Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.
Advertisement
Communications Psychology volume 4, Article number: 61 (2026)
1785
3
Metrics details
Cost-benefit analyses of whether environmental scientists should engage in activism currently rest on a thin empirical base, despite a lively debate on the topic. There are several potential benefits of scientists’ activism, but some have argued that these benefits might be offset by the potential for activism to undermine public perception of environmental scientists as unbiased and competent. To explore these potential consequences, we asked participants to read two (ostensibly real) profiles of climate scientists that either described themselves as activists or not. Study 1 (N = 491) found that a scientist who engaged in conventional activism was seen as slightly less competent and more hypocritical than a scientist who engaged in public science communication, but there was no impact on their persuasiveness. Study 2 (N = 636, pre-registered) found that a scientist who engaged in civil disobedience, a more disruptive form of activism, was seen as less competent and more hypocritical than a non-activist scientist who only engaged in teaching and research, with predicted spill-over effects on trust in the scientist’s field. Scientist activists were also downgraded on a range of other dimensions. We conclude that engaging in activism has small but reliable costs for environmental scientists.
There is a long-standing debate within the scientific community about whether environmental scientists should engage in activism. On one side are those who believe it is environmental scientists’ responsibility to engage in activism, ranging from conventional protests to civil disobedience1. Several arguments have been proposed in favour of this notion: that traditional influence strategies are not working2, that activism has documented success as an influence strategy3,4, that science cannot (and must not) be value-free5, that scientific activism is a signal of the urgency of the environmental problem6, that scientists’ voices have especially strong influence on the public, policymakers and other activists7, and that there is an intrinsic moral or ethical imperative to use that influence8,9. According to this view, activism is an effective tool to increase the salience of the environment as an important issue, to show one’s conviction, and to sway public opinion.
On the other side of the debate are those who are concerned that activism threatens the credibility of scientists as unbiased, honest brokers of information. From this perspective, it is scientists’ reputation as impartial observers that makes scientific voices uniquely influential, and if activism threatens that reputation, the long-term cost outweighs the benefits10,11,12. Concerns about the reputational cost of activism are reinforced by research showing that the general population associates activists with negative stereotypes such as being militant, self-righteous, and ideologically biased13,14. Surveys of scientists themselves suggest generally positive views of environmental activism, although approximately 10% expressed concerns that participating in protests might create tensions with their role as a scientist (e.g., reduce their perceived objectivity)15. An international survey of climate scientists also revealed a generally harmonious perception of the relationship between science and activism, although this perception was most pronounced among the roughly half of the sample who were already members of an activist group such as Extinction Rebellion or Greenpeace16. Members of activist groups were also less uncertain about the effectiveness of activism.
In weighing up the potential credibility costs of engaging in activism, environmental scientists have a surprisingly thin evidence base to draw on. In evaluating the evidence, it is important to distinguish between advocacy and activism. Advocacy typically refers to activities in which scientists use their expertise to communicate, recommend, or support evidence-based solutions to policymakers or the public, usually within conventional institutional channels. Activism, in contrast, typically refers to direct public action—such as protests, demonstrations, or campaigns—that explicitly calls for social or political change and may be perceived as more confrontational. Although both are forms of public engagement, activism carries a stronger association with partisanship and moral urgency, which may make it more contested in terms of public perceptions of scientific credibility.
A host of research has focused on advocacy (taking a public stance in support of a cause) rather than activism as traditionally interpreted (e.g., petitioning, marching, civil disobedience). The most commonly cited paper reported that advocacy (operationalised as an interview with Associated Press) did not have consistently negative effects on scientists’ perceived credibility, and none at all on trust in the broader climate science community17. A further experiment found that non-controversial advocacy (operationalised as writing Op-Eds presenting information and suggesting education campaigns) increased climate scientists’ credibility compared to only providing scientific information, although this benefit disappeared when the advocacy included recommendations of regulation and mandating of solutions 18.
Non-experimental work on public perceptions has typically comprised surveys or focus groups. Surveys in the U.S. and Germany indicated that the general public is less supportive than is the scientific community of scientists’ political and public engagement19. Furthermore, qualitative analyses of 15 focus groups of citizens, journalists, and scientists in Denmark indicated high levels of support for the notion that scientists ought to engage in neutral rather than politically prescriptive communication11. A recent survey corroborates this view, as 4 out of 10 Americans indicated they expected scientists to focus on establishing sound scientific facts rather than contributing to policy debates20. Others have observed that a slight majority of the overall U.S. population (51%) agrees that scientists should take an active role in public policy debates, although this agreement was substantially higher among Democrats (67%) than Republicans (35%)21. Globally, support for scientists’ advocacy was observed at a comparable level (54%), with only 20% disagreeing or strongly disagreeing22. Overall, these studies suggest generally neutral-to-positive views about scientists’ advocacy, although this support does not appear to be universal.
