The effects of social identity threat and social identity affirmation on laypersons’ perception of scientists

Abstract

Public debates about socio-scientific issues (e.g. climate change or violent video games) are often accompanied by attacks on the reputation of the involved scientists. Drawing on the social identity approach, we report a minimal group experiment investigating the conditions under which scientists are perceived as non-prototypical, non-reputable, and incompetent. Results show that in-group affirming and threatening scientific findings (compared to a control condition) both alter laypersons’ evaluations of the study: in-group affirming findings lead to more positive and in-group threatening findings to more negative evaluations. However, only in-group threatening findings alter laypersons’ perceptions of the scientists who published the study: scientists were perceived as less prototypical, less reputable, and less competent when their research results imply a threat to participants’ social identity compared to a non-threat condition. Our findings add to the literature on science reception research and have implications for understanding the public engagement with science.

Keywords

public engagement with science science attitudes and perceptions science communication social identity subtyping threat

Over the last 20 years, public engagement with science activities, such as public meetings, consensus conferences, or science cafés, have become the state-of-the-art approach of science communication (Gregory and Lock, 2008). By shifting away from the public deficit model (Ziman, 1991), according to which laypersons simply have to be “educated” about science, science communicators tried to engage laypersons, scientists, and policymakers in dialogic settings in order to create a sphere of mutual understanding (Stilgoe et al., 2014). A premise of this open, involving, and participatory approach in order to be successful is that laypersons are willing and capable to engage in a constructive and open-minded dialogue about socio-scientific problems (Carpini et al., 2004). However, psychological research casts doubt on whether this premise is easily satisfied. This research demonstrated that laypersons are prone to various biases aiming at fostering pre-existing beliefs and one’s positive sense of the self (Kunda, 1990) that might ultimately impede public engagement activities. For instance, laypersons readily dismiss scientific evidence that is belief-inconsistent (Greitemeyer, 2014) or identity-threatening (Nauroth et al., 2015) independent of its objective quality.

These psychological findings challenge the assumption that laypersons are generally open-minded when they engage in a science–public dialogue and question public engagement as the silver bullet to overcome the science–public divide. At the same time, advocates of engagement activities may argue that factual disagreements between laypersons and scientists are at the very heart of these activities and are even necessary for establishing a mutual understanding between scientists and laypersons (Lezaun and Soneryd, 2007). However, engagement activities would be severely complicated if psychological biases do not only lead to disagreement about the matter of fact but entailed distrust and doubts about the professional stance of the participating scientists. Such motivated skepticism about the scientists themselves may actually be harmful for a constructive discussion environment (Schrodt et al., 2009). For instance, when laypersons question the competence and reputation of scientists participating in engagement activities only because these scientists advocate undesirable research findings, a constructive dialogue may be significantly impaired (Binder et al., 2011). This problem would be enhanced if laypersons perceived participating scientists as excluded within the scientific community representing a minority opinion rather than as representative members of the scientific community. Why should laypersons sincerely discuss scientific issues with scientists of whom they think that they are incompetent, disreputable, and not advocating the scientific state-of-the-art?

One may argue that these concerns are far-fetched and there are no theoretical reasons to assume such comprehensive biases in science reception. However, when conceptualizing the layperson–scientist relationship as an intergroup situation, social identity theory (Tajfel and Turner, 1986) and the burgeoning literature on social influences in laypersons’ science reception (see Maier et al., 2014) allow the deduction of specific predictions about how laypersons perceive scientists advocating in-group favorable or in-group unfavorable scientific findings. Specifically, the present research investigates whether the extent to which scientific findings affirm or jeopardize laypersons’ social identity can alter laypersons’ perceptions about scientists who publish or advocate the findings. In the following, we derive the theoretical rational for our argument that social identity threat (Tajfel and Turner, 1986) and social identity affirmation (Ellemers et al., 2002) bias not only people’s evaluation of research findings but also their perception of the researchers who publish and advocate these findings.

1. Social identity threat and affirmation

To date, research on laypersons’ science perception mainly rests on the assumption that science reception biases are rooted in individual beliefs, attitudes, or knowledge (e.g. Sinatra et al., 2014). More recent research, however, suggests that a biased science reception may also be related to social identity concerns (Postmes, 2015). A particular scientific finding, such as the finding that a vegetarian diet has positive consequences for one’s health, affirms not only a vegetarian’s personal beliefs and values but also his or her social identity—the self-worth one derives from being a member of a social category (Tajfel and Turner, 1986), in this case, being a vegetarian. By contrast, a study demonstrating negative health effects of a vegetarian diet should implicitly threaten a vegetarian’s social identity. One way to cope with that threat would be to devalue the threatening study (De Hoog, 2013).

A number of findings support these assumptions. Morton et al. (2006) showed that participants evaluated a study as more “scientific” and its research topic to be more eligible (i.e. deserving more government funding) when the findings affirmed their gender identity. They concluded that scientific findings are more likely to be perceived as credible and plausible to the extent that they provide people with a positive sense of their social identity. Complementing these results, Nauroth et al. (2014) showed that research findings implicitly threatening one’s social identity led participants to devalue these findings. Strongly identified video game players negatively evaluated a scientific study when it found that playing these games increases players’ aggressive tendencies (compared to a study which showed no negative effects of playing violent video games). These studies were the first to shed light on the psychological underpinnings of the denial and acceptance of group-relevant research findings, but they fall short of investigating whether these group biases are so pervasive to also alter more basic psychological processes—that is whether these biases also affect laypersons’ impressions about the scientists personally.

It is plausible to assume that social identity–affirming and threatening findings not only influence laypersons’ evaluation of the research but also their perceptions of the respective researchers. Notably, two opposing hypotheses can be derived from social identity theory (Tajfel and Turner, 1979) and from the literature on group perception regarding how social identity threat and affirmation may alter people’s perception of researchers. One hypothesis, which we refer to as the “homogeneity hypothesis,” builds upon group perception research (Ostrom and Sedikides, 1992). This research suggests that threatening results lead to a more homogeneous perception of the entire scientific community, whereas affirming results do not alter people’s homogeneity perceptions. The other hypothesis, which we refer to as the “subtyping hypothesis,” argues that being confronted with threatening research results leads people to cognitively exclude the respective authors from the scientific community, whereas affirming results lead people to perceive the authors as more prototypical. Both hypotheses also allow to derive predictions regarding laypersons’ perception of the competence and the reputability of the researchers who published threatening versus affirming research and, therefore, for laypersons’ engagement with science.

