,” “discover empirical reality,” “interpret the constitution,” or “enforce the law.”

This perceived politicization may be undermining trust. For example, highlighting the politicization of scientific information undermined perceived scientific consensus surrounding anthropogenic climate change²⁹. The politicization of COVID-19 information was associated with distrust toward the media and government³⁰, and mentions of political conflict over HPV vaccines were associated with declines in confidence toward doctors³¹. A recent experiment found that exposure to a real political candidate endorsement of Joe Biden by the science journal Nature reduced trust both in the journal Nature and scientists in general³².

In six studies, we test how perceived politicization of institutions impacts public trust, support, and willingness to defer to expertise for a wide-ranging set of institutions, organizations, groups of professionals, and academic disciplines. We experimentally test the causal impact of politicization and the broader consequences for professional groups. We also test whether these effects vary among members of the public who are more or less ideologically aligned with institutions. On one hand, people might prefer when ideologically-aligned institutions pursue political values in their work because this would support their own ideological beliefs and goals. On the other hand, people might oppose institutions using their power and influence to enforce their own political agenda or consider the non-political mission of institutions too important to risk alienating a substantial subset of the population. In these cases, people might resent all stripes of politicization, even that which supports their own side. Thus, we tested two competing hypotheses: (1) Participants evaluate ingroup institutions as more trustworthy the more they are perceived as politicized, and evaluate outgroup institutions as less trustworthy the more they are perceived as politicized or (2) Participants evaluate both ingroup and outgroup institutions as less trustworthy the more they are perceived as politicized. The former hypothesis would indicate that politicization is polarizing, increasing trust among ingroup members but reducing trust among outgroup members. The latter would suggest that apparent politicization undermines trust almost universally.

Study 1 reports the results of a politically-balanced survey in the U.S. testing how perceived politicization relates to trust towards a broad set of institutions. Studies 2a, 2b and 3 use experimental methods to test the causal impact of politicization on a smaller set of institutions. Studies 4 and 5 replicate these effects among academic disciplines. We highlight the most important results in the main text and Extended Data, but full results and additional discussion for all six studies are included in the Supplement.

Study 1: Institutions Survey

In an initial exploratory study, 621 American participants evaluated a subset of 40 institutions, organizations, and groups of professionals (henceforth referred to as institutions) on (1) perceived ideological slant, or the proportion of liberal/conservative individuals within the institution, (2) perceived politicization, or the amount to which political values influence the institution’s work, and (3) trustworthiness, and they reported their own political ideology.

Table 1 reports descriptive statistics for perceived slant (organized right-leaning to left-leaning), perceived politicization (organized high to low), and trust (organized low to high) for each group, and Table S1 in the Supplement reports correlations between trust, politicization, slant, and participant (Ps) ideology for all institutions. Higher perceived politicization was associated with lower trust for 34 of the 40 institutions (rs=-0.57 to 0.01, M_r=-0.28). There was no consistent pattern between perceived ideological slant (i.e., proportion of conservatives vs. liberals) and trust (rs=-0.28 to 0.38, M_r=0.06).

Table 1 Descriptives for Slant, Politicization, and Trust Across Institutions (ns 307-309) in Study 1

Slant from right to left			Politicization from high to low			Trust from low to high
Group	M	SD	Group	M	SD	Group	M	SD
Catholic Church	75.06	19.61	Congress	5.52	0.84	Facebook	1.74	0.96
Police Officers	70.67	16.49	Supreme Court	5.01	1.36	Congress	2.00	1.13
Supreme Court	64.56	16.01	United Nations	4.78	1.34	State Judges	2.02	1.13
Banks	63.36	20.31	Journalists	4.77	1.19	Pharm. Companies	2.27	1.31
Mechanics	62.19	16.83	Crim. Just. System	4.46	1.36	Journalists	2.62	1.31
Firefighters	61.36	13.94	Catholic Church	4.40	1.44	Catholic Church	2.71	1.63
Local Business Owners	57.93	13.49	Facebook	4.39	1.32	NFL	2.77	1.18
Crim. Just. System	57.58	19.27	State Judges	4.38	1.35	Crim. Just. System	2.81	1.38
State Judges	55.78	13.77	Hollywood	4.33	1.45	Pollsters	2.83	1.34
Pharm. Companies	55.31	23.30	PETA	4.29	1.60	PETA	2.83	1.61
MLB	54.86	14.41	Professors	3.98	1.40	CIA	2.84	1.65
Real Estate Agents	53.90	13.43	CIA	3.92	1.56	Think Tanks	2.87	1.20
Dentists	53.66	10.56	Pollsters	3.84	1.54	Lawyers	2.93	1.33
Park Rangers	52.35	15.56	Police Officers	3.79	1.52	Supreme Court	2.94	1.65
Economists	52.26	15.83	Think Tanks	3.76	1.65	Banks	2.96	1.50
CIA	51.88	21.34	Pharm. Companies	3.73	1.58	United Nations	3.11	1.45
Lawyers	50.37	13.95	WHO	3.69	1.60	Real Estate Agents	3.18	1.12
Doctors	48.89	12.71	K12 Teachers	3.63	1.36	Police Officers	3.26	1.70
Congress	48.28	14.70	Lawyers	3.49	1.36	Hollywood	3.27	1.33
Toll Booth Workers	48.26	11.82	Economists	3.42	1.33	MLB	3.38	1.31
Tailors	47.50	11.59	Banks	3.41	1.61	Economists	3.49	1.16
USPS	47.45	12.27	Psychologists	3.02	1.38	Mechanics	3.53	1.18
NFL	46.69	20.11	Local Business Owners	3.02	1.27	Professors	3.72	1.43
Think Tanks	46.49	16.65	Scientists	2.91	1.51	WHO	3.74	1.76
Veterinarians	44.26	13.66	ASPCA	2.85	1.52	Local Business Owners	3.98	1.03
Pollsters	43.00	14.11	Librarians	2.80	1.37	Psychologists	4.02	1.26
Chefs	41.07	12.85	NFL	2.70	1.39	K12 Teachers	4.14	1.35
Physicists	39.82	15.41	Doctors	2.55	1.36	ASPCA	4.17	1.44
United Nations	38.15	18.15	Real Estate Agents	2.37	1.23	USPS	4.18	1.32
Facebook	36.76	24.09	Physicists	2.22	1.26	Toll Booth Workers	4.19	1.18
Scientists	36.55	15.30	Park Rangers	2.20	1.22	Chefs	4.35	1.12
K12 Teachers	35.95	15.82	Mechanics	2.17	1.25	Scientists	4.41	1.56
Librarians	35.69	20.21	Firefighters	1.98	1.16	Tailors	4.41	1.09
Psychologists	34.23	14.98	Chefs	1.95	1.15	Doctors	4.44	1.36
ASPCA	34.12	18.07	USPS	1.87	1.19	Dentists	4.51	1.22
Journalists	34.11	16.59	MLB	1.85	1.12	Physicists	4.56	1.41
WHO	33.26	18.03	Veterinarians	1.79	1.06	Park Rangers	4.66	1.22
Professors	31.00	17.71	Dentists	1.73	0.99	Veterinarians	4.75	1.22
PETA	24.63	21.99	Tailors	1.68	1.00	Firefighters	4.89	1.30
Hollywood	22.54	19.21	Toll Booth Workers	1.53	1.00	Librarians	4.93	1.34