A smaller set of studies has investigated the influence of scientists’ activism. A survey conducted immediately before and after the 2017 March for Science rallies in the U.S. revealed a polarisation effect: Liberals’ attitudes toward scientists became more positive, whereas conservatives’ attitudes became more negative23. A recent, high-powered registered report used vignettes to investigate the influence of scientists’ participation in protests against oil and gas drilling in an experimental design. Vignettes either did not mention scientists (control condition), mentioned a scientist’s endorsement of protests, or mentioned a scientist’s active participation. The study found no evidence that scientists’ participation in activism undermined their credibility or public trust in environmental science as compared to only endorsing the protests (this variable was not assessed in the control condition). However, it also found little evidence that scientists’ endorsement of or engagement in activism increased public support for related climate policies, activist movements, or donations to environmental causes24. Crucially, no non-activist scientist condition was included in the study, and it thus cannot directly evaluate the pros and cons of scientists engaging in activism. Finally, a recent experiment observed that college students indicated greater personal worry about climate change after reading a mock newspaper article by a scientist engaging in activism rather than a scientist just teaching25. Indicating that the scientist engaged in civil disobedience (i.e., was arrested for protesting) further increased reported personal worry about climate change. This study did not observe negative effects on scientists’ apparent competence or credibility. However, the presented information source was highly trustworthy (a newspaper article), and the sample of college students was left-leaning and highly supportive of environmental action.
To provide direct causal evidence for the benefits or costs of scientists’ environmental activism, the current studies involved demographically diverse Americans reading two (ostensibly real) profiles of environmental scientists. Study 1 compared a scientist who engaged in conventional environmental activism with a scientist who engaged in public science communication but with no reference to activism. Scientists were presented with carefully controlled profile pictures merged onto AI-generated background scenes in a city, either including environmental protesters or not. Study 2 compared an environmental scientist who engaged in civil disobedience, a more disruptive form of activism5,6,25, with a scientist who engaged in environmental research and teaching activities. Moreover, the civil disobedience manipulation and measures were adapted from past work25. As such, both studies controlled for whether the target engaged in environmental research, manipulating only whether the target engaged in activist behaviour as part of their engagement. Potential confounding factors such as target sex, matching of the materials to conditions, and the presentation order of materials were fully randomised (counter-balanced). After each comment, participants evaluated the scientist (Studies 1 and 2) and their research field (Study 2). We compared the mean evaluations of scientist activists versus scientist non-activists to test the causal effect of scientists’ activism on evaluations.
To summarize, we explored (Study 1) and tested (Study 2) the following hypotheses:
H1: Environmental scientists engaging in non-disruptive activism (vs. public science communication; Study 1) or disruptive activism (vs. not engaging in activism at all; Study 2) will damage their scientific reputation.
H1a: Environmental scientist-activists will be evaluated as less competent (lower expertise).
H1b: Environmental scientist-activists’ findings will be evaluated as less credible.
H2: Environmental scientists engaging in disruptive activism (vs. not engaging in activism) will be evaluated to be more hypocritical.
H3: Reading a description of environmental scientists engaging in disruptive activism (vs. not engaging in activism) will lead to lower trustworthiness evaluations of their research field (Study 2 only).
Both studies also included exploratory moderators of these effects, based on our intuition and past research. Specifically, we measured past environmental behaviour as people who already engage in pro-environmental behaviours may be more sympathetic to environmental scientists’ activism than less engaged individuals. Similarly, we included measures of trust in science, as participants with higher trust in science may be more likely to extend this trust to scientists engaging in activism as compared to less trusting individuals. Conversely, we included conspiracy mentality as a higher conspiracy mentality could intensify scepticism toward scientists when they are seen as political actors. We also included trait-level reactance and resistance to change as exploratory moderators, reflecting the idea that responses to scientist activism may depend on people’s general motivation to protect autonomy and stability. Those high in reactance may view activist scientists as overstepping or coercive, triggering defensive scepticism, whereas those high in resistance to change may oppose activism because it symbolises disruption to the social or political status quo.
We report all measures, manipulations, and participant exclusions. Materials, analyses, and data (Studies 1 and 2) are available at https://osf.io/pshak. Study 2 was pre-registered on September 19, 2025 before launching the data collection at https://osf.io/2ujrg. Unless otherwise noted, our analyses followed the pre-registered procedures. The University of Salzburg internal review board provided ethics approval for the studies (GZ10/2020). All researchers involved, including a PhD candidate, are listed as authors on this paper.
In Study 1, we used CloudResearch26 to recruit Amazon’s Mechanical Turk workers as well as Prime-Panels participants located in the U.S., with >85% approval rate and at least 100 completed HITs. CloudResearch was set up to block suspicious geocode locations, workers on its universal exclude list, and duplicate IP addresses, as well as to verify worker country and state location, and micro-batch our HIT. Workers received a base-flat-payment of $1USD. On July 22, 2024, 554 participants completed the study. Our raw dataset includes 222 incomplete entries and test-runs which are not included in this number; analyses did not show systematic attrition (see Supplementary for details). Two participants reported technical issues and 61 failed the attention check (see below) leaving N = 491 for analyses (sample demographics in Table 1). Sensitivity analyses27 indicated that this sample was sufficient to detect a small effect of d = 0.16 at 1 – β = 0.95 or a negligible effect of d = 0.13 at 1 – β = 0.80. The sample was also sufficient to explore moderators yielding a small-to-medium interaction effect of f = 0.16 at 1 – β = 0.95 or a small effect of f = 0.13 at 1 – β = 0.80.
To provide a confirmatory test of our hypotheses, Study 2 was pre-registered and recruited a larger sample, providing a high power to detect a small-to-medium effect. A power analysis using G*power 3.1 (Faul, Erdfelder, Buchner, & Lang, 2009), assuming the effect observed in Study 1 (d = 0.15) and setting 1 – β = 0.95, resulted in a minimum sample size of N = 580 for a paired-samples t-test (two-tailed). We aimed to recruit 650 subjects to account for potential drop-outs (see below). The sample was also sufficient to explore moderators yielding a small-to-medium interaction effect of f = 0.14 at 1 – β = 0.95 or a small interaction effect of f = 0.11 at 1 – β = 0.80. We used Prolific Academic to recruit a nationally representative sample (stratified by age, gender, and political affiliation) of participants located in the U.S. Participants received a base-flat-payment of the equivalent of £1.50 (about $2USD). On September 19, 2025, 651 participants completed the study. Our raw dataset includes 129 incomplete entries, which are not included in this number; analyses did not show systematic attrition (see Supplementary for details). Among the complete data sets, 15 participants failed the attention check (see below), leaving N = 636 for analyses (demographics in Table 1).