2. The homogeneity hypothesis

Group perception research indicates that people often generalize their impression about one or a few out-group members to the entire out-group (e.g. Crawford et al., 2002). Negative behaviors performed by members of non-familiar out-groups are particularly likely to be generalized to the entire out-group (Hamilton and Gifford, 1976). Subsequent research supported this contention (e.g. Hamilton et al., 1985; Risen et al., 2007) and demonstrated that negative behaviors are also more readily generalized than positive behaviors (Barlow et al., 2012; Stark et al., 2013). For the average layperson, the scientific community represents a non-familiar out-group with which they seldom interact. Thus, when laypersons are confronted with researchers publishing or advocating threatening research findings (a “negative” behavior; Nauroth et al., 2014), laypersons might generalize their impressions about these researchers to the entire scientific community. Conversely, these researchers should be perceived as rather typical members of the scientific community.

People perceive threatening out-groups to be more homogeneous than non-threatening out-groups (e.g. Corneille et al., 2001; Dépret and Fiske, 1999; Rothgerber, 1997; Wilder, 1978). For instance, Rothgerber (1997) showed that students perceive more intragroup similarities among students from a rivaling university when students were told that out-group students were negatively biased in judging essays of in-group students. Applied to the case where laypersons see themselves confronted with a threatening scientific study, they may perceive researchers of the respective scientific community as rather homogeneous. Taken together, the homogeneity hypothesis for social identity threat states that being confronted with threatening (vs non-threatening) research results leads to higher homogeneity perceptions of the respective scientific community and to higher typicality perception of the researchers who published the research results.

In contrast to threatening research, social identity–affirming research should not affect laypersons’ out-group perceptions (or their typicality perceptions). Since increased homogeneity perceptions are assumed to foster in-group members’ defenses against negative social identity implications, group members should not alter their out-group perception when being confronted with non-threatening groups (Dépret and Fiske, 1999; Wilder, 1978). This is in line with results from Rothgerber’s (1997) first study. In this experiment, he incorporated a benevolent out-group condition (besides the threatening out-group condition; see above) and showed that a benevolent out-group (i.e. out-group students judging in-group students’ essays very good) did not alter homogeneity perceptions compared to a control group. Thus, a scientific result affirming one’s social identity should have no effect on homogeneity (or typicality) perceptions of the scientific community.

3. The subtyping hypothesis

The homogeneity hypothesis rests on the assumption that researchers publishing threatening scientific findings are perceived as representative prototypes of the scientific community. However, this might not necessarily be the case. Since laypersons are motivated to re-establish a positive social identity after being confronted with threatening research findings (Nauroth et al., 2015), one way to maintain a positive social identity in the face of threatening scientific evidence could be to dismiss a single threatening finding—and the researcher(s) who published this finding—as dissenting and not representative. This prediction resonates with the observation made by Gilbert and Mulkay (1984) that even scientists themselves despise other scientists who advocate a scientific opinion differing from their own. From a psychological perspective, this would be in line with research on subtyping (Weber and Crocker, 1983). Subtyping describes the phenomenon that members of an out-group who disconfirm the group stereotype are sometimes cognitively excluded from the respective social category (Riek et al., 2013). Applied to the case where laypersons read about a threatening scientific finding, researchers who published this finding may be excluded from the social category of scientists. The key variable indicating subtyping is perceived prototypicality (see Richards and Hewstone, 2001): group members being perceived as non-prototypical are the ones also being cognitively excluded from the respective social category (i.e. subtyped). Thus, researchers publishing threatening findings should be perceived as less prototypical members of the scientific community than researchers publishing non-threatening findings. Taken together, the subtyping hypothesis for social identity threat states that being confronted with threatening (vs non-threatening) research results leads to reduced typicality perceptions of the researchers for the respective scientific community.

The logic underlying the subtyping hypothesis also allows for predictions concerning affirmative research results. When laypersons downplay a threatening research finding in order to protect their social identity, they may also exaggerate the importance of an affirming research finding in order to boost their social identity. More precisely, the identity-enhancing effect of an affirming research result should be strongest when people also believe in its validity and perceive the researchers as prototypical and reputable. Thus, when being confronted with an identity-affirming research result, laypersons should be inclined to perceive the researchers as prototypical for the scientific community. This prediction would also be in line with the results of Morton et al. (2006), who showed that identity-affirming scientific results lead to more positive research evaluations.

4. The present research

Summing up, the homogeneity hypothesis and the subtyping hypothesis predict distinctly different patterns regarding the effect of identity-threatening and identity-affirming research results on laypersons’ perceptions of researchers. According to the homogeneity hypothesis, people should perceive researchers who publish identity-threatening research results as more typical for the respective scientific community and perceive the community as more homogeneous compared to a non-threat control condition. However, identity-affirming research results should not alter typicality or homogeneity perceptions compared to a control condition. By contrast, the subtyping hypothesis predicts that people perceive researchers who publish identity-threatening research results as less typical for the respective scientific community compared to a control condition, whereas homogeneity perceptions of the (remaining) scientific community should remain unaffected. However, identity-affirming research results should lead to higher typicality perceptions compared to a control condition in order to enhance their effect on one’s social identity.

Both hypotheses have immediate implications for how competent and reputable scientists might be perceived. Whereas the homogeneity hypothesis implies that threatening (and affirming) research results should not alter laypersons’ assessment of competence and reputability of an individual researcher, the subtyping hypothesis implies that laypersons perceive scientists publishing threatening (vs affirming) research results as less (vs more) competent and reputable (in line with the predicted decreased (vs increased) typicality perception). Thus, we also measured laypersons’ competence and reputation assessments in order to shed light on possible downstream consequences of researcher typicality perceptions. Potential effects on the competence and reputation impressions of laypersons are particularly important for public engagement activities: if the subtyping hypothesis was true, researchers participating in engagement activities would face the risk of being perceived as atypical, incompetent, and disreputable when advocating threatening research findings and as a stereotypical, competent, and reputable when advocating affirming research. However, if the homogeneity hypothesis was true, the individual researcher would be perceived as representing the scientific community regardless of the scientific evidence he or she advocates.

