Abstract
This study examined if creating intimacy in a group discussion is more effective toward reaching consensus about climate change than a focus on information. Participants were randomly assigned to either a group that spent the first part of an online discussion engaging in self-disclosure and focusing on shared values (intimacy condition) or discussing information from an article about climate change (information condition). Afterward, all groups were given the same instructions to try to come to group consensus on their opinions about climate change. Participants in the intimacy condition had higher ratings of social cohesion, group attraction, task interdependence, and collective engagement and lower ratings of ostracism than the information condition. Intimacy groups were more likely to reach consensus, with ostracism and the emotional tone of discussion mediating this effect. Participants were more likely to change their opinion to reflect that climate change is real in the intimacy than information condition.
Large-scale action is needed to stave off the worst effects of climate change. Indeed, our carbon dioxide budget to limit warming to 1.5 C is dwindling (Mercator Research Institute, 2020). This will require broad support and collaboration between members of the public, global actors, and even scholars in the sciences and humanities (Tschirhart and Bloomfield, 2020). Polling trends indicate that a growing number of Americans are concerned and alarmed about climate change (Leiserowitz et al., 2019). However, concern does not always translate into advocacy or action (Bloomfield, 2019), so to shift the needle into political momentum, it is imperative to continue exploring strategies for climate communication. This includes the understudied role of group dynamics and interpersonal conversations as a potential avenue for public scientific engagement (Goldberg et al., 2019).
Given the scientific nature of climate change, one strategy for climate communication is to educate people to encourage action and political support. Some research proposes that establishing baseline beliefs, such as the scientific consensus on climate change, can shift public opinion toward support for climate action (e.g. van der Linden et al., 2019). Accurate information is an important part of the public’s understanding of science, yet additional research has found that focusing on information and facts alone is limited in its ability to drive attitude change on scientific topics (McFadden, 2016; Miller, 2001). In the case of climate change, a focus on information may even backfire due to the entanglement of climate change with divisive political ideology in the United States (Bloomfield et al., 2020; Hart and Nisbet, 2012; Trevors et al., 2016). Therefore, frameworks for discussing climate change in addition to information exchange are needed.
Drawing from research on the importance of creating common ground, social identity, cohesion, and social influence in groups, we examine if fostering intimacy among group members may affect participants’ attitudes and facilitate reaching group consensus. First, we review research on the knowledge deficit model. Then, we present an alternative method by fostering commonality and intimacy through discussion of self-defining questions.
1. Information and facts
According to the knowledge deficit model of science communication, the public’s misunderstanding about climate change can be largely attributed to lack of information and discussion of it. Therefore, increased communication of facts and information contributes to the alignment of public opinion with the scientific majority (Bauer et al., 2007; Nisbet, 2005; Simmons, 2007). This model proposes that more information and the associated deliberation of that information will help diminish the knowledge gap between scientists and lay public (Bauer et al., 2007; Hart and Nisbet, 2012). Cook (2017) argued that discussing the expert consensus on climate change “is a stepping stone” toward productive scientific conversations and that failing to correct misinformation and misconceptions can “delay substantive policy discussion” (p. 733). However, despite some support in the literature for the importance of science literacy, the knowledge deficit model has at times not received support across scientific topics and circumstances in terms of the predicted effects of increased information (Goidel et al., 1997; Moser and Dilling, 2007; Nisbet, 2005; Nisbet and Goidel, 2007). Dryzek and Lo (2015) note “more and better information does not have an effect” on the public’s opinion (p. 1). For example, on the issue of stem cell research, when presented with more information, people still tended to rely on pre-existing opinions and highly accessible information (Nisbet, 2005). With climate change specifically, some research has found that information does not reduce the knowledge gap and increased knowledge may even lead to disagreement with the scientific consensus on climate change (McFadden, 2016; Nisbet, 2005; Nisbet and Goidel, 2007).