Both experiments followed 2-level (scientist activist vs. scientist non-activist) within-participants designs. Order of vignettes, order of conditions, matching of faces to conditions (Study 1), the assignment of scientists’ names to the vignettes (Study 2), and the matching of the scientists’ research field to the vignettes (Study 2) were fully counterbalanced (i.e., randomised). To control for target sex, participants either viewed two male or two female targets (i.e., target sex was randomised between participants), either by showing respective photos (Study 1) or by varying the scientist’s name (Study 2). These variations were applied automatically by the employed survey software formR 28.
Participants responded to the study advertisements on Prime Panels or MTurk (Study 1) or Prolific Academic (Study 2) by clicking on a link to a web-based survey in formR28. In both studies, participants provided informed consent and, after the main part of the study, were fully debriefed regarding origin of the comment at the end of the respective survey. All participants were given the option to withdraw their consent, which none of them did. The scales in both studies contained an attention check asking participants to select a specific response (Select strongly agree so that we are sure you are reading attentively).
The main part of Study 1 proceeded as follows: After providing informed consent, participants were asked to consecutively read two statements, ostensibly written by the target. Target person images were selected from the Chicago Face Database29, a set of standardised and rated portrait pictures. Pictures were selected to reduce the variance in the stimulus materials; that is, all were Caucasian men and women, matched for average attractiveness, trustworthiness, and prototypicality, with a rated age range of 32.88 to 41.69 years. Four sets of faces were chosen (Chicago Face Database items CFD-WF-217−085-N, CFD-WF-245−084-N, CFD-WM−018−002-N, CFD-WM-248−036-N).
We generated raw scenes of climate activist demonstrations in a city (Fig. 1) using the AI tool Microsoft Designer/Copilot (model: DALL-E 3) and replaced the faces in the generated scene with the selected Chicago Face Database portraits using Adobe Photoshop. For the scientist non-activist condition, we instructed Adobe Photoshop’s Generative Fill to adapt the generated scenes so that no climate demonstration took place (i.e., we removed all people except the target person and activism materials such as ‘save the planet’ signs). The comments all indicated that the target was an environmental scientist working on a topic related to climate change (i.e., energy use or waste reduction). In the scientist activist condition, the statement indicated that they like to attend rallies and write to politicians to influence the public. In the scientist-only condition, the statement indicated that they frequently share their findings with the media and in public lectures (Table 2).
Chicago Face Database photos were used (CFD-WF-217−085-N, CFD-WF-245−084-N, CFD-WM−018−002-N, CFD-WM-248−036-N); since no personal consent for the publication of these images could be obtained from the photographed people, faces are blurred here.
After reading about each target, participants then completed measures of nine dependent variables, details of which can be found in Table 3. The first five measures were designed to assess the target’s characteristics, with two items each measuring facets of friendliness, morality, competence, and assertiveness30, and an 8-item scale measuring the target’s perceived hypocrisy31. The person evaluation scale covered the most common dimensions of social evaluation30; we additionally included hypocrisy as it represents a common response to social motive violations and is related to message rejection31. Participants then rated the extent to which the message was perceived to be constructive (1 item) and participants’ attribution of environmental concern to the message source (2 items)32,33 as well as their agreement with the target’s message (2 items) and issue salience (3 items)34. Finally, for exploratory purposes, we incorporated three open-ended questions (What do you think about the issues addressed in the comment?, What do you think about the person who made the comment?, What do you think are the reasons why this person made that comment?).
After evaluating the second comment, participants completed additional measures to assess potential moderators. As detailed in Table 3, moderators were reactance (how much regulation attempts and rules typically instil countering behaviour in participants), pro-environmental behaviour (engagement in more environmentally friendly behaviour such as biking instead of driving), conspiracy mentality (the endorsement of the idea that secret powers control world affairs), trust in science (participants’ general trust that scientists and the scientific method yield reliable results), and resistance to change (8 items).
Finally, participants provided demographic information (age, gender, race, educational attainment, income) and indicated whether they were activists/scientists themselves (How much would you describe yourself as an activist?; If so, in which areas? How much would you describe yourself as a scientist?; If so, in which areas?). Age, participant gender, and target sex were included in the suite of moderation analyses.
The main part of Study 2 proceeded as follows: After providing informed consent, participants were first asked to respond to scales and questions designed to assess demographics and potential moderators. We moved the moderators to the beginning of the survey to preclude an effect of reading the scientist vignettes on these measurements. Specifically, participants provided demographic information (age, gender, race, educational attainment, income) and indicated whether they were activists/scientists themselves (How much would you describe yourself as an activist?; If so, in which areas?; How much would you describe yourself as a scientist?; If so, in which areas?). We then assessed the moderators observed in Study 1, pro-environmental behaviour, trust in science, and conspiracy mentality (Table 3). The measures on reactance and resistance to change were not included because they did not moderate any of the effects observed in Study 1.