Additionally, we set out to replicate the effects of affirming and threatening research studies on the evaluation of these studies found by Morton et al. (2006) and Nauroth et al. (2014, 2015). These authors investigated the biasing effects of social identity threat and affirmation only in one context (i.e. the effect of social identity threat was only investigated in the context of research on violent video games, and the effect of social identity affirmation was only investigated in the context of research on gender differences). In order to test whether their findings reflect a general principle or are context-specific, we tested their predictions in a minimal group paradigm (Tajfel, 1970). A minimal group paradigm categorizes individuals on the basis of arbitrary distinctions (in our case a faked psychological test determining one’s “association style”), thereby being able to experimentally manipulate individuals’ group affiliation. Arbitrary categorizations are preferable over “naturalistic” categorizations because socially shared stereotypes and intergroup dynamics in natural group memberships could potentially confound social identity effects. Furthermore, the minimal group paradigm is well-established in social psychology in general (e.g. Diehl, 1990) and in social perception research in particular (e.g. Rubin et al., 2001).

5. Method

Procedure

The experiment took place at the department of psychology of the Philipp University of Marburg. Participants arrived in the lab for an experiment entitled “How do people perceive science?” After participants were provided with an overview of the experiment and a set of general instructions, informed consent was obtained. Each participant was seated in an individual cubicle and all further instructions and the questionnaire were delivered via computer. In order to manipulate participants’ group membership, participants were ostensibly categorized as either exhibiting a “holistic” or a “detailed” association style (see Scheepers and Ellemers, 2005) on the basis of a word- and figure-association task. Participants had to decide which word/figure out of four presented words/figures subjectively stuck out and did not fit in well. Participants were told that their association style would be assessed by the computer. All participants were informed that they exhibited a “holistic association style.” After receiving this feedback, they were asked to read a randomly chosen article on association styles that had ostensibly been published recently in a scientific journal. Depending on experimental conditions, they received (a) an identity-threatening, (b) a neutral (control), or (c) an identity-affirming article. The manipulation material was a (faked) typical first page of a research article consisting of a short presentation of the author (name, title, university affiliation, research interest, and picture) and the abstract. Whereas the author description was held constant across all conditions, the abstract was altered depending on conditions. In the threat condition, participants read about an experiment showing that people with a holistic association style are less prosocial compared to people with a detailed association style. In the affirmation condition, people with a holistic association style were found to be more prosocial, and in the control condition, people with a holistic association style did not differ from people with a detailed association style regarding their prosocial behavior.

Immediately after reading the article, participants had to answer several multiple choice comprehension questions and were only allowed to proceed when these questions were answered correctly. This was done in order to ensure that they had read the article conscientiously and attentively. After the comprehension questions, the dependent variables and manipulation check measures were assessed.¹

Measures

The items of the scales consisting of more than one item can be found in Supplementary Appendix (including the items’ descriptive statistics) (Supplementary Appendix is available at: http://pus.sagepub.com/content/by/supplemental-data). Participants were neither forced to answer the following items nor were they informed that they could skip items.

Participants were asked to evaluate the study and the competence of the author (adapted from Nauroth et al., 2014, 2015). Participants’ evaluation was assessed with six items (Cronbach’s α = .76) and the competence of the author with four items (Cronbach’s α = .75). Response scales ranged from 1 (not at all true) to 6 (very much true), with higher values indicating a more positive evaluation of the study and a positive assessment of the competence of its author.

Next, participants were asked to indicate the degree to which they think that scientists working in the field of association styles share the opinion that a holistic association style has positive or negative consequences (predominant opinion in the scientific community). This was done by varying the sizes of two circles (one for the number of scientists having the opinion that a holistic association style has negative consequences and the other for the number of scientists having the opinion that a holistic association style has positive consequences); the bigger the circle, the more supporters the respective opinion had. The pair of circles varied in seven graduations from 1 (clear majority for negative consequences; i.e. very large circle for the number of scientists believing in negative consequences and very small circle for the number of scientists believing in positive consequences) to 7 (clear majority for positive consequences; i.e. very large circle for the number of scientists believing in positive consequences and very small circle for the number of scientists believing in negative consequences). This item served as a manipulation check; it indicates the degree to which participants believed the study results. Participants in the affirmation condition should perceive positive effects to be the dominating scientific opinion, participants in the threat condition should perceive negative effects to be the dominating scientific opinion, and participants in the control condition should perceive negative and positive effect opinions as equally prevalent.

Homogeneity of the scientific community was assessed with the item “How similar are the members of the scientific community among each other?” (ranging from 1 (extremely different) to 6 (extremely similar)).

Typicality of the author was assessed both directly and indirectly. Direct typicality was measured with two scales, typicality of the author and typicality of the author’s opinion. Typicality of the author was assessed with two items (“How typical is the author for the scientific community in general?” ranging from 1 (not at all typical) to 6 (extremely typical) and “How similar is the author to the scientific community?” with a response scale varying the distance between a small circle for the author and a large circle for the community ranging from 1 (author circle far outside the community circle) to 7 (author being exactly in the middle); Cronbach’s α = .54; see Riek et al., 2013; Schubert and Otten, 2002). Typicality of the author’s opinion was assessed by four items (“How many percent of the scientific community share the opinion of the author?” implemented via a sliding bar returning integer percentages ranging from 0% to 100%, as well as three items using Likert-type responses ranging from 1 (not at all true) to 6 (very much true); Cronbach’s α = .76). Due to different response scales, we z-standardized responses prior to computing the respective composite. High values on the direct measures of typicality indicate a high typicality of the author for the scientific community.

In order to include an unobtrusive measure of typicality, we additionally assessed typicality indirectly. Following Wenzel et al. (2003), participants rated the author and the scientific community separately on 15 traits using 7-point semantic differentials (ranging from −3 to +3). Ten of these traits were science-related (e.g. unscientific–scientific, subjective–objective, stupid–intelligent) and the remaining five were filler traits (e.g. unattractive–attractive, indifferent–passionate). We computed Euclidean distances between the author and the community on the science-related traits as a standard measure of profile dissimilarity (see Machunsky et al., 2009). The obtained Euclidian distances were z-standardized and multiplied with −1, such that high values indicate a high typicality.