Treating institutions as the unit of analysis, there was also a very strong negative association between perceived politicization and trust (r = -.76, t = -7.16, p < 0.001, 95% CI = [-0.87, -0.58]), such that organizations with higher group-level averages in perceived politicization also were viewed as less trustworthy on average. For ease of visualization, Figure 1 maps this relationship. There was no significant association between ideological slant and trust, (r = -.15, t = -0.91, p = 0.369, 95% CI = [-0.44, 0.17]).

Next, we regressed participant ideology, perceived slant, perceived politicization, and all two- and three-way interactions on trust for all 40 institutions (full results presented in Table S2 of Supplement). In this and all subsequent analyses, we interpret significant effects as those that reach a minimum threshold of |semipartial r| ³ .10. Participant ideology predicted trust for 26 of 40 institutions (rs = -.38 to .40). Perceived slant predicted trust for only five institutions and with small effects (rs = -.12 to .13), indicating that an imbalance of liberals and conservatives alone had little relation to trust. However, the interaction between participant ideology and perceived slant predicted trust (rs = .04 to .36) for 34 institutions, such that greater congruence between participant ideology and perceived slant (hereafter, referred to as ideological congruence or ideological alignment) predicted greater trust: left-leaning participants were more trusting of institutions perceived as having a left-slanted workforce (e.g., K12 teachers, journalists, professors), and right-leaning participants were more trusting of institutions perceived as having a right-slanted workforce (e.g., police officers, the Catholic Church, banks).

Turning to the critical question, greater perceived politicization predicted lower trust for 32 of 40 institutions, with effect sizes similar to those for ideological congruence (rs = .01 to -.33), indicating that perceived politicization was similarly important for trust as ideological alignment. Outside of ideological congruence, the other two- and three-way interactions were rarely significant, indicating that for most institutions, perceived politicization was associated with reduced trust regardless of whether participants saw the institution as ideologically congruent. In other words, both left- and right-leaning participants trusted both left- and right-leaning institutions less the more they were perceived as politicized. For discussion of the three significant three-way interactions, please refer to the Supplement.

Item E1 in the Extended Data maps the interactions between ideological congruence between participant and institution (whether participants view the institution as on the same side of the ideological spectrum as they view themselves) and trust for all 40 institutions. Two patterns are worth noting: (1) politicization tended to be associated with lower trust, and (2) the negative associations between politicization and trust tended to be weaker among ideologically aligned (vs. misaligned) participants, but the associations still tended to be negative nonetheless (negative coefficients for 35 of the 40 institutions, Wilcoxon Signed Rank p < 0.001).

To summarize, across 40 diverse institutions, we found that perceived politicization tended to be associated with lower public trust, often substantially so. To no surprise, participants were also less trusting of institutions they perceived as slanted against their own ideology. Perhaps more surprising, however, was that perceived politicization tended to be associated with lower trust for both ingroup and outgroup institutions. Or in other words, when left-leaning institutions were perceived as politicized, even left-leaning participants trusted them less, and when right-leaning institutions were perceived as politicized, even right-leaning participants trusted them less.

Studies 2a, 2b and 3: Institutions Experiments

To test the causal effect of politicization on trust, as well as on willingness to defer to institutions’ expertise, we conducted three experiments. Study 2a focused on six candidate (and real) organizations that, based on similar institutions in Study 1, were likely to be perceived as left-leaning (American Association of University Professors, the WHO), politically balanced (International Association of Fire Fighters, American Medical Association), or right-leaning (The National Association of Police Organizations, American Judges Association). Study 2b focused on just the professor and police organizations. In control conditions, participants merely read about the organization. To attempt to manipulate perceptions of politicization, in Studies 2a and 2b, participants were presented with statements reporting that upon discovering the ideological composition of their organization, the organization either intended to take efforts to ensure the political values of their members did not affect the work (depoliticization condition) or commented on how the ideological composition helps them pursue their shared values and priorities in their work (politicization condition).