Participants then consecutively read two descriptions of ostensibly real scientists (Table 4). Study 2 used text-only materials as the AI-generated background-scenes used in Study 1 may have been unconvincing for some participants. Both vignettes briefly described a scientist working on environmental issues (consumer products or energy consumption). The rest of the paragraph was designed to manipulate the scientists’ engagement in activism. The scientist activism condition was directly adapted from Friedman (2024)25 and described the targets’ involvement in civil disobedience (e.g., being arrested for blocking train lines to stop the transportation of pollutants), a disruptive form of activism. In the scientist non-activism condition, the scientist was described as engaging in climate-related teaching and research activities (e.g., organizing an environmental science lecture series).
After reading about each target, participants indicated the target’s name and research topic (as comprehension checks) and then completed measures of eight dependent variables, details of which can be found in Table 3. The first three measures were designed to assess trust in the target, with six items measuring expertise-based trust, four items measuring integrity-based trust, and four items measuring benevolence-based trust35. Expertise-based trust served as our registered dependent variable to test Hypothesis 1a. The next three items were directly adapted from Friedman (2024)25. Two items were designed to assess participants’ trust in the scientists’ findings and one to assess participants’ intention to act on them. The first measure served as our dependent variable to test Hypothesis 1b. The 8-item hypocrisy scale31 (DV to test H2) and two items on the attribution of environmental concern to the scientist32,33 used in Study 1 followed. A final measure of general trust in the scientists’ research field (three items) served as the dependent measure of Hypothesis 3. The measures on message constructiveness, agreement with the target’s message, issue salience, and the free text items were not included because they did not yield useful information in Study 1.
We used R36, including multiple packages to analyze our data (see Supplementary for details). All scale reliabilities were satisfactory, and we thus computed single indexes by averaging the items (Table 3). Since our study followed a within-subjects design, checks for variance homogeneity were not necessary. Data distribution was assumed to be normal, but this was not formally tested. We assumed normality of differences and residuals due to the large sample sizes (Study 1 N = 491; Study 2 N = 636).
Condition-independent means (across 982 data points, Study 1, and 1,272 data points, Study 2) for the dependent variables are displayed in Table 3. Evaluations of scientists were generally positive, as indicated by relatively high ratings of competence, high attributions of concern for the environment, and low hypocrisy scores (Study 1), as well as trait ratings and attributions of concern for the environment above the mid-point (Study 2) across the conditions. We observed no significant differences across Topics (Energy saving vs. Recycling [Study 1] or Energy consumption vs. Consumer products [Study 2]) and no significant Condition × Completion Time interactions (see Supplementary for details).
Comparisons of means across conditions are summarised in Table 5 (Study 1) and Table 6 (Study 2). The tables also include the results of the paired samples t-tests used to test our hypotheses, as we pre-registered for Study 2.
Activist scientists were evaluated as significantly less competent than non-activist scientists, and also as more hypocritical. These effects were small but reliable after a Bonferroni adjustment for 9 repeated significance tests. Activists’ comments were also evaluated to be less constructive than non-activists’ comments, although this effect was no longer significant after Bonferroni correction. We observed no effects of scientists’ activism on their perceived morality or friendliness. Furthermore, there were no significant effects of condition on the attribution of environmental concern, issue salience, or agreement with the comment. Thus, while we observed negative consequences of scientists’ activism on person evaluation, no significant effects on message perception emerged.
Environmental activists were evaluated as lower in expertise (H1a) and their findings as less credible (H1b) than their non-activist counterparts. Environmental scientists engaging in disruptive activism (vs. not engaging in activism at all) were also evaluated to be more hypocritical, confirming Hypothesis 2. Effects regarding Hypotheses 1 and 2 were medium in size and thus larger than in Study 1 (see Tables 5 and 6). Finally, in line with Hypothesis 3, engaging in disruptive activism (vs. not engaging in activism) led to lower trustworthiness evaluations of the described scientist’s research field.
As pre-registered, we further explored the effect of scientist activism on several additional dependent measures (Table 6). Engaging in activism reduced perceptions of scientists’ integrity-based trust, benevolence-based trust, and their apparent environmental concern, although the latter effect was small. Finally, scientists who engaged (vs. did not engage) in activism led participants to report a lower intention to act pro-environmentally.
To explore potential moderators of the observed effects of scientists’ activism on their perceived competence and hypocrisy, we included condition, the respective moderator, and the respective dependent variable (competence and hypocrisy) in a generalised linear model, including the participant as a random intercept. We followed up on significant Condition × Moderator interactions with Johnson-Neyman analyses to identify regions of significance (Figs. 2 & 3). Johnson-Neyman analyses consider the complete range of moderator values and indicate at which of these values a significant condition effect was observed. Accordingly, Johnson-Neyman analyses yield more unbiased information than do classic approaches (e.g., simple slopes) that focus only on a few arbitrary values of the moderator37. Here, we only report significant moderations (see Supplementary for a full report of the conducted analyses).
The message source effect (scientist activist vs. scientist non-activist) was not significant in red-shaded areas and significant at p <0.05 in the green-shaded areas (N=491 participants for pro-environmental behaviour, trust in science, and conspiracy mentality; N = 486 participants for age [5 participants did not indicate their age]). The y-axis indicates the observed effect size of the activist vs. non-activist scientist manipulation. Values above 0 indicate that evaluation values on the dependent variable were greater for activists than non-activists; values below 0 indicate that values on the dependent variable were lower for activists than non-activists. Thick horizontal line indicates the range of observed values of the moderator; dashed green lines indicate transition points (i.e., moderator values where the condition effect transitions between significant and non-significant).