Furthermore, participants were asked to indicate the reputation of the author within the scientific community on four items (Cronbach’s α = .84). Response scales ranged from 1 (not at all true) to 6 (very much true), with higher values indicating a higher reputation within the community.

Additionally, participants were asked how strongly they identified with the group of people exhibiting a holistic association style (“I identify with the group of holistic-associating people;” response scales ranging from 1 (not at all true) to 6 (very much true); see Postmes et al., 2013) and how strongly they feel threatened by the research results in their social identity as a holistic-associating person (response scales ranging from 1 (not at all true) to 6 (very much true); Cronbach’s α = .78). These two measures served as additional manipulation checks. We expected that identification as a holistic-associating person should only be increased in the affirmation condition and threat perceptions should only be increased in the threat condition.

Finally, demographic information was assessed. Upon completion of the study, which took about 30 minutes, participants were debriefed and thanked. The mean and standard deviations (SDs) of all measured variables can be found in Table 1 and their intercorrelations in Table 2.

Table 1.

Descriptive statistics.

Variable	M (SD)
	Threat	Control	Affirmation
Predominant opinion in scientific community	3.78 (1.03)	4.27 (0.66)	5.04 (0.93)
Identification	3.27 (1.62)	3.32 (1.64)	4.13 (1.47)
Threat	2.56 (1.31)	1.29 (0.53)	1.41 (0.65)
Evaluation of the study	3.28 (0.91)	3.72 (0.80)	4.09 (0.94)
Competence of the author	4.33 (0.93)	4.77 (0.77)	4.77 (0.88)
Reputation of the author	3.31 (0.86)	3.74 (0.91)	3.92 (0.82)
Direct measures of typicality
Typicality of author	−0.28 (0.81)	0.08 (0.79)	0.23 (0.82)
Typicality of author’s opinion	−0.46 (0.70)	0.27 (0.75)	0.23 (0.65)
Indirect measures of typicality
Typicality of author	−0.39 (1.35)	0.12 (0.66)	0.31 (0.67)
Homogeneity of scientific community	3.24 (1.09)	3.57 (0.87)	3.20 (1.19)

M: mean; SD: standard deviation.

N = 138.

Table 2.

Correlations between variables (Pearson’s r).

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)
Predominant opinion in scientific community (1)	1.00
Identification (2)	.17*	1.00
Threat (3)	−.20*	−.04	1.00
Evaluation of the study (4)	.23**	.26**	−.26**	1.00
Competence of the author (5)	.23**	.13	−.23**	.69***	1.00
Reputation of the author (6)	.21*	.26**	−.07	.50***	.38***	1.00
Direct measures of typicality
Typicality of author (7)	.11	.15	−.09	.39***	.34***	.42***	1.00
Typicality of author’s opinion (8)	.12	.08	−.30***	.51***	.42***	.51***	.65***	1.00
Indirect measures of typicality
Typicality of author (9)	.14	.01	−.36***	.33***	.28**	.30***	.32***	.35***	1.00
Homogeneity of scientific community (10)	.00	−.16	−.08	−.23**	−.14	−.05	−.05	−.02	−.11	1.00

N = 138.

p < .05; **p < .01; and ***p < .001.

Power analysis and participants

Morton et al. (2006) and Nauroth et al. (2014) reported moderate effect sizes between their control and experimental conditions on their evaluative measures. Since we conducted a laboratory study (which should lead to less noise in our data compared to these previous studies), we expected medium to large effects on our dependent variables (Cohen’s d = .65). In order to obtain a power of .80 when comparing each experimental condition with the control group, power analysis indicated a minimum sample of N = 117 (G*Power 3; see Faul et al., 2007). As suggested by Oppenheimer et al. (2009), the survey also comprised attention checks in the latter part of the experiment (“If I complete this questionnaire attentively and focused, I do not respond to this item,” and “My association style is [holistic, detailed]”). Based on the drop-out rates reported by Oppenheimer et al. (2009), we expected a failure rate of 25% on these checks and therefore tried to sample at least 156 participants.

In total, 160 students of the Philipp University of Marburg (studying various subjects: 59% social sciences (including psychology), 20% natural sciences, 11% humanities, 10% teacher training) agreed to participate and either received course credit or €6. Twenty-two participants (14%) failed the attention checks leading to a final sample of 138 participants (67% female, M_age = 24, SD_age = 2.99),² which well exceeds the minimal sample size indicated by the power analysis.

6. Results

We tested our hypotheses by comparing the control condition to both experimental conditions (i.e. affirmation threat) on each dependent variable employing a priori contrasts (contrast 1: 0 = affirmation, −1 = control, 1 = threat; contrast 2: 1 = affirmation, −1 = control, 0 = threat). Thus, the first contrast directly compares the threat versus control conditions; the second contrast directly compares the affirmation versus control conditions. The results of both contrast analyses can be found in Table 3. Due to heteroscedastic SDs, we used Welch’s t statistic (Welch, 1947) to probe the contrasts and Glass’ g coefficient (Hedges and Olkin, 1985) to estimate the respective effect size. For reasons of readability, we refrain from presenting all test statistics in the text but refer readers to Table 3 for this information. Additionally, we ran the same set of analysis and controlled for age and gender. Since this did not alter the pattern of results, we do not report these results here.

Table 3.

Results of the contrast analysis.

Variable	Contrast 1 (threat vs control)				Contrast 2 (affirmation vs control)
	t	p	g	95% CI	t	p	g	95% CI
Predominant opinion in scientific community	2.81	.006	−0.75	(−1.19, −0.31)	4.53	<.001	1.17	(0.69, 1.64)
Identification	0.16	.87	−0.03	(−0.44, 0.38)	2.47	.02	0.50	(0.07, 0.91)
Threat	6.25	<.001	2.41	(1.76, 3.06)	0.96	.34	0.23	(−0.18, 0.64)
Evaluation of the study	2.48	.02	−0.55	(−0.97, −0.12)	2.03	.05	0.47	(0.04, 0.89)
Competence of the author	2.48	.02	−0.57	(−0.99, −0.14)	0.03	.98	0.01	(−0.41, 0.42)
Reputation of the author	2.35	.02	−0.48	(−0.89, −0.06)	0.94	.35	0.19	(−0.23, 0.61)
Direct measures of typicality
Typicality of author	2.20	.03	−0.46	(−0.88, −0.04)	0.84	.35	0.18	(−0.24, 0.60)
Typicality of author’s opinion	4.85	<.001	−0.98	(−1.43, −0.51)	0.31	.76	−0.06	(−0.48, 0.36)
Indirect measures of typicality
Typicality of author	2.35	.02	−0.78	(−1.22, −0.34)	1.35	.36	0.29	(−0.13, 0.71)
Homogeneity of scientific community	1.68	.10	−0.37	(−0.78, 0.05)	1.59	.12	−0.42	(−0.84, 0.01)

CI: confidence interval.