Information may not be helpful for reaching consensus due to motivated reasoning and conflicts with identity. People are motivated to biasedly process information to be consistent with their beliefs, especially ones connected to their identity (Kahan et al., 2013; Lord et al., 1979). In the United States, the topic of climate change is ideological and politically partisan (Hart and Nisbet, 2012; Nisbet, 2005). A position on climate change can be an identity marker (Fiorina and Abrams, 2008), which people use to show their identity and to “provide the interpretative cues that differentiate the ‘self’ from the ‘other’” (Hart and Nisbet, 2012: 707). Therefore, focusing on information may be divisive with those who downplay climate change because it may activate a desire to protect their ideological identity (Nyhan and Reifler, 2010). The goal of identity protection can compete with understanding information, which may lead to information being processed biasedly. This can strengthen original opinions, increase polarization, and underscore differences (Lord et al., 1979; Prasad et al., 2009). Indeed, researchers found that when educated about climate change, perception of climate change as an environmental danger increased among Democrats but decreased among Republicans (Hamilton, 2011; Hamilton et al., 2010).
Trevors et al. (2016) note that individuals are motivated to reject information to “restore a sense of self-worth” (p. 343). Information that threatens one’s identity leads to negative emotion, which can reduce a person’s ability and willingness to learn from information or reach consensus (Trevors et al., 2016). This suggests that, in some cases, a focus on information can be counterproductive in that they lead to negative emotions which have adverse effects on openness to new information. In addition, motivated reasoning can emphasize one’s own ideological identity at the expense of creating a collective social identity and promoting consensus (Hart and Nisbet, 2012). Because information that threatens identity and self-concept can increase differentiation of self from others and negative emotions, we examine a different focus for climate communication. Instead, fostering commonality and intimacy and affirming people’s self-concept may facilitate consensus and avoid the potential backfire effects of information-only interventions.
2. Fostering intimacy
Environmental policy and research have often measured whether exposure to information and awareness will lead to a decision to act (e.g. Bamberg and Moser, 2007). Yet as Fritsche et al. (2018, p. 250) note and research on motivated reasoning supports, information is “interpreted in the social context. . .social identity processes determine this context.” The Social Identity Model of Pro-Environmental Action (SIMPEA; Fritsche et al., 2018) examines how social identity in a group can shape norms, emotions, and goals in relation to environmental behaviors. People define their self-concept both as a personal identity through individual characteristics and a social identity based on collective beliefs and norms of group membership (Tajfel and Turner, 1979).
Given that climate change cannot be solved without collective effort and that the consequences of climate change are so long term or unobservable to a single individual (Comstock and Hocks, 2016; Nerlich et al., 2010), fostering a group social identity may be an effective approach to climate communication. Social identity helps people to “adopt collective appraisals” (Fritsche et al., 2018: 246) and can be a means of persuasion. Social identity can move people beyond their disparate identities and ideologies, as people adopt norms and goals within their group. One way to facilitate a social identity is through establishing common ground, as social identities can arise from discussions about shared and common opinions or behaviors (Postmes et al., 2005; Thomas et al., 2016; Thomas et al., 2012). Scholars have argued that locating common values can open people up to persuasive messages (e.g. Bloomfield and Tscholl, 2018; Burke, 1950) and increase the personal relevance of climate change consequences. Using SIMPEA as a theoretical framework, we examine research on the importance of establishing commonality and research on group decision-making and cohesion.
Establishing a social identity in a group through discussion of commonalities can override differences. Dryzek and Lo (2015) found that the use of “bridging rhetoric,” which established connections between the members of a group who disagreed, helped climate change deniers reach consensus with climate change supporters on policies to combat climate change (Dryzek and Lo, 2015: 1). Bridging differences and establishing social identity become the foundation for further conversation and even persuasion, opening dialogues that may otherwise be difficult by the polarizing effects of information. Perceived boundaries and assumptions can be destabilized through a shared social identity, which “bridg[es] conflicting ideologies” and creates “larger unities” (Zappen, 2009: 280–281). Common ground and a social identity in a group can override other ideological identities (Fritsche et al., 2018). Social identification and common ground open up communicative possibilities for defusing conflicts and collaborating productively. When people receive affirmation for a social identity based on commonalities, they may be open to new information (Cohen et al., 2007), even when this information touches on strongly held political partisanship (Binning et al., 2010).
Sargent and Sue-Chan (2001) note that social interaction builds cohesion and interdependence among group members, facilitating a social identity. Research on group communication has highlighted the importance of establishing cohesion and psychological safety for effective collaboration (Pollack and Matous, 2019). Cohesion is both socio-emotional and task-related (Van Vianen and De Dreu, 2001). Socio-emotional cohesion reflects positive affective climate and attraction to a group (Graupenperger et al., 2019), whereas task cohesion reflects task interdependence and engagement (Van Vianen and De Dreu, 2001). Task interdependence refers to group members communicating effectively, listening to and relying on each other, and sharing knowledge; groups with better social dynamics have higher task interdependence (Sargent and Sue-Chan, 2001). Task engagement is a “positive, fulfilling, work-related state of mind that is characterized by vigor, dedication, and absorption” (Schaufeli et al., 2002: 74).