Across Studies 2a and 2b, these manipulations largely failed to change perceptions of politicization, rendering the studies ill-suited for testing causal hypotheses about the impact of perceived politicization. Nevertheless, the correlational results replicated the findings from Study 1, and extended these results to a new outcome: willingness to support the institution and defer to their expertise. Perceived politicization was associated with lower trust, rs = -.16 to -.54, and support/deference, rs = -10 to -.55, for all institutions across the two studies. When regressing participant ideology, perceived slant, perceived politicization, and all two- and three-way interactions on trust and deference, results also largely followed the same pattern as Study 1. Perceived politicization continued to predict lower trust and deference across all institutions in both studies, rs = -.10 to -.37. There were modest interactions between ideology and politicization on trust and deference in 11 out of the 16 regressions, rs = -.20 to .15, indicating that the negative relationships between politicization with trust and deference were weaker when participants were more ideologically aligned with institutions. Nevertheless, in almost all cases, the associations were still negative. See Item E2 in the Extended Data to visualize these relationships and the Supplement for discussion and full tables of all results.

One reason the manipulations failed to change perceptions of politicization may be that people already had strong intuitions about the degree of politicization of these institutions, making experimental attempts at depoliticization untenable. Thus, in Study 3, we sought to test whether increasing politicization of an unfamiliar and politically neutral organization undermined trust. In Study 1, Park Rangers were rated as trustworthy, low in politicization, and ideologically balanced. And so in Study 3, participants read about a fictional organization, Get Outdoors: Park Rangers of America, whose mission was “Helping America experience the awe of our national parks.” Participants were randomized into three conditions, in which the organization endorsed Joe Biden or Donald Trump for the 2020 election, or a control condition with no such endorsement. And participants responded to the same questions as in Studies 2a and 2b, which were then repeated for park rangers in general to test whether politicization of Get Outdoors also undermined trust and deference toward the entire professional group of rangers. See the Supplement for full results.

This manipulation was successful: participants in the control condition saw the organization as less politicized than those who read about the organization endorsing either presidential candidate, ps<.001, whereas the two politicization conditions did not differ, p = .636.

As seen in Item E3 in the Extended Data, the Biden and Trump endorsements (vs. control) significantly reduced trust and deference toward Get Outdoors with very large effects, η²s= .29-.34, simple contrasts ps<.001, and significantly reduced trust and deference toward all park rangers in general with medium to large effects, η²s= .06-.13, simple contrasts ps<.001, except the Biden endorsement only marginally reduced trust toward rangers in general, p = .077.

Similarly, in regressions (see Item E4 in Extended Data), the Trump and Biden endorsement dummy variables predicted lower trust and deference to both Get Outdoors and park rangers in general, rs = -.20 to -.56 (except for the Biden dummy on trust toward all rangers, which was trending in the same direction, r = -.09). The interactions between the Trump and Biden endorsement dummy variables and participant ideology predicted more negative outcomes in three out of eight cases, rs = -.10 to .27. As seen in Figure 2 Panel A, although outgroup endorsements tended to undermine trust and deference more than ingroup endorsements, endorsements tended to have a negative effect regardless of ideological alignment.

As can be seen in Item E5 in the Extended Data, similar patterns emerged if we tested moderation by participant candidate endorsement in the 2020 election rather than by participant ideology: Among Trump supporters, Biden supports, and participants who supported neither Trump or Biden, trust and deference toward both Get Outdoors and all park rangers were always highest in the control condition. Outgroup endorsements again had larger negative effects, and for the one outcome “trust toward all rangers”, ingroup endorsements did not significantly differ from the control, but in no cases and for no groups did political endorsements improve outcomes, and in most cases, political endorsements worsened them. In other words, even Trump supporters reacted negatively to Get Outdoors and all park rangers when Get Outdoors endorsed Trump, and even Biden supporters reactively negatively to Get Outdoors and all park rangers when Get Outdoors endorsed Biden.

A final question asked participants to vote on a $100 donation to one of three organizations—Get Outdoors, or two other organizations they were hearing about for the first time (Firefighters for Community Risk Prevention and National Dentists for Oral Health). Endorsements of any kind reduced votes for Get Outdoors. In the control condition, participants were roughly twice as likely to vote for Get Outdoors (48.1%) compared to the other two organizations (24.2%, 21.7%), whereas in the Biden endorsement condition, they were similarly likely to vote for Get Outdoors (35.1%) and the other two organizations (31.8%, 27.5%), and in the Trump endorsement, they were roughly one third as likely to vote for Get Outdoors (16.8% vs. 44.0% and 50.7%), χ2= 64.72, p<.001.^{^[1]} Again, the costs of politicization were seen even when it involved an endorsement of one’s own preferred candidate. Trump supporters’ votes for Get Outdoors fell from 72.5% in the control condition to 54.1% when the organization endorsed Trump (and 33.3% when they endorsed Biden). Biden supporters’ votes for Get Outdoors fell from 63.5% in the control condition to 52.3% when the organization endorsed Biden (and 12.6% when they endorsed Trump). Among those who did not support Biden or Trump, votes went from 45.3% in the control to 37.1% in the Biden condition and 20.0% in the Trump condition. See Figure 2 Panel B.