N = 138. T-values based upon Welch’s t-test. Bold-type contrasts are significant at p < .05. Positive (negative) g-values indicate a higher (lower) mean in the experimental (i.e. threat and affirmation) compared to the control condition.

Manipulation checks

Our manipulation successfully affected participants’ perceptions of the predominant opinion in the field in both experimental conditions: Participants in the threat (vs control) condition less strongly believed that the majority of scientists thinks that a holistic association style has positive consequences (M_threat = 3.78, M_control = 4.27), whereas the opposite was true for the affirmation versus control contrast (M_affirmation = 5.04). As expected, the affirmative article increased identification with the in-group (M_affirmation = 4.13, M_control = 3.32), whereas the threatening article did not (M_threat = 3.27). The opposite pattern of results emerged on the perceived threat scale: only the threatening article increased threat perceptions (M_threat = 2.56, M_control = 1.29), whereas the affirmative article did not alter threat perceptions (M_affirmation = 1.41). These results indicate that our manipulation was successful in altering people’s perception of the predominant opinion in the scientific community and also concerning the two assumed psychological process variables affecting a biased assimilation of research results—namely, the degree of threat emanating from scientific findings and the affirmative appeal of scientific findings affecting the in-group.

Main analysis

We expected that people’s evaluation of the study would be most positive in the affirmation condition and most negative in the threat condition. Investigating participants’ evaluation of the study revealed that the threat condition differed significantly from the control condition (M_threat = 3.28, M_control = 3.72), and this was also the case for the affirmation condition (M_affirmation = 4.09). The assessment of the competence of the author only differed between the threat and the control condition (M_threat = 4.33, M_control = 4.77) and not between the affirmation and the control condition (M_affirmation = 4.77). These results were also mirrored in participants’ assessment of the reputation of the author within the scientific community and were only affected by the threatening article (M_threat = 3.31, M_control = 3.74), whereas the affirming article did not alter reputation assessments (M_affirmation = 3.92).

Concerning typicality perceptions, a similar pattern emerged: directly assessed typicality of the author differed between the threat and control condition (M_threat = −0.28, M_control = 0.08) but not between the affirmation and control condition (M_affirmation = 0.23). This was also the case for typicality of the author’s opinion (M_affirmation = 0.23, M_control = 0.08, M_threat = −0.28) and for indirectly assessed typicality of the author (M_threat = −0.39, M_control = 0.12, M_affirmation = 0.31). Notably, no reliable differences between conditions were found for homogeneity of the scientific community (M_threat = 3.24, M_control = 3.57, M_affirmation = 3.20).

7. Discussion

The present research investigated the effects of social identity-affirming versus identity-threatening scientific findings on perceptions of scientists and the scientific community. Participants read a threatening, an affirming, or a neutral study; evaluated the study; and assessed the competence, reputability, and typicality of its author and the homogeneity of the respective scientific community. Participants who read the threatening study perceived the author to be less competent, less reputable, and less typical for the scientific community, but they did not perceive the scientific community as more homogeneous than participants who read the neutral study. Furthermore, they evaluated the study more negatively. Surprisingly, reading an affirming study affected neither author- nor community-related perceptions: social identity–affirming research findings only led to more positive evaluations of the study. Thus, concerning the effects of social identity–threatening research findings, our results favor the subtyping hypothesis, which suggests that researchers publishing threatening research are not seen as prototypical members of the scientific community. However, concerning affirming research findings, our results favor the homogeneity hypothesis, which assumed no effect on homogeneity and typicality perceptions. In general, our findings corroborate the notion that social identity threat triggers a biased reception and perception of science and scientists—that is, people refrain from acknowledging the scientific community as a threatening homogeneous out-group and rather perceive the evidence and its researchers as a minority voice.

We derived our predictions from social identity theory and its central premise that people are motivated to maintain a positive identity. Although we did not directly test whether identity concerns are the motivational basis for the effects we found, our data allow an indirect investigation of this premise. In fact, one could have expected that a social identity threat leads to a decreased identification (see Owuamalam and Zagefka, 2011), but in our study social identification in the threat condition did not differ from social identification in the control condition (only the affirmation condition led to an increased identification). Subtyping threatening researchers as unscientific and disreputable may constitute an effective identity protection strategy buffering the social self-concept from negative implications. Thus, interpreting our results along this line, subtyping researchers may indeed be functional for maintaining a positive social identity.

Another point worth discussing is the role of typicality in our study. One could argue that typicality perceptions mediate the effect of a social identity threat on science evaluations. Although the design of our study does not allow for a direct investigation of this hypothesis (we neither manipulated typicality perceptions nor assessed typicality before the evaluative measures), typicality perceptions and evaluation of the study correlated significantly with participants’ evaluation of the study (.33 ≤ rs ≤ .51). Taking research on source credibility into consideration, this line of research shows that when people perceive somebody as an expert they are also more likely to trust his or her assessments (Chaiken and Maheswaran, 1994). Thus, one could cautiously interpret these correlations as indicating a mediating effect of typicality: the more strongly laypersons perceive researchers as prototypical scientists and, therefore, trustworthy and knowledgeable, the more positively they should evaluate their research. To test this reasoning with our data, we additionally analyzed the indirect effect of the threat (vs control) condition on evaluation via typicality perceptions by inspecting the bias-corrected 95% confidence intervals obtained by bootstrapping using Hayes’ (2008) PROCESS macro (10,000 resamples). Concerning the direct measures of typicality, both indirect effects were significant at the .05 level and only the indirect effect via the indirect measure failed to reach significance. This offers tentative support for the reasoning that typicality perceptions might be a mediator in explaining the effect of threatening research on science evaluations. However, since a reversed causality cannot be ruled out, future research should investigate whether typicality perceptions truly mediate the effect of social identity threat on evaluations by experimentally manipulating typicality.