For groups to have higher social and task cohesion (task interdependence and engagement), group members need to communicate effectively to build trust, respect, and psychological safety. For example, Renger and Reese (2017) found that group members treating each other respectfully helps build a greater social identity, and Pollack and Matous (2019) note that self-disclosure builds trust. Edmondson (1999) defines psychological safety as members feeling like they can speak up and will not be punished or embarrassed for disagreement, which reflects a sense of interpersonal trust and respect. One way to establish psychological safety is through self-disclosure exercises (Pollack and Matous, 2019). In his review of Google’s research to improve team performance, Duhigg (2016) notes for psychological safety in teams, “we need to establish a bond” (p. 6). Self-disclosure can establish bonds and social identity and create a group norm toward more communication, engagement, and interdependence.
In addition, when groups feel more cohesion, attraction, and engagement, members feel more included and less ostracized. Ostracism involves feeling excluded or ignored by others in a group (Williams, 2008), which threatens member’s self-esteem, sense of social identity to the group, and sense of belonging (Williams et al., 2000). Psychological consequences of ostracism are immediate and strong (Smith and Williams, 2004; Williams and Nida, 2011), including increased negative emotions (Williams, 2009; Williams and Nida, 2011). Ostracism, by reducing a sense of belonging, leads to perceptions of social exclusion and low group cohesion (Williams et al., 2000). A focus on information may highlight differences, rather than commonalities, between group members. Van Swol et al. (2019) found that a focus on presenting new information in a group discussion increased feelings of ostracism in comparison to a condition in which discussion focused on what members already knew. Therefore, we hypothesize that when groups start discussion focusing on information, rather than building cohesion and identity through self-disclosure, feelings of ostracism will increase. In addition, members who feel ostracized often react with social avoidance (e.g. see Wesselmann et al., 2012), and this could reduce the ability to reach consensus. Thus, the feelings of ostracism are likely to mediate the effect of condition (intimacy or information) on the ability to reach consensus in the group.
Concepts, such as cohesion, attraction, and ostracism, are strongly connected to emotions. Fritsche et al. (2018) note in their SIMPEA model that emotions often mediate the effect of group-based processes on outcomes. Ostracism engenders more negative emotions (Williams, 2008; 2009), such as anxiety and anger. Social and task cohesion and attraction to a group generate more positive emotion. For example, Salas et al. (2015) note a relationship between cohesion and positive affect. One way to unobtrusively measure emotion in groups is through negative and positive emotion words. Generally, when members are more satisfied, identified with the group, and experience more positive emotions in a group discussion, this is reflected in more use of positive emotion words, while conflict and lower well-being are reflected with more negative emotion words (for review see van Swol and Kane, 2019). Therefore, the use of more positive emotion words should be positively correlated with social cohesion, attraction of the group, task interdependence, and collective engagement and negatively correlated with ostracism. While, the opposite relationships should be found for negative emotion words. Given that emotion words are related to our constructs, they are likely to mediate the effect of condition on reaching consensus. This allows us to directly examine how intimacy may affect communication processes that then affect outcomes, such as consensus. Thus, we predict the use of more positive than negative emotion words will mediate the effect condition on reaching consensus.
The creation of intimacy and commonality is explored here as potential inroads for fostering productive climate communication and consensus-building (Bloomfield et al., 2020). Aron et al. (1997) developed a method to generate interpersonal closeness. Their method includes self-disclosure and targeted questions. This temporarily produces closeness between study participants such that participants rate this relationship “at about the 30th percentile of ratings of one’s closest, deepest, most involved, and most intimate relationship” (Aron et al., 1997: 371). Developing intimacy can foster social identity, cohesion, and consensus, which may increase participants’ willingness to change their opinion about climate change. This study uses a shortened version of Aron et al. (1997) method to foster intimacy in a group discussion, to compare to the information condition, in which participants were instructed to directly and only focus on information underlying their opinions. To summarize, we hypothesize: Hypothesis 1: Participants in the intimacy condition will have higher ratings of (a) social cohesion, (b) group attraction, (c) task interdependence, and (d) collective engagement than the information condition. Hypothesis 2: Participants in the intimacy condition will have lower ratings of ostracism than the information condition. Hypothesis 3: (a) Participants in the intimacy condition will be more likely to reach consensus in their group than the information condition, (b) this will be mediated by lower ostracism, and (c) more use of positive than negative emotion language. Hypothesis 4: Participants in the intimacy condition will be more likely to change their opinion toward the view that climate change is really happening than in the information condition.