Study 4: Academic Disciplines Survey

One of the industries in which there has been a great deal of concern surrounding ideological slant, ideological alignment, politicization, trust, and deference to expertise is academia^23,33,34. Study 4 replicated a design similar to Study 1 using a set of 30 academic disciplines to test whether politicization is similarly costly for academic disciplines. We also tested the relationships between perceived politicization and beliefs about how much skepticism students should have about what they are taught by professors in each discipline. Descriptives for all primary variables are available in Item E6 in the Extended Data.

Results largely replicated those from Study 1. Higher perceived politicization was associated with lower trust, rs = -.16 to -.37, and higher skepticism across all 30 disciplines, rs = .24 to .48, all ps<.001. At the discipline level, as can be seen in Figure 3, disciplines perceived as more politicized were trusted much less (r = -.88, p < 0.001, 95% CI = [-0.94, -0.76]) and were far more likely to elicit skepticism (r = .80, p < 0.001, 95% CI = [0.62, 0.90]). Ideological slant was again mostly unrelated to trust (r = .05, p = 0.776, 95% CI = [-0.31, 0. 41]) and skepticism (r = -.13, p = 0.489, 95% CI = [-0.47, 0.24]).

We regressed participant ideology, slant, politicization, and all two- and three-way interactions on trust and skepticism. Politicization consistently predicted lower trust and higher skepticism across all 30 disciplines. The one exception was for trust toward Religious Studies, which, at r = -.097, did not quite hit a minimum threshold of |r| ≥ .10, but was in the same direction and statistically significant, p = .027. Perceived slant (more right-leaning) predicted higher trust for only four (highly left-leaning) disciplines and did not predict skepticism.

The interactions between participant ideology and politicization reached the minimum threshold for trust only in environmental science and mathematics and for skepticism only in music and art, but with small effects. Critically, even among congruent participants—those who saw disciplines as slanted in the same direction as their own ideology—greater politicization was associated with lower trust (for 25 of the 30 disciplines, Wilcoxon Signed-Rank p < 0.001) and higher skepticism (for all 30 disciplines, Wilcoxon Signed-Rank p < 0.001). Similar to what was seen for organizations, institutions, and groups of professionals, perceived politicization was generally associated with lower trust and higher skepticism toward academic disciplines regardless of participants’ ideological alignment with the discipline. Item E7 in the Extended Data displays the interactions between ideological congruence and perceived politicization on trust. Figure 4 Panel A displays the interactions between ideological congruence and perceived politicization on skepticism, and Panel B plots the coefficients of the slopes between politicization and trust and skepticism for congruent, incongruent, and neutral participants in Studies 1 and 4. The Supplement reports full results for Study 4.

Study 5: Academic Disciplines - Experiments

Last, we conducted an experiment to test whether the politicization or depoliticization of a fictional academic organization would lead to changes in trust and skepticism. Participants read about a fictional economics academic society (Economics Professors of America [EconPA]) and were randomly assigned to either a control condition or one of three experimental conditions that included extra information: Republican politicization, Democrat politicization, depoliticization. In the politicization conditions, the society leadership invited a former Democratic or Republican governor to deliver the keynote address at their annual meeting, and the speech was followed by a public discussion with the society’s advisory board on how Democratic or Republican values should shape the society’s research agenda. In the depoliticization condition, the keynote was the director of a fictional organization called Political Neutrality and Tolerance in Science, and the advisory board discussed a number of funded initiatives to prevent the influence of politics on research. Participants then responded to the same politicization and trust questions as Study 4 for both EconPA and economics professors in general, and the same skepticism item as in Study 4 for economics professors in general. Also, as in Study 3, participants voted to donate to one of three organizations—EconPA, Political Neutrality and Tolerance in Science, or one they had not yet heard of (The American Society for the Preservation of Historical Literature).

The manipulation was again successful, with EconPA evaluated as more politicized in the two politicization conditions (which did not differ from each other, p = 1.00) than the control condition, ps<.001, which was evaluated as more politicized than the depoliticization condition, p<.001. In a MANOVA (see Item E8 in Extended Data), participants in both the Democrat and Republican politicization conditions (vs. control condition) trusted EconPA and all economists less and thought students should be more skeptical of all economists, all simple effect ps<.021. The depoliticization condition, however, had no positive impact on these outcomes compared to the control condition, simple effect ps = 1.00. Unlike Study 3, however, the experimental manipulation had no significant impact on donation voting behavior, c² = 9.11, p < .168, possibly due to floor effects (EconPA was disfavored across all conditions including the control, receiving only 14.8%-19.8% of the vote across conditions).

We next tested the interactions between participant ideology and dummy-coded condition variables in a series of regressions (see Item E9 in the Extended Data). The two politicization dummy variables predicted lower trust toward EconPA and lower trust and higher skepticism toward economists in general. Only for trust toward EconPA was this relationship moderated by ideology (see Figure 5), such that ingroup politicization had a smaller negative impact on trust than outgroup politicization. Nonetheless, following the pattern seen in the previous studies, the politicization conditions tended to predict lower trust toward EconPA and lower trust and higher skepticism toward economists in general across the ideological spectrum—with most conservatives reporting lower trust even when EconPA was politicized in a conservative direction, and most liberals reporting lower trust even when EconPA was politicized in a liberal direction.[2] See the Supplement for full results.