The source activism effect (scientist activist vs. scientist non-activist) was not significant in red-shaded areas and significant at p < 0.05 in the green-shaded areas (N = 636 participants for pro-environmental behaviour, trust in science, and political orientation, N = 633 participants for age [3 participants did not indicate their age]). The y-axis indicates the observed effect size of the activist vs. non-activist scientist manipulation. Values above 0 indicate that evaluation values on the dependent variable were greater for activists than non-activists; values below 0 indicate that values on the dependent variable were lower for activists than non-activists. A thick horizontal line indicates the range of observed values of the moderator; dashed green lines indicate transition points (i.e., moderator values where the condition effect transitions between significant and non-significant). See Supplementary for all Johnson-Neyman graphs.
Condition effects on competence evaluations were consistently moderated by reported environmental behaviour, B = 0.15, SE = 0.04, t(489) = 3.69, p <0.001, trust in science, B = 0.15, SE = 0.04, t(489) = 3.71, p <0.001, and conspiracy mentality, B = 0.10, SE = 0.04, t(489) = 2.71, p = 0.007. Johnson-Neyman analyses indicated that the condition effect on competence was not significant for participants reporting high levels of pro-environmental behaviour (above 5.51), a very high trust in science (above 5.73), or a high conspiracy mentality (above 5.27, Fig. 2). We did not observe parallel moderation effects on hypocrisy.
Regarding demographics, condition effects on hypocrisy (but not on competence) were moderated by age and condition effects on competence (but not on hypocrisy) were moderated by self-description as an activist, such that younger participants (<39 years) and those who identified as activists exhibited no significant condition effect (Fig. 2).
As pre-registered, we explored if the expected effects (i.e., H1-H3) were moderated by participants’ own pro-environmental behaviour, conspiracy mentality, trust in science, or demographics (political orientation, age, gender, participants’ identification as activists and scientists). Condition effects on all pre-registered dependent measures were moderated by reported environmental behaviour: expertise-based trust, B = 0.12, SE = 0.04, t(634) = 3.03, p = .003, hypocrisy, B = −0.18, SE = 0.04, t(634) = -4.43, p <0.001, credibility, B = 0.27, SE = 0.07, t(634) = 3.72, p <0.001, and trust in researcher’s field, B = 0.15, SE = 0.04, t(634) = 4.15, p <0.001. Trust in science also emerged as a consistent moderator of the condition effects on expertise-based trust, B = 0.17, SE = 0.04, t(634) = 4.07, p <0.001, hypocrisy, B = −0.20, SE = 0.04, t(634) = -4.78, p <0.001, credibility, B = 0.34, SE = 0.07, t(634) = 4.55, p <0.001, and trust in researcher’s field, B = 0.16, SE = 0.04, t(634) = 4.11, p <0.001. Participants engaging in more pro-environmental behaviour and with higher trust in science showed smaller negative effects of scientists’ activism. Johnson-Neyman analyses indicated that the condition effect on trust in the researchers’ field was not significant for participants reporting very high levels of pro-environmental behaviour (above 5.96) or a very high trust in science (above 6.64, Fig. 3). No other transition points were observed.
Regarding demographics, condition effects on all pre-registered dependent measures were moderated by participants’ political orientation. Johnson-Neyman analyses indicated that more left-leaning participants showed smaller negative effects of scientists’ activism. Significant transition points emerged on hypocrisy (<1.18), credibility (<0.43) and the trust in the researchers’ field (<1.82; Fig. 3). Condition effects on credibility (but not the other pre-registered dependent measures, expertise-based trust, hypocrisy, trust in researcher’s field) were moderated by participants’ age, B = −0.02, SE = 0.01, t(631) = -2.86, p = .004. Older participants showed a larger condition effect on credibility, but no transition points were observed (Fig. 3).
Finally, condition effects on all pre-registered dependent measures were moderated by participants’ self-description as an activist (yes vs. no). A stronger Source Activism effect emerged among non-activist participants than among activist participants on expertise-based trust, F(1, 617) = 13.62, p <0.001, η²p = 0.02, 90% CI [0.01, 0.04], hypocrisy, F(1, 617) = 20.10, p <0.001, η²p = 0.03, 90% CI [0.01, 0.06], credibility, F(1, 617) = 26.08, p <0.001, η²p = 0.04, 90% CI [0.02, 0.07], and trust in researcher’s field, F(1, 617) = 20.26, p <0.001, η²p = 0.03, 90% CI [0.01, 0.06] (Table 7).
There is an ongoing debate about the reputational costs and persuasive benefits of engaging in activism for environmental scientists, but experimental research on this topic is limited. To help fill this empirical gap, Study 1 compared community perceptions of environmental scientists who engaged in non-disruptive activism with those who engaged in public science communication. In Study 2, we compared perceptions of environmental scientists who engaged in disruptive activism with a non-activist control.
Overall, scientists were evaluated quite positively, as indicated by high average ratings of competence, morality, and trust as well as low hypocrisy (Table 3). Nevertheless, both studies detected negative effects of activism on perceptions of the target scientists’ competence and hypocrisy. Study 2 revealed that these negative effects extended to participants’ reduced trust in the field of the researcher and exploratory evidence of reduced intent to act on the researchers’ findings, as well as other evaluations of trust and environmental concern. The observed effects in Study 2 were not only more consistent but also substantially larger than in Study 1, potentially due to the more extreme forms of activism described in the manipulation. In sum, engaging in activism had reliable negative effects on the perception of scientists and (in the case of disruptive activism) some effects on the effectiveness of their message and perception of their field. In both studies, these effects were particularly pronounced among those members of the community who need to be reached to have an impact: those with relatively low trust in science and low levels of pro-environmental behaviour.