3. Method
Participants and design
Undergraduate participants from two large, public universities in the United States located in the Midwest (n = 233) and Southwest (n = 96) were randomly assigned to either the intimacy (n = 163, Midwest = 119, Southwest = 44) or information condition (n = 166, Midwest = 113, Southwest = 52). A priori power analysis using G*Power (Faul et al., 2007) suggested that, for our design, a total sample of 128 would be sufficient to detect a medium-size effect (f = .25) with 0.80 power and α = .05. The majority of participants were female (60.18%). There were no significant differences between universities in random assignment to condition, χ2 = .81, ϕ = −.05, p = .37. Participants were randomly assigned to a group (information n = 50, intimacy n = 47). Most groups included three (n = 51) or four members (n = 39), but due to variation in attendance, some groups only included two members (n = 5) or had five members (n = 2). Participants received extra credit.
Procedure
Participants signed up for a timeslot and arrived in person for the experiment. Each timeslot was randomly assigned to either intimacy or information condition. Participants were seated at a computer and were asked to give informed consent. Then, they were handed an article about climate change condensed and adapted from the New York Times (Gillis, 2017). This article was not “two-sided,” presenting both pro and con sides of the view that climate change is real, as this would create a false equivalency (Brüggemann and Engesser, 2016) and contradict near universal scientific consensus (Leiserowitz, 2019). The article discussed anthropogenic climate change as real, serious, and offered an overview of potential solutions. While reading, participants were randomly assigned to groups. After reading the article, participants were directed to their computer to answer questions about their opinion of climate change.
The experiment was conducted concurrently at both universities, so groups generally comprised participants from both universities. After answering initial individual opinion online, participants waited until all group members were ready and then a chat room opened for each group. Up to three groups could be run during an experimental session, and each group had their own chat room. On entering the chat room, groups in the information condition were given these instructions from the moderator: For the next 8-10 minutes, we would like you to talk with each other about information about climate change, focusing on what you know and on what you learned from the article. Please do not discuss your opinion about climate change at this time, but simply pick 4-5 pieces of information that you think are important for the group to focus on for later discussion.
In the intimacy condition, the instructions were, For the next 8-10 minutes, we would like you to talk with each other by following the directions below. Please do not discuss your opinion about climate change at this time, but simply answer the questions with each other. Tasks:
Greet your fellow conversation partners and introduce yourself
Answer the following questions to get to know one another
What would constitute a “perfect” day for you? What is the greatest accomplishment of your life? Find two things you and your group members appear to have in common.
We used these questions because an effective self-affirmation manipulation should be irrelevant to the domain of interest to avoid defensive attitudes arising when individuals discuss values (McQueen and Klein, 2006). After instructions, participants had 8 minutes to interact before the next part of the experiment. In the chat, each participant had a username that randomly merged a color and an animal (e.g. goldenlion, purplemonkey). Participants were told their group could include participants from their own and other participating university. After 8 minutes, the moderator entered the chat and stated: Now we want you to discuss your opinions about climate change as a group and try to come to consensus in answering the questions on the group questionnaire. The experimenter will pass out a paper copy of the questions you should answer in your group.
The experimenter gave each participant a paper copy of the questions the group was to answer (see measures). In addition, groups were asked to discuss whether the following were viable policies to mitigate climate change: instituting caps on carbon emissions and fining industries that go over this cap; raising taxes to invest funds into green and eco-friendly technology; taxing gasoline and high emissions/low efficiency cars; taxing meat and animal products; changing personal actions only (no government actions); or do nothing. Groups had 12 minutes to discuss and reach consensus. The moderator gave a 1-minute warning before the chat closed automatically. Throughout the experiment, the moderator monitored group discussion to ensure active participation. When the chat closed after 20 minutes (8-minute manipulation/12-minute discussion), participants were directed to a questionnaire. Participants were thanked, debriefed, and given extra credit.