[1]Note in the preregistration we said we would analyze this is a dichotomous outcome, but because this approach resulted in a loss of information and did not alter the interpretation of the results, we retained the three possible donation options.

[2] Even the most extreme partisans in our sample—those who selected a 1 (1.44 SDs below the mean) or 7 (1.98 SDs above the mean) on a 7-point scale—did not report significantly increased trust in the condition where EconPA endorsed the values of their political ingroup relative to the control condition (ps > 0.69, Wilcoxon Rank Sum Tests), though this analysis was not preregistered and our sample size at these points became quite small (ns = 12 to 34 per condition).

In six studies, we found that perceived politicization—the extent to which political values impact work—of institutions and academic disciplines was associated with lower trust, lower willingness to defer to expertise, lower financial support, and greater skepticism. We also found that increased politicization caused these negative outcomes, that these negative outcomes were observed even among the ideologically aligned, and that these negative outcomes extended out from individual politicized organizations to entire associated professional groups. This indicates that the politicization of one organization can negatively impact trust towards related groups of professionals who may not be politicized themselves.

Across studies, ideological slant was rarely an important predictor of trust, whereas politicization was often as important or even more important than ideological alignment. And, within academic disciplines, politicization was a more robust predictor of trust than left-leaning ideology. Although scholars have argued that ideological imbalances of institutions such as journalism, higher education, and many scientific institutions (or corresponding ideological misalignment with conservatives) are to blame for lower trust among the public (and especially conservatives), we see that perceived politicization may be as or more important a determinant for whether individuals trust institutions and sciences.

We also had relative difficulty experimentally decreasing perceived politicization and subsequently increasing trust, whereas increasing perceived politicization and subsequently decreasing trust was easier. These findings suggest the possibility that individual instances of apparent politicization may tarnish the reputation of an institution, and that once an institution or discipline has this reputation, it may be difficult to undo.

Many institutions may be motivated to get involved in contemporary political issues, especially when they believe their relevant community (be it employees, customers, consumers, or audience) is relatively ideologically homogeneous and has predictable political desires. Although these efforts may achieve certain important and instrumental goals, our results suggest that if these efforts result in perceptions of politicization, they may involve tradeoffs with costs in public deference, financial support, and trust. Furthermore, although we had difficulty reducing perceived politicization, our findings suggest the possibility that even ideologically imbalanced or misaligned institutions could conceivably gain public trust if they could maintain political neutrality, and subsequently, the appearance of it, in their work.

This research was approved by the Behavioural Research Ethics Board at University of British Columbia.

Fulmer, C. A. & Gelfand, M. J. At what level (and in whom) we trust: Trust across multiple organizational levels. J Manage 38, 1167-1230 (2012).
Bicchieri, C. et al. In science we (should) trust: Expectations and compliance across nine countries during the COVID-19 pandemic. PLoS One 16, e0252892 (2021). https://doi.org:10.1371/journal.pone.0252892
Caplanova, A., Sivak, R. & Szakadatova, E. Institutional trust and compliance with measures to fight COVID-19. Int Adv Econ Res 27, 47-60 (2021).
Hornsey, M. J., Lobera, J. & Diaz-Catalan, C. Vaccine hesitancy is strongly associated with distrust of conventional medicine, and only weakly associated with trust in alternative medicine. Soc Sci Med 255, 113019 (2020). https://doi.org:10.1016/j.socscimed.2020.113019
Jones, P. & Menon, A. Trust in Religious Leaders and Voluntary Compliance: Lessons from Social Distancing during COVID‐19 in Central Asia. J Sci Stud Relig 61, 583-602 (2022).
Schmelz, K. Enforcement may crowd out voluntary support for COVID-19 policies, especially where trust in government is weak and in a liberal society. Proc Natl Acad Sci U S A 118 (2021). https://doi.org:10.1073/pnas.2016385118
Clark, C., Davila, A., Regis, M. & Kraus, S. Predictors of COVID-19 voluntary compliance behaviors: An international investigation. Glob Transit 2, 76-82 (2020). https://doi.org:10.1016/j.glt.2020.06.003
Hamilton, L. C., Hartter, J. & Saito, K. Trust in scientists on climate change and vaccines. Sage Open 5, 2158244015602752 (2015).
Strömbäck, J. et al. News media trust and its impact on media use: Toward a framework for future research. Ann Int Commun Assoc 44, 139-156 (2020).
Wilholt, T. Epistemic trust in science. Brit J Philos Sci (2013).
Goldsmith, A. Police reform and the problem of trust. Theor Criminol 9, 443-470 (2005).
Murphy, K., Mazerolle, L. & Bennett, S. Promoting trust in police: Findings from a randomised experimental field trial of procedural justice policing. Polic Soc 24, 405-424 (2014).
Kennedy, B., Tyson, A. & Funk, C. Americans’ trust in scientists, other groups declines. Pew Research Center (2022).
Twenge, J. M., Campbell, W. K. & Carter, N. T. Declines in trust in others and confidence in institutions among American adults and late adolescents, 1972-2012. Psychol Sci 25, 1914-1923 (2014). https://doi.org:10.1177/0956797614545133
Brenan, M. Americans’ confidence in major US institutions dips. Gallup Polls (2021).
Brenan, M. Americans’ trust in media remains near record low. Gallup News (2022).
Doherty, C. & Kiley, J. Americans have become much less positive about tech companies’ impact on the US. Pew Research Center (2019).
Saad, L. Historically Low Faith in U.S. Institutions Continues. Gallup News (2023).
Jones, J. M. Confidence in higher education down since 2015. Gallup News (2018).
Jones, J. M. Trust in Federal Government Branches Continues to Falter. Gallup News (2022).
Levin, J. M., Bukowski, L. A., Minson, J. A. & Kahn, J. M. The political polarization of COVID-19 treatments among physicians and laypeople in the United States. Proc Natl Acad Sci U S A 120, e2216179120 (2023). https://doi.org:10.1073/pnas.2216179120
Clark, C. J. & Winegard, B. M. Tribalism in war and peace: The nature and evolution of ideological epistemology and its significance for modern social science. Psychol Inq 31, 1-22 (2020).
Duarte, J. L., Crawford, J. T., Stern, C., Haidt, J., Jussim, L. & Tetlock, P. E. Political diversity will improve social psychological science. Behav Brain Sci 38, e130 (2015). https://doi.org:10.1017/S0140525X14000430
McLaughlin, D. M., Mewhirter, J. & Sanders, R. The belief that politics drive scientific research & its impact on COVID-19 risk assessment. PLoS One 16, e0249937 (2021). https://doi.org:10.1371/journal.pone.0249937
Barnidge, M. et al. Politically motivated selective exposure and perceived media bias. Commun Res 47, 82-103 (2020).
Clark, C. J., Liu, B. S., Winegard, B. M. & Ditto, P. H. Tribalism is human nature. Curr Dir Psychol Sci 28, 587-592 (2019).
Haidt, J. The righteous mind: Why good people are divided by politics and religion. (Vintage, 2012).
Finkel, E. J. et al. Political sectarianism in America. Science 370, 533-536 (2020). https://doi.org:10.1126/science.abe1715
Bolsen, T. & Druckman, J. N. Do partisanship and politicization undermine the impact of a scientific consensus message about climate change? Group Process Interg Relat 21, 389-402 (2018).
Van Scoy, L. J. et al. Public anxiety and distrust due to perceived politicization and media sensationalism during early COVID-19 media messaging. J Commun Healthc 14, 193-205 (2021).
Fowler, E. F. & Gollust, S. E. The content and effect of politicized health controversies. Ann Am Acad Polit SS 658, 155-171 (2015).
Zhang, F. J. Political endorsement by Nature and trust in scientific expertise during COVID-19. Nat Hum Behav 7, 696-706 (2023). https://doi.org:10.1038/s41562-023-01537-5
Chilton, A. S. & Posner, E. A. An empirical study of political bias in legal scholarship. J Legal Stud 44, 277-314 (2015).
Honeycutt & Freberg. The liberal and conservative experience across academic disciplines: An extension of Inbar and Lammers. Soc Psychol Pers Sci (2017).