Our findings stand in contrast to past experiments that did not observe negative consequences of scientists’ activism on the credibility of their work or participants’ intent to engage in climate action24,25. However, neither of these past experiments included no-activism control conditions but rather contrasted different forms of advocacy. For instance, Friedman25 observed no negative effects of disruptive activism over other types of civic engagement (i.e., non-disruptive activism or advocacy) on scientists’ credibility and participants’ intent to act. In contrast, we sought to support scientists who wish to make an informed decision whether to get actively involved in activism or not and thus included a “pure” no-activism control condition.
In fact, our Study 2 used Friedman’s disruptive activism manipulation and some of their dependent measures. Interestingly, our participants responded to the disruptive activism description (credibility M = 6.71, SD = 2.94, action intent M = 6.31, SD = 3.06) in a manner similar to Friedman’s participants (credibility M = 6.11, SD = 1.66, action intent M = 5.88, SD = 2.23). However, our control condition without activism, which was not included in Friedman’s study, yielded significantly more positive ratings (credibility M = 8.15, SD = 2.06, action intent M = 6.99, SD = 2.72). Apparently, activism reduced positive perceptions of scientists rather than triggering their rejection.
A few caveats should be noted. First, across both experimental conditions, evaluations of the scientists were generally positive, indicating that we are observing shifts within an overall favourable impression rather than outright rejection or hostility toward activist scientists. The effects, therefore, reflect relative differences in warmth and credibility rather than categorical disapproval. Second, public approval is not the sole metric of effective activism. Those engaging in activism may, in fact, not aim for approval but seek disruption to signal power38, build momentum39, or attract media attention40. The assumption is that the resulting tension is required for societal change—even if it comes at the cost of alienating some members of the public. According to this view, disruptive and divisive actions may nonetheless be consequential if they attract media attention, shift public agendas, or pressure decision-makers to act (see, e.g., Ostarek, et al.41).
From this perspective, activism can have instrumental or environmental benefits even when it provokes ambivalence or resistance in the community, highlighting that reputational costs and social impact do not always move in tandem. Alternatively, activist events may serve an important function for the climate community, such as helping create a joint identity or attracting new members to join the movement. Whether increased media attention and a large, strongly-identified climate movement increase readiness for change in the broader population are important questions, but this was outside the auspices of the current study. We hope that the current research will help scientists to make an informed decision on whether to engage in activism.
Both reported studies were conducted in the US, a context highly polarized on science and environmental issues. For instance, Cologna et al.22 observed that while average endorsement of scientists’ public engagement was moderate in the US, these ratings were strongly correlated with participants’ reported trust in science. This relationship between trust in science and the endorsement of scientists’ engagement was weaker or even reversed in other countries. It is thus possible that the observed effects would be attenuated in contexts where science and the environment are less polarizing topics. Accordingly, future research should assess how engaging in activism influences the perception of scientists across cultures.
To resolve the activist dilemma described here—balancing the need to create change with the need to maintain a reputation as non-hypocritical and competent—scientists who wish to engage in activism have several options. One option is to separate science and activism activities by highlighting one’s role as a private citizen, not an expert, during activism. This approach is generally in line with the pledge to only voice recommendations in line with one’s own scientific expertise and in a non-defensive, neutral manner42, but see5. Scientists could also choose to engage in advocacy, working behind the scenes for policy change with political actors they have unique access to, although advocacy may come with its own challenges17,18,19,20. To overcome these challenges, related research on advocacy shows that taking the position of an honest broker of information rather than emphasizing the epistemic implications of their knowledge for political decisions (i.e., an epistocrat) can go a long way43. Similar approaches may help attain the key goal to enhance the benefits of scientists’ engagement in activism (e.g., provide skills, access to public stakeholders) while securing their scientific reputation.
In sum, our results show that, all else being equal, engaging in activism had moderate but reliable costs in terms of reduced competence and increased hypocrisy, as well as some small costs on downstream trust and perceptions of science. We hope our research will help scientists make informed decisions on whether the benefits of activist engagement outweigh the potential costs.
Anonymized raw data are available at https://osf.io/pshak/.
Code for analyses is available at https://osf.io/pshak.
Castiglione, A., Brick, C., Holden, S., Miles-Urdan, E. & Aron, A. R. Discovering the psychological building blocks underlying climate action—a longitudinal study of real-world activism. R. Soc. Open Sci. 9, 210006 (2022).
Article PubMed PubMed Central Google Scholar
Racimo, F. et al. The biospheric emergency calls for scientists to change tactics. eLife 11, e83292 (2022).
Article PubMed PubMed Central Google Scholar
Bennett, H., Macmillan, A., Jones, R., Blaiklock, A. & McMillan, J. Should health professionals participate in civil disobedience in response to the climate change health emergency? Lancet 395, 304–308 (2020).
Article PubMed Google Scholar
Finnerty, S., Piazza, J. & Levine, M. Between two worlds: The scientist’s dilemma in climate activism. npj Clim. Action 3, 77 (2024).
Article Google Scholar
van Eck, C. W., Messling, L. & Hayhoe, K. Challenging the neutrality myth in climate science and activism. npj Clim. Action 3, 81 (2024).
Article Google Scholar
Gardner, C. J. & Wordley, C. F. R. Scientists must act on our own warnings to humanity. Nat. Ecol. Evol. 3, 1271–1272 (2019).