Measures
Opinion
Before and after group discussion, participants rated their agreement with the statement “I believe that climate change is really happening” on a scale from 1 (strongly disagree) to 7 (strongly agree). In addition, groups were asked to reach consensus about their agreement with this statement. Group members were asked whether they were able to reach consensus by choosing between yes (1), somewhat (2), and no (3). After rating their opinion, participants were asked “How confident are you (is your group) in your opinions (their opinion) about climate change?” and answered on a scale from 1 (not at all confident) to 7 (very confident).
Social cohesion
Four items from Sargent and Sue-Chan (2001) measured social cohesion on a 7-point scale from 1 (strongly disagree) to 7 (strongly agree). Items included “I liked my group” and “I got along with members of my group.” The four items were averaged for an index of social cohesion (M = 5.98, SD = 1.03; Cronbach’s α = .92).
Group attraction
Eight items from Evans and Jarvis (1986) measured group attraction on a 7-point scale from 1 (strongly disagree) to 7 (strongly agree). Items included “I want to remain a member of this group,” “If it were possible to move to another group at this time, I would (reverse coded),” and “I felt included in the group.” Factor analysis indicated two items—“I feel it would have made a difference to the group if I were not there” and “I feel my absence would not matter to the group (reverse coded)”—did not load well. Therefore, six items were averaged to form a group attraction index (M = 5.45, SD = 0.88; Cronbach’s α = .73).
Task interdependence
Three items were adapted from Sargent and Sue-Chan (2001) to measure task interdependence on a 7-point scale from 1 (strongly disagree) to 7 (strongly agree). Items included “All group members’ opinions were taken into consideration” and “All group members listened to each other.” Items were averaged to form an index of task interdependence (M = 6.21, SD = 0.85; Cronbach’s α = .82).
Collective engagement
Six items adapted from Salanova et al. (2003) measured collective engagement on a 7-point scale from 1 (strongly disagree) to 7 (strongly agree). Items included “My group was involved in the task,” “My group felt enthusiastic about the task,” and “My group liked doing the task.” The items were averaged for an index of collective engagement (M = 5.62, SD = 1.00; Cronbach’s α = .91).
Ostracism
Three items developed from Van Swol et al. (2019) measured ostracism on a scale from 1 (strongly disagree) to 7 (strongly agree). Items included “I was excluded from the group discussion,” “Group members paid attention to my contributions (reverse coded),” and “The other group members ignored me.” Reliability of the three items was low (Cronbach’s α = .64); factor analysis indicated that the second item did not load well. Therefore, the first and third items were averaged for an index of ostracism (M = 5.36, SD = 1.13; Cronbach’s α = .69).
Tone of language
Transcripts of the 12-minute group discussion during which participants tried to reach consensus were analyzed with Linguistic Inquiry Word Count (LIWC) software (Pennebaker et al., 2015) to measure emotional tone of the language used. Emotional tone compares positive and negative emotion language. A higher number indicates a more upbeat and positive emotional tone; a lower number indicates more negative emotions, such as sadness, anxiety, and anger.
4. Results
Perception of group
To account for non-independence among individuals in interacting groups, multi-level modeling with group ID as a random factor with correlated error variance was used for these analyses. In addition, variables were also analyzed in hierarchical linear modeling with HLM 8 (shown in Table 1) to calculate an effect size for non-independent data. Type of group (intimacy/information) and group size were group-level independent variables. Participants’ responses were answered individually. Group size is included in analysis as a covariate. In support of Hypothesis 1A, intimacy groups had higher ratings of group cohesion than information groups, SE = 0.08, F(1, 91.82) = 11.59, p < .001; effect size was an increase in 0.58 standard deviation of group cohesion. The effect size was calculated by dividing γ02 from HLM analysis by the pooled standard deviation of the outcome variable. See Table 1 for means and standard deviations.
Perception of group between conditions.
SD: standard deviation.
p < .10 *p < .05, ** p < .01, ***p < .001.
In support of Hypothesis 1B, intimacy groups had higher ratings of group attraction in comparison with information groups, SE = 0.05, F(1, 94.30) = 5.95, p = .017; effect size was an increase in 0.39 standard deviation.