Open Practices Statement

Study 1 was exploratory and was not preregistered. Studies 2a (https://aspredicted.org/3PZ_WK1), 2b (https://aspredicted.org/K58_4WG), 3 (https://aspredicted.org/FPD_THN), 4 (https://aspredicted.org/P3P_YDZ) and 5 (https://aspredicted.org/5R8_6JT) were preregistered. Methods were followed as described except the online platforms slightly overrecruited participants. Only a subset of planned analyses are reported in the main text for space, but full results are reported in the Supplement. Data, analysis code, Qualtrics materials, and preregistrations are available here: https://osf.io/2auj6/?view_only=56584edcc0ad4205aacb64ea5445600c. All data that were ever collected to test the present research question are reported—there are no file-drawered studies.

Study 1

Participants. We recruited an ideologically balanced U.S. sample from Prolific Academic (n = 621; M_age = 38.78, SD = 14.65; 306 men, 300 women, 10 non-binary, 5 other or missing; M_conservatism = 3.89, SD = 2.13).

Procedure. Participants were randomly assigned to respond to questions about one of two sets of 20 institutions (see Table 1 for all institutions). We measured ideological slant by having participants first estimate the political breakdown of each institution on a sliding scale from 0 (nearly all liberals) to 100 (nearly all conservatives), and measured perceived politicization by having participants indicate how much they thought political values affected the institution’s work on a scale from 1 (not at all) to 6 (extremely). After participants completed these two questions for all 20 institutions, they reported how trustworthy each was on a 1 (not at all trustworthy) to 6 (extremely trustworthy) scale. Last, participants reported some demographic information, including their political ideology reported on a 1 (very progressive/left) to 7 (very conservative/right) scale.

Study 2a

Participants. We recruited an ideologically balanced U.S. sample from Prolific Academic (n = 1211; M_age = 45.55, SD = 16.14; 578 men, 589 women, 18 non-binary, 9 other or missing; M_conservatism = 3.14 out of 7, SD = 1.63).

Procedure. Participants read a statement about three out of six real organizations and the professionals within them. Based on participant ratings of similar institutions in Study 1, we selected two institutions considered left-leaning (American Association of University Professors, the WHO), two considered politically balanced (International Association of Fire Fighters, American Medical Association), and two considered right-leaning (The National Association of Police Organizations, American Judges Association). Aside from their perceived political breakdowns, these institutions were chosen because their members receive public funding and/or make public recommendations—topics asked about among our dependent measures. Participants were randomly assigned to read about one of the two organizations from each of the three ideological slant categories. This manipulation was crossed with a politicization manipulation, such that each organization was paired with a politicization statement, a depoliticization statement, or no statement (control), so that each participant received each condition just once. See sample statements below.

Politicization: An association-wide survey of the American Association of University Professors discovered that over 82% of university professors identified as politically to the left. The association remarked that this ideological composition helps them pursue their shared values and priorities in their professional activities and human resource management.