Our findings resonate with the “hostile media effect” that describes partisans’ tendency to ascribe a biased stance of media reports in favor of the other side (e.g. Gunther and Liebhart, 2006). In accordance with our approach, Hartmann and Tanis (2013) investigated the hostile media effect from a social identity perspective. They showed that in-group identification and the societal group status qualify the hostile media effect such that the effect only occurs for strong identifiers and group members perceiving the in-group as having a low societal status. Notably, the hostile media effect occurs for partisans on both sides of a controversial issue toward the same, partisan-neutral message. Recently, Sjöström et al. (2013) theorized about a “hostile science effect” as a counterpart of the hostile media effect. Taking our own and Hartmann and Tanis’ results together, a hostile science effect might indeed exist. Future research should therefore investigate whether antagonistic groups of laypersons are biased against even factually neutral (but still group-relevant) scientific findings.

Implications for public engagement with science

Our findings have immediate implications for public engagement with science activities. When laypersons perceive scientists as less competent, less reputable, and not representative of the scientific community and the scientist’s opinion as deviating from the current scientific state-of-the-art, laypersons might be less willing to participate in constructive discussions (Schrodt et al., 2009). Furthermore, our mediation analysis suggests that these negative perceptions deepen the trench between scientists and laypersons concerning the current scientific state-of-the-art. We speculate that these biases might actually even lead to engagement activities to backfire: instead of developing a mutual understanding they might intensify laypersons’ misconceptions about the scientific state-of-the-art. Corroborating this hypothesis, Binder et al. (2011) demonstrated that discussions about controversial science topics may in fact polarize different groups around a priori positions. Additional preliminary support for this hypothesis can also be found in case studies about public engagement activities in controversial socio-scientific issues. Some of these reports (for two examples, see Lezaun and Soneryd, 2007) indicate problems to maintain a productive atmosphere between laypersons and experts in the discussion sessions.

Besides these practical implications, our results also add further evidence to the growing body of literature questioning the validity of the deficit model in science communication according to which people’s attitudes toward science are mainly determined by their knowledge about science (Sturgis and Allum, 2004). We demonstrated that social identity concerns profoundly influence laypersons’ perceptions and evaluations of scientific results regardless of laypersons’ knowledge. However, our results also question whether involving laypersons in policy decision processes based upon scientific evidence is reasonable in all socio-scientific issues. Particularly when the scientific evidence has potential negative consequences for social groups, our research suggests that laypersons may be prone to biases based upon their social affiliations. For example, if regular video game players were involved in decision-making processes concerning potential sales restrictions of violent video games, they would be likely to perceive scientific evidence demonstrating detrimental effects of violent video games as shoddy and the respective researchers as disreputable (Greitemeyer, 2014; Nauroth et al., 2014, 2015).

Limitations of the present research

Our results generally support the notion that researchers publishing threatening scientific findings are perceived as atypical and are devalued. However, one might argue that the external validity of our study was limited. First, minimal group paradigms have often been criticized for lacking ecological validity (e.g. Rabbie and Schot, 1990). This is also true for our study since participants only had a very limited sense about what it means to be a member of the group they were artificially assigned to. However, this is also advantageous since it allowed us to investigate the unbiased effect of group membership per se on science evaluation. Extending this argument, effects found in a minimal group paradigm may even be more generalizable compared to effects found in natural group settings, precisely because minimal group effects are not confounded with group specifics inherently connected to natural groups (cf. Diehl, 1990).

Second, since we only tested university students, our sample is not necessarily representative for a general population. Undergraduate students might be more confident to evaluate research endeavors negatively compared to less educated laypersons. However, our results replicate earlier findings from real group settings (Nauroth et al., 2014, 2015), which also demonstrated similar biasing effects in other populations. Furthermore, our predictions are based on well-established findings from group perception research, which have been shown to be applicable to a wide array of populations (Abrams and Hogg, 1999). Nonetheless, future studies could recruit participants from a more heterogeneous population in order to test the replicability of our findings.

Conclusion

On a theoretical level, our research contributed to the literature in the domains of science communication and group perception. Science communication has just begun to explicitly include and investigate group influences on science reception (Postmes, 2015). Likewise, social identity models can be tested and refined by applying them to science reception phenomena (Bliuc et al., 2015; Morton et al., 2006; Nauroth et al., 2014, 2015). Against this background, our research demonstrates the importance of considering social and group influences on laypersons’ science reception. On a practical level, our research shows that science reception is affected by group identities and that this influence spans from negatively biased evaluations to distorted perceptions of scientists. Furthermore, it points to psychological obstacles when conducting public engagement with science activities about controversial socio-scientific issues.

Footnotes

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by German Research Foundation (Deutsche Forschungsgemeinschaft) grants (no. GO 1674/2-1 and no. GO 1674/2-2) to the second author.

Notes

Author biographies

Peter Nauroth is a Research Assistant at the Philipp University of Marburg, Germany. His research focuses on social identity, science communication, and social cognition.

Mario Gollwitzer is Professor for methodology and social psychology at Philipp University of Marburg, Germany. His research focuses on (a) social psychological analyses of individuals’ reactions to injustice, (b) applied methodology in the social sciences, and (c) laypersons’ perceptions of and reactions to scientific information.

Henrik Kozuchowski is Psychologist and a Trainee for psychotherapy at the IPP in Bielefeld, Germany. His research focuses on the influence of social characteristics on science reception and science communication.

Jens Bender is a Research Assistant at the University of Koblenz-Landau. His research focuses on morality, threat, and motivated information processing.

Tobias Rothmund is Assistant Professor for political psychology at the University of Koblenz-Landau and visiting professor at the Technische Universität Darmstadt, Germany. His research focuses on how moral dispositions and mass media communication conjointly shape political attitudes and behavior.

References

Abrams

Hogg

(1999) Social Identity and Social Cognition. Malden, MA: Blackwell Publishing.