Article PubMed Google Scholar
Capstick, S. et al. Civil disobedience by scientists helps press for urgent climate action. Nat. Clim. Change 12, 773–774 (2022).
Article Google Scholar
Garcia-Gibson, F. The ethics of climate activism. WIREs Clim. Change 14, e831 (2023).
Article Google Scholar
Anguelovski, I. et al. The activism responsibility of climate scientists and the value of science-based activism. npj Clim. Action 4, 40 (2025).
Article Google Scholar
Donner, S. D. Finding your place on the science – advocacy continuum: An editorial essay. Clim. Change 124, 1–8 (2014).
Article Google Scholar
Nicolaisen, P. B. A state of emergency or business as usual in climate science communication? A three-dimensional perspective on the role perceptions of climate scientists, climate journalists, and citizens. Sci. Commun. 44, 667–692 (2022).
Article Google Scholar
Büntgen, U. The importance of distinguishing climate science from climate activism. npj Clim. Action 3, 36 (2024).
Article Google Scholar
Farinha, C. & Rosa, M. Just chill! An experimental approach to stereotypical attributions regarding young activists. Soc. Sci. 11, 427 (2022).
Article Google Scholar
Bashir, N. Y., Lockwood, P., Chasteen, A. L., Nadolny, D. & Noyes, I. The ironic impact of activists: Negative stereotypes reduce social change influence. Eur. J. Soc. Psychol. 43, 614–626 (2013).
Article Google Scholar
Dablander F. et al. Climate change engagement of scientists. Nat. Clim. Change 14, 1033–1039 (2024).
Finnerty, S., Piazza, J. & Levine, M. Scientists’ identities shape engagement with environmental activism. Commun. Earth Environ. 5, 240 (2024).
Article Google Scholar
Kotcher, J. E., Myers, T. A., Vraga, E. K., Stenhouse, N. & Maibach, E. W. Does engagement in advocacy hurt the credibility of scientists? Results from a randomized national survey experiment. Environ. Commun. 11, 415–429 (2017).
Article Google Scholar
Beall, L., Myers, T. A., Kotcher, J. E., Vraga, E. K. & Maibach, E. W. Controversy matters: Impacts of topic and solution controversy on the perceived credibility of a scientist who advocates. PLOS ONE 12, e0187511 (2017).
Article PubMed PubMed Central Google Scholar
Cologna, V., Knutti, R., Oreskes, N. & Siegrist, M. Majority of German citizens, US citizens and climate scientists support policy advocacy by climate researchers and expect greater political engagement. Environ. Res. Lett. 16, 024011 (2021).
Article Google Scholar
Pew Research Center. Trust and mistrust in Americans’ views of scientific experts. https://www.pewresearch.org/science/2019/08/02/trust-and-mistrust-in-americans-views-of-scientific-experts/ (2019).
Pew Research Center. Public trust in scientists and views on their role in policymaking. https://pewrsr.ch/3URbOGZ (2024).
Cologna, V. et al. Trust in scientists and their role in society across 68 countries. Nat. Hum. Behav. 9, 713–730 (2025).
Article PubMed PubMed Central Google Scholar
Motta, M. The polarizing effect of the march for science on attitudes toward scientists. PS: Political Sci. Polit. 51, 782–788 (2018).
Google Scholar
Dablander, F., Sachisthal, M. S. M. & Aron, A. R. Out of the labs and into the streets: Effects of climate protests by environmental scientists. R. Soc. Open Sci. 12, 241001 (2025).
Article PubMed PubMed Central Google Scholar
Friedman, R. S. Civil disobedience by environmental scientists: An experimental study of its influence on the impact and credibility of climate change research*. Environ. Commun. 18, 451–464 (2024).
Article Google Scholar
Litman, L., Robinson, J. & Abberbock, T. TurkPrime.com: A versatile crowdsourcing data acquisition platform for the behavioral sciences. Behav. Res. Methods 49, 433–442 (2017).
Article PubMed PubMed Central Google Scholar
Faul, F., Erdfelder, E., Buchner, A. & Lang, A.-G. Statistical power analyses using G*Power 3.1: Tests for correlation and regression analyses. Behav. Res. Methods 41, 1149–1160 (2009).
Article PubMed Google Scholar
Arslan, R. C., Walther, M. P. & Tata, C. S. formr: A study framework allowing for automated feedback generation and complex longitudinal experience-sampling studies using R. Behav. Res. Methods 52, 376–387 (2020).
Article PubMed PubMed Central Google Scholar
Ma, D. S., Correll, J. & Wittenbrink, B. The Chicago face database: A free stimulus set of faces and norming data. Behav. Res. Methods 47, 1122–1135 (2015).
Article PubMed Google Scholar
Koch, A. et al. Validating a brief measure of four facets of social evaluation. Behav. Res. Methods 56, 8521–8539 (2024).
Article PubMed Google Scholar
Pillow, D. R., Hale, W. J., Kohler, J., Mills, S. & Soler, J. The motivated appeal to hypocrisy: the relation of motivational threats to message rejection. Front. Psychol. 14, 1253132 (2023).
Hornsey, M. J. & Imani, A. Criticizing groups from the inside and the outside: An identity perspective on the intergroup sensitivity effect. Personal. Soc. Psychol. Bull. 30, 365–383 (2004).
Article Google Scholar
Thürmer, J. L. & McCrea, S. M. Disentangling the Intergroup Sensitivity Effect: Defending the ingroup or enforcing general norms? Eur. J. Soc. Psychol. 51, 1061–1072 (2021).