De-politicization: An association-wide survey of the American Association of University Professors discovered that over 82% of university professors identified as politically to the left. The association has forwarded a new policy agenda to ensure their political leanings do not impact their professional activities or human resource management.

Control: An association-wide survey of the American Association of University Professors discovered that over 82% of university professors identified as politically to the left.

Participants were first asked to estimate the political breakdown on a scale from 1 (completely liberal) to 7 (completely conservative) as a measure of ideological slant, and to rate politicization and trustworthiness on the same items as in Study 1. Participants reported support and deference toward organization: “If you were to vote on whether this group should receive more or less public funding, how would you vote?” on a scale from 1 (large funding cut) to 6 (large funding increase), “If this group were to make public recommendations, how likely would you be to take their advice?” on a scale from 1 (very unlikely) to 6 (very likely), “When this group makes decisions, should the public challenge those decisions, or defer to/accept those decisions?” on a scale from 1 (definitely challenge) to 6 (definitely defer), as=.674-.918. Last, participants reported demographic information.

Study 2b

Participants. We recruited an ideologically balanced U.S. sample from Prolific Academic (n = 1806; M_age = 38.58, SD = 13.08; 927 men, 829 women, 47 non-binary, 8 other or missing; M_conservatism = 3.82 out of 7, SD = 1.83). As preregistered, we excluded participants who failed the politicization attention check (85 in the professor conditions, 63 in the police conditions), resulting in a final sample of 1658.

Procedure. Procedures were nearly identical to Study 2a, except for two changes. First, participants were randomly assigned to read only about either the police or professor organizations, and to receive only one of the three politicization conditions, for a fully between-subjects design. See sample statements below.

Politicization: An association-wide survey of The National Association of Police Organizations (a large group of police officers in the United States) discovered that over 82% of police officers identified as politically to the right/conservative. The association shared the results of this survey with their members in an email along with another survey asking these police officers their views on the results.

The results of this new survey showed that police officers were generally not concerned about this political imbalance. Instead, the majority of members agreed that this ideological composition helps them pursue their shared values and priorities in their professional activities and hiring practices. And most members reported that it was appropriate for their work to support a right-leaning agenda. The association decided no action was needed.

Depoliticization: An association-wide survey of The National Association of Police Organizations (a large group of police officers in the United States) discovered that over 82% of police officers identified as politically to the right/conservative. The association shared the results of this survey with their members in an email along with another survey asking these police officers their views on the results.

The results of this new survey showed that police officers generally were concerned about this political imbalance. The majority of members agreed that this ideological composition could interfere with their professional activities and hiring practices and that this should be proactively avoided. And most members reported that it was not appropriate for their work to support any political agenda. With support from their members, the association empowered a new committee to detect possible political influence on their practices in order to preempt any potential bias.

Control: An association-wide survey of The National Association of Police Organizations (a large group of police officers in the United States) discovered that over 82% of police officers identified as politically to the right/conservative. The association shared the results of this survey with their members in an email along with another survey asking these police officers their views on the results. The results of this new survey showed that police officers had various reactions to this political imbalance.

Second, immediately following the vignette, participants responded to one attention check regarding the politicization manipulation (included in Study 2b only to rule out the possibility that our failed manipulation in Study 2a resulted from inattention), “Did professors/police think it was appropriate for their work to support a political agenda?” (response options: Yes, Not Sure, No). As preregistered, participants in the politicization conditions were excluded if they did not select “yes”, participants in the depoliticization conditions were excluded if they did not select “no”, and participants in the control condition were not excluded regardless of their response because no clear reaction was specified. Participants then responded to the same questions as in Study 2a.

Study 3

Participants. We recruited an ideologically balanced U.S. sample from Prolific Academic (n = 603; M_age = 38.19, SD = 13.97; 362 men, 233 women, 12 non-binary, 2 other or missing; M_conservatism = 3.21, SD = 1.70).

Procedure. Participants read about an institution of park rangers (made up for purposes of this study):

Get Outdoors: Park Rangers of America is an organization of park rangers from across the United States that works to educate the public about the various outdoor activities our national parks have to offer and to promote safety during visits to national parks. For example, they provide informational materials about the best hikes in America and on how to store food safely while camping, and they post pictures of some of our parks’ most breathtaking vistas. Their outreach activities include writing blogposts, going on podcasts, and giving talks at schools and other public events. Together, these activities have reached hundreds of thousands of community members. Their mission statement is, “Helping America experience the awe of our national parks.”

Park rangers were selected because they were considered relatively low in politicization, politically balanced, and highly trustworthy in Study 1. As our experimental manipulation, participants were randomly assigned to one of three conditions: (1) only the description of the organization above (control), or this same description plus information about the organization endorsing (2) Biden, or (3) Trump:

In September, 2020, the Get Outdoors: Park Rangers of America posted a blogpost expressing their fulsome support for Donald Trump/Joe Biden for the 2020 election, titled “Why Get Outdoors: Park Rangers of America supports Donald Trump/Joe Biden for US president.” They stated “As an organization that advocates for public well-being, we take our responsibility to this country very seriously. We cannot stand by and let our communities suffer. Donald Trump’s/Joe Biden’s commitment to protecting the health, safety, and flourishing of this nation make him the only choice in the US election.” They put together a pamphlet that was shared at their headquarters and on their website explaining why Trump/Biden was the only reasonable choice for the future of our country, and they urged visitors to distribute them widely. They also added a button to their website where visitors could easily click to donate to the Trump/Biden campaign. In their blogpost, they concluded, “In these challenging times, we feel it is our responsibility to use our voice and influence to protect our future. Vote Trump/Biden in 2020.”