Barlow

Paolini

Pedersen

Hornsey

Radke

Harwood

et al . (2012) The contact caveat negative contact predicts increased prejudice more than positive contact predicts reduced prejudice. Personality and Social Psychology Bulletin 38(12): 1629–1643.Available at: http://doi.org/10.1177/0146167212457953

Binder

Scheufele

Brossard

Gunther

(2011) Interpersonal amplification of risk? Citizen discussions and their impact on perceptions of risks and benefits of a biological research facility. Risk Analysis 31(2): 324–334.

Bliuc

A-M

McGarty

Thomas

Lala

Berndsen

Misajon

(2015) Public division about climate change rooted in conflicting socio-political identities. Nature Climate Change 5: 226–229. Available at: http://doi.org/10.1038/nclimate2507

Carpini

MXD

Cook

Jacobs

(2004) Public deliberation, discursive participation, and citizen engagement: A review of the empirical literature. Annual Review of Political Science 7: 315–344. Available at: http://doi.org/10.1146/annurev.polisci.7.121003.091630

Chaiken

Maheswaran

(1994) Heuristic processing can bias systematic processing: Effects of source credibility, argument ambiguity, and task importance on attitude judgment. Journal of Personality and Social Psychology 66(3): 460–473. Available at: http://doi.org/10.1037/0022-3514.66.3.460

Corneille

Yzerbyt

Rogier

Buidin

(2001) Threat and the group attribution error: When threat elicits judgments of extremity and homogeneity. Personality and Social Psychology Bulletin 27(4): 437–446. Available at: http://doi.org/10.1177/0146167201274005

Crawford

Sherman

Hamilton

(2002) Perceived entitativity, stereotype formation, and the interchangeability of group members. Journal of Personality and Social Psychology 83(5): 1076–1094. Available at: http://doi.org/10.1037//0022-3514.83.5.1076

De Hoog

(2013) Processing of social identity threats: A defense motivation perspective. Social Psychology 44(6): 361–372. Available at: http://doi.org/10.1027/1864-9335/a000133

10.

Dépret

Fiske

(1999) Perceiving the powerful: Intriguing individuals versus threatening groups. Journal of Experimental Social Psychology 35: 461–480. Available at: http://doi.org/10.1006/jesp.1999.1380

11.

Diehl

(1990) The minimal group paradigm: Theoretical explanations and empirical findings. European Review of Social Psychology 1(1): 263–292. Available at: http://doi.org/10.1080/14792779108401864

12.

Ellemers

Spears

Doosje

(2002) Self and social identity. Annual Review of Psychology 53(1): 161–186. Available at: http://doi.org/10.1146/annurev.psych.53.100901.135228

13.

Faul

Erdfelder

Lang

A-G

Buchner

(2007) G* Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behavior Research Methods 39(2): 175–191. Available at: http://doi.org/10.3758/BF03193146

14.

Gilbert

Mulkay

(1984) Opening Pandora’s Box: A Sociological Analysis of Scientists’ Discourse. Cambridge: Cambridge University Press.

15.

Gregory

Lock

(2008) The evolution of “Public Understanding of Science”: Public engagement as a tool of science policy in the UK. Sociology Compass 2(4): 1252–1265. Available at: http://doi.org/10.1111/j.1751-9020.2008.00137.x

16.

Greitemeyer

(2014) I am right, you are wrong: How biased assimilation increases the perceived gap between believers and skeptics of violent video game effects. PLoS ONE 9(4): e93440. Available at: http://doi.org/10.1371/journal.pone.0093440

17.

Gunther

Liebhart

(2006) Broad reach or biased source? Decomposing the hostile media effect. Journal of Communication 56(3): 449–466. Available at: http://doi.org/10.1111/j.1460-2466.2006.00295.x

18.

Hamilton

Gifford

(1976) Illusory correlation in interpersonal perception: A cognitive basis of stereotypic judgments. Journal of Experimental Social Psychology 12(4): 392–407. Available at: http://doi.org/10.1016/S0022-1031(76)80006–6

19.

Hamilton

Dugan

Trolier

(1985) The formation of stereotypic beliefs: Further evidence for distinctiveness-based illusory correlations. Journal of Personality and Social Psychology 48(1): 5–17. Available at: http://doi.org/10.1037/0022-3514.48.1.5

20.

Hartmann

Tanis

(2013) Examining the hostile media effect as an intergroup phenomenon: The role of ingroup identification and status. Journal of Communication 63(3): 535–555. Available at: http://doi.org/10.1111/jcom.12031

21.

Hayes

(2008) Introduction to Mediation, Moderation, and Conditional Process Analysis: A Regression-Based Approach. New York, NY: Guilford Press.

22.

Hedges

Olkin

(1985) Statistical Methods for Meta-Analysis. Orlando, FL: Academic Press.

23.

Kunda

(1990) The case for motivated reasoning. Psychological Bulletin 108(3): 480–498. Available at: http://doi.org/10.1037/0033-2909.108.3.480

24.

Lezaun

Soneryd

(2007) Consulting citizens: Technologies of elicitation and the mobility of publics. Public Understanding of Science 16(3): 279–297. Available at: http://doi.org/10.1177/0963662507079371

25.

Machunsky

Meiser

Mummendey

(2009) On the crucial role of mental ingroup representation for ingroup bias and the ingroup prototypicality-ingroup bias link. Experimental Psychology 56(3): 156–164. Available at: http://doi.org/10.1027/1618-3169.56.3.156

26.

Maier

Rothmund

Retzbach

Otto

Besley

(2014) Informal learning through science media usage. Educational Psychologist 49(2): 86–103. Available at: http://doi.org/10.1080/00461520.2014.916215

27.

Morton

Haslam

Postmes

Ryan

(2006) We value what values us: The appeal of identity-affirming science. Political Psychology 27(6): 823–838. Available at: http://doi.org/10.1111/j.1467-9221.2006.00539.x

28.

Nauroth

Gollwitzer

Bender

Rothmund

(2014) Gamers against science: The case of the violent video games debate. European Journal of Social Psychology 44(2): 104–116. Available at: http://doi.org/10.1002/ejsp.1998

29.

Nauroth

Gollwitzer

Bender

Rothmund

(2015) Social identity threat motivates science-discrediting online comments. PLoS ONE 10(2): e0117476. Available at: http://doi.org/10.1371/journal.pone.0117476

30.