Article Google Scholar
Halfmann, E., Thürmer, J. L. & McCrea, S. M. Group boundaries impede organizational change: Intergroup sensitivity in the organic food industry. (in prep.).
Hendriks, F., Kienhues, D. & Bromme, R. Measuring laypeople’s trust in experts in a digital age: The muenster epistemic trustworthiness inventory (METI). PLOS ONE 10, e0139309 (2015).
Article PubMed PubMed Central Google Scholar
R-Core-Team. R: A language and environment for statistical computing. R Foundation for Statistical Computing (2020).
Finsaas, M. C. & Goldstein, B. L. Do simple slopes follow-up tests lead us astray? Advancements in the visualization and reporting of interactions. Psychol. Methods 26, 38–60 (2021).
Article PubMed Google Scholar
Tufekci, Z. Twitter and tear gas: The power and fragility of networked protest. Yale University Press (2017).
Han, H. & Barnett-Loro, C. To support a stronger climate movement, focus research on building collective power. Front. Commun. 3, (2018).
Tufekci, Z. Not this one”: Social movements, the attention economy, and microcelebrity networked activism. Am. Behav. Sci. 57, 848–870 (2013).
Article Google Scholar
Ostarek, M., Simpson, B., Rogers, C. & Ozden, J. Radical climate protests linked to increases in public support for moderate organizations. Nat. Sustain. 7, 1626–1632 (2024).
Article Google Scholar
Kotcher, J., Luong, K., Charles, J., Gould, R. & Maibach, E. Calling attention to opponents of climate action in climate and health messaging. Lancet Planet. Health 7, e938–e946 (2023).
Article PubMed Google Scholar
Post, S. & Bienzeisler, N. The honest broker versus the epistocrat: Attenuating distrust in science by disentangling science from politics. Political Commun. 41, 763–785 (2024).
Article Google Scholar
McCrea, S. M., Erion, C. J. G. & Thürmer, J. L. Why punish critical outgroup commenters? Social identity, general norms, and retribution. Br. J. Soc. Psychol. 61, 711–728 (2022).
Article PubMed Google Scholar
Sturgis, P., Brunton-Smith, I. & Jackson, J. Trust in science, social consensus and vaccine confidence. Nat. Hum. Behav. 5, 1528–1534 (2021).
Article PubMed Google Scholar
White, K. R. G., Kinney, D., Danek, R. H., Smith, B. & Harben, C. The Resistance to Change-Beliefs Scale: Validation of a new measure of conservative ideology. Personal. Soc. Psychol. Bull. 46, 20–35 (2020).
Article Google Scholar
Hong, S.-M. & Faedda, S. Refinement of the Hong Psychological Reactance Scale. Educ. Psychol. Meas. 56, 173–182 (1996).
Article Google Scholar
Brick, C., Sherman, D. K. & Kim, H. S. Green to be seen” and “brown to keep down”: Visibility moderates the effect of identity on pro-environmental behavior. J. Environ. Psychol. 51, 226–238 (2017).
Article Google Scholar
Stojanov, A. & Halberstadt, J. The Conspiracy Mentality Scale: Distinguishing between irrational and rational suspicion. Hogrefe Publishing (2019).
Download references
This research was funded in part by the Austrian Science Fund (FWF) [https://doi.org/10.55776/P37261] awarded to J.L.T and ARC Laureate fellowship awarded to M.J.H. (FL230100022). The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript. We thank Julia Prohaska for her assistance in preparing the manuscript.
Department of Psychology, University of Salzburg, Salzburg, Austria
J. Lukas Thürmer & Jeremias Braid
Economic Psychology, Seeburg Castle University, Seekirchen am Wallersee, Austria
J. Lukas Thürmer
Salzburg Center for European Union Studies, University of Salzburg, Salzburg, Austria
J. Lukas Thürmer
Department of Psychology, University of Wyoming, Laramie, WY, USA
Sean M. McCrea
Business School, The University of Queensland, St Lucia, QLD, Australia
Matthew J. Hornsey
PubMed Google Scholar
PubMed Google Scholar
PubMed Google Scholar
PubMed Google Scholar
Conceptualization, J.L.T., S.M.M., & M.J.H.; methodology, J.B. & J.L.T.; formal analysis, J.B.; resources, M.J.H & J.L.T.; data curation, J.B.; writing—original draft preparation, J.L.T. & J. B.; writing—review and editing, M.J.H & S.M.M.; visualization, J.B.; supervision, J.L.T.; project administration, J.B.; funding acquisition, M.J.H & J.L.T. All authors have read and agreed to the published version of the manuscript.
Correspondence to J. Lukas Thürmer.
The authors declare that they have no competing interests.
Communications Psychology thanks Viktoria Cologna and the other anonymous reviewer(s) for their contribution to the peer review of this work. Primary Handling Editors: John Aruta and Troby Ka-Yan Lui. A peer review file is available.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Reprints and permissions
Thürmer, J.L., Braid, J., McCrea, S.M. et al. Navigating the credibility risks of environmental scientists’ activism. Commun Psychol 4, 61 (2026). https://doi.org/10.1038/s44271-026-00409-8
Download citation
Received:
Accepted:
Published:
Version of record:
DOI: https://doi.org/10.1038/s44271-026-00409-8
Anyone you share the following link with will be able to read this content:
Sorry, a shareable link is not currently available for this article.
Provided by the Springer Nature SharedIt content-sharing initiative
Collection
Advertisement
Communications Psychology (Commun Psychol)
ISSN 2731-9121 (online)
© 2026 Springer Nature Limited
Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.