Participants responded to the same politicization, trust, and deference questions as in Studies 2a and 2b. These questions were then repeated for park rangers in general to test whether a loss of trust in the suborganization Get Outdoors also impacted trust and deference toward the entire professional group of park rangers.

Participants were then told that the research team would be making a $100 donation to one of three organizations based on which organization received the most votes among participants. One of the included organizations was Get Outdoors, the other two were also made up for purposes of this study: Firefighters for Community Risk Prevention and National Dentists for Oral Health (dentists and firefighters were selected because, like park rangers, they were considered low in politicization and highly trustworthy in Study 1). This donation outcome would allow us to test whether participants were more likely to donate to the familiar organization, Get Outdoors, than the two unfamiliar organizations in the control condition, and whether this familiarity benefit declined when the organization endorsed a political candidate. Participants reported the same demographics as in earlier studies as well as which candidate they had wanted to win in the 2020 U.S. presidential election with options Trump, Biden, and Other/Neither.

Participants were debriefed that the organizations were not real and thus that we could not donate to them, however they were informed that we would make a $100 donation to a similar organization to the one that received the most votes. Because our experimental manipulation reduced support for Get Outdoors, this ended up being Firefighters for Community Risk Prevention, and we thus made the donation to The Leary Firefighters Foundation.

Study 4

Participants. To extend generalizability to a new online sample of U.S. adults, we recruited participants from CloudResearch (n = 878; M_age = 39.95, SD = 12.41; 450 men, 410 women, 12 non-binary, 6 other or missing; M_conservatism = 3.89, SD = 2.13).

Procedure. To generate a list of academic disciplines, we prompted ChatGPT, “What are the 30 most common fields for American professors in the US in recent years?”.

Participants were randomly assigned to evaluate one of two subsets of 15 academic disciplines. On the same items as in Study 1, they first evaluated the ideological slant and politicization of all disciplines and then rated the trustworthiness. Following this, as a measure of the perceived appropriateness of politicization, participants evaluated the extent to which professors in each discipline should allow their political values to influence their work on 1 (not at all) to 6 (very much) scales. These results are reported only in the Supplement. Last, participants reported how much skepticism students should have about what they are taught by professors in each discipline on 1 (no skepticism) to 6 (a great deal of skepticism) scales and completed demographics.

Study 5

Participants. We recruited sample of U.S. adults from CloudResearch (n = 806; M_age = 39.38, SD = 12.30; 424 men, 367 women, 12 non-binary, 4 other or missing; M_conservatism = 3.52, SD = 1.76).

Procedure. Participants read about an economics academic organization and were randomly assigned to either a control condition or one of three experimental conditions that included extra information: Republican politicization, Democrat politicization, depoliticization.

Control: Economics Professors of America is a professional society of academic economists in the United States. Each year, they host a very large annual conference (Economics Professors of America Annual Convention or EPAAC) where economics professors can present their latest research and network with their peers. This organization also votes on and presents awards to economists who publish highly impactful papers as well as early career economists who having promising career trajectories. They also engage with the public on social media and through their blog, aiming to inspire interest in economics and economics professions.

Politicization (conditions separated by a /): At last year's EPAAC, society leadership invited a former Democratic/Republican governor to be the keynote speaker. During the keynote speech, the speaker urged society members to support and advocate for Democratic/Republican policies in their research. Following his remarks, the Advisory Board for Economics Professors of America came on stage and discussed how Democratic/Republican values should shape the society's research agenda and outreach activities.

Depoliticization: At last year's EPAAC, society leadership invited the Executive Director of the non-profit, Political Neutrality and Tolerance in Science, to be the keynote speaker. During the keynote speech, the speaker commended society members for maintaining political neutrality in their research and teaching and provided strategies for maintaining this high standard of neutrality and tolerance. Following his remarks, the Advisory Board for Economics Professors of America came on stage and discussed three new initiatives the society had launched, supported by millions of dollars of funding and new staff positions. These would prohibit politicians and lobbyists from interfering with research outputs, help identify politically representative panels for award and review committees, and provide research funds to politically diverse research teams to encourage collaborations across political difference.

Participants responded to the same politicization and trust questions as in other studies. These questions were then repeated along with the skepticism question from Study 4 for Economics Professors in general to test whether the politicization of a suborganization impacts trust toward the broader professional group.

Participants were then told that the research team would be making a $100 donation to one of three organizations based on which organization received the most votes among participants. These included Economics Professors of America, Political Neutrality and Tolerance in Science, and The American Society for the Preservation of Historical Literature. Participants reported the same demographics as in earlier studies.

As in Study 3, participants were debriefed that the organizations were not real and thus that we could not donate to them, however they were informed that we would make a $100 donation to a similar organization to the one that received the most votes. This ended up being Political Neutrality and Tolerance in Science, and we thus made the donation to Society for Open Inquiry in Behavioral Science.

There is NO Competing Interest.

PoliticizationUnderminesTrustSupplement.docx
Supplement
PoliticizationUnderminesTrustExtendedData.docx
Extended Data

Politicization Undermines Trust in Institutions, Even Among the Ideologically Aligned Public

Status:

Version 1

Abstract

Figures

Summary Paragraph

Full Text

Conclusion

Declarations

References

Methods

Additional Declarations

Supplementary Files

Status:

Version 1