Oppenheimer

Meyvis

Davidenko

(2009) Instructional manipulation checks: Detecting satisficing to increase statistical power. Journal of Experimental Social Psychology 45(4): 867–872. Available at: http://doi.org/10.1016/j.jesp.2009.03.009

31.

Ostrom

Sedikides

(1992) Out-group homogeneity effects in natural and minimal groups. Psychological Bulletin 112(3): 536. Available at: http://doi.org/10.1037/0033-2909.112.3.536

32.

Owuamalam

Zagefka

(2011) Downplaying a compromised social image: The effect of metastereotype valence on social identification: Metastereotyping and social identification. European Journal of Social Psychology 41(4): 528–537. Available at: http://doi.org/10.1002/ejsp.805

33.

Postmes

(2015) Psychology: Climate change and group dynamics. Nature Climate Change 5: 195–196. Available at: http://doi.org/10.1038/nclimate2537

34.

Postmes

Haslam

Jans

(2013) A single-item measure of social identification: Reliability, validity, and utility. British Journal of Social Psychology 52(4): 597–617. Available at: http://doi.org/10.1111/bjso.12006

35.

Rabbie

Schot

(1990) Group behavior in the minimal group paradigm: Fact or fiction? In: Drenth

Sergeant

Takens

(eds) European Perspectives in Psychology. New York, NY: Wiley, pp. 253–263.

36.

Richards

Hewstone

(2001) Subtyping and subgrouping: Processes for the prevention and promotion of stereotype change. Personality and Social Psychology Review 5(1): 52–73. Available at: http://doi.org/10.1207/S15327957PSPR0501_4

37.

Riek

Mania

Gaertner

(2013) Reverse subtyping: The effects of prejudice level on the subtyping of counterstereotypic outgroup members. Basic and Applied Social Psychology 35(5): 409–417. Available at: http://doi.org/10.1080/01973533.2013.823616

38.

Risen

Gilovich

Dunning

(2007) One-shot illusory correlations and stereotype formation. Personality and Social Psychology Bulletin 33(11): 1492–1502. Available at: http://doi.org/10.1177/0146167207305862

39.

Rothgerber

(1997) External intergroup threat as an antecedent to perceptions in in-group and out-group homogeneity. Journal of Personality and Social Psychology 73(6): 1206–1212. Available at: http://doi.org/10.1037/0022-3514.73.6.1206

40.

Rubin

Hewstone

Voci

(2001) Stretching the boundaries: Strategic perceptions of intragroup variability. European Journal of Social Psychology 31(4): 413–429. Available at: http://doi.org/10.1002/ejsp.51

41.

Scheepers

Ellemers

(2005) When the pressure is up: The assessment of social identity threat in low and high status groups. Journal of Experimental Social Psychology 41(2): 192–200. Available at: http://doi.org/10.1016/j.jesp.2004.06.002

42.

Schrodt

Witt

Turman

Myers

Barton

Jernberg

(2009) Instructor credibility as a mediator of instructors’ prosocial communication behaviors and students’ learning outcomes. Communication Education 58(3): 350–371. Available at: http://doi.org/10.1080/03634520902926851

43.

Schubert

Otten

(2002) Overlap of self, ingroup, and outgroup: Pictorial measures of self-categorization. Self and Identity 1(4): 353–376. Available at: http://doi.org/10.1080/152988602760328012

44.

Sinatra

Kienhues

Hofer

(2014) Addressing challenges to public understanding of science: Epistemic cognition, motivated reasoning, and conceptual change. Educational Psychologist 49(2): 123–138. Available at: http://doi.org/10.1080/00461520.2014.916216

45.

Sjöström

Sowka

Gollwitzer

Klimmt

Rothmund

(2013) Exploring audience judgments of social science in media discourse: The case of the violent video games debate. Journal of Media Psychology: Theories, Methods, and Applications 25(1): 27–38. Available at: http://doi.org/10.1027/1864-1105/a000077

46.

Stark

Flache

Veenstra

(2013) Generalization of positive and negative attitudes toward individuals to outgroup attitudes. Personality and Social Psychology Bulletin 39(5): 608–622. Available at: http://doi.org/10.1177/0146167213480890

47.

Stilgoe

Lock

Wilsdon

(2014) Why should we promote public engagement with science? Public Understanding of Science 23(1): 4–15. Available at: http://doi.org/10.1177/0963662513518154

48.

Sturgis

Allum

(2004) Science in society: Re-evaluating the deficit model of public attitudes. Public Understanding of Science 13(1): 55–74. Available at: http://doi.org/10.1177/0963662504042690

49.

Tajfel

(1970) Experiments in intergroup discrimination. Scientific American 223(5): 96–102.

50.

Tajfel

Turner

(1979) An integrative theory of intergroup conflict. In: Austin

Worchel

(eds) The Social Psychology of Intergroup Relations. Monterey, CA: Brooks/Cole, pp. 33–47.

51.

Tajfel

Turner

(1986) The social identity theory of intergroup behavior. In: Worchel

Austin

(eds) The Social Psychology of Intergroup Relations. Chicago, IL: Nelson-Hall, pp. 7–24.

52.

Weber

Crocker

(1983) Cognitive processes in the revision of stereotypic beliefs. Journal of Personality and Social Psychology 45(5): 961–977. Available at: http://doi.org/10.1037/0022-3514.45.5.961

53.

Welch

(1947) The generalization of “Student’s” problem when several different population variances are involved. Biometrika 34(1–2): 28–35. Available at: http://doi.org/10.1093/biomet/34.1-2.28

54.

Wenzel

Mummendey

Weber

Waldzus

(2003) The ingroup as pars pro toto: Projection from the ingroup onto the inclusive category as a precursor to social discrimination. Personality and Social Psychology Bulletin 29(4): 461–473. Available at: http://doi.org/10.1177/0146167202250913

55.

Wilder

(1978) Reduction of intergroup discrimination through individuation of the out-group. Journal of Personality and Social Psychology 36(12): 1361. Available at: http://doi.org/10.1037/0022-3514.36.12.1361

56.

Ziman

(1991) Public understanding of science. Science, Technology & Human Values 16(1): 99–105. Available at: http://doi.org/10.1177/016224399101600106

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.18 MB