In support of Hypothesis 1C, intimacy groups had higher ratings of task interdependence in comparison with information groups, SE = 0.05, F(1, 91.06) = 10.83, p < .001; effect size was an increase in 0.63 standard deviation of interdependence.
In support of Hypothesis 1D, intimacy groups had higher ratings of collective engagement than information groups, SE = 0.06, F(1, 94.18) = 11.15, p < .001; effect size was an increase in 0.46 standard deviation of engagement.
In support of Hypothesis 2, intimacy groups had lower ratings of ostracism than information groups, SE = 0.04, F(1, 89.88) = 6.33, p = .01; effect size was an increase in 0.39 standard deviation of ostracism.
Using the group mean as unit of analysis, group cohesion (r = .46, p < .001), group attraction (r = .40, p < .001), collective engagement (r = .33, p = .001), and task interdependence (r = .33, p = .001) positively correlated with emotional tone of discussion, while ostracism (r = −.38, p < .001) negatively correlated with emotional tone.
Group consensus
Participants were asked whether their group reached consensus. Using group-level response, an analysis of covariance found intimacy groups (M = 1.11, SD = 0.29) reported reaching consensus more often than information groups (M = 1.41, SD = 0.77), F(1, 94) = 6.07, η2 = .06, p = .016. This supports Hypothesis 3A. In addition, participants rated how confident their group was in the decision. Intimacy groups (M = 6.13, SD = 0.71) were more confident than information groups (M = 5.56, SD = 1.26), F(1, 94) = 6.84, η2 = .07, p = .01.
Using Preacher and Hayes (2004) bootstrapping method with process macro for SPSS, we tested the mediation of ostracism for the effect of condition on reaching consensus. Groups with higher levels of ostracism were less likely to reach consensus (r = .41, p < .0001), and the confidence interval testing the mediation did not contain zero (LL CI = .001; UL CI = .075), Z = 2.04, p = .04. Thus, ostracism mediated the effect of condition on reaching consensus. Participants in the intimacy condition felt less ostracized in their group and this increased their ability to reach consensus. This supports Hypothesis 3B.
In addition, tone of language was related to reaching consensus (r = −.46, p < .0001); as the tone of discussion was more positive, groups were better able to reach consensus. We tested if tone mediated the effect of condition on reaching consensus. The confidence interval did not contain zero (LL CI = .061; UL CI = .331), Z = 2.84, p = .005. Therefore, tone of language mediated the effect of condition on the ability to successfully reach consensus, supporting Hypothesis 3C.
Opinion change
Finally, we examined how much participants’ opinions shifted toward the view espoused in the article that climate change is real. We conducted a repeated measures ANCOVA with condition as a between-subjects’ variable and opinion on whether climate change is really happening (before/after group discussion) as a within-subjects’ variable with group size as a covariate. The interaction between condition and opinion (before and after) was significant, F(1, 324) = 4.76, η2 = .014, p = .03. To understand the interaction, we tested participants’ change in opinion before and after group discussion in each condition separately in paired-samples t-tests. In the intimacy condition, change from pre-group (M = 6.32, SD = 1.33) to post-group opinion (M = 6.67, SD = 0.76) was significant, t(161) = −3.23, p < .001. In the information condition, the change from pre-group opinion (M = 6.55, SD = 0.82) to post-group opinion (M = 6.61, SD = 0.93) was not significant, t(164) = −0.82, p = .42. This supports Hypothesis 4.
In post hoc analysis, we examined if change in opinion in the intimacy condition was driven more by participants who were skeptical of climate change. We ran a regression with initial opinion of whether climate change is happening and condition predicting change in opinion after group discussion. The regression with an interaction term between initial opinion and condition was significant, r = .97, R2 = .94, SE = 0.67, F(3, 10) = 38.86, p < .001. There was a significant interaction between condition and initial opinion, unstandardized β = −2.20, SE = 0.52, standardized β = −1.54, t = −4.27, p = .004. To understand this interaction, we examined the relationship between initial opinion and change in opinion for the information condition, r = −.45, p < .001, and the intimacy condition, r = −.84, p < .001. While in both conditions, participants whose initial opinion was closer to the climate denial side of the scale (1) were more likely to change their opinion toward the majority, this trend was stronger in the intimacy condition. For participants who rated their initial opinion as five or above, their change in opinion toward the end point that climate change is really happening (7) was small in both the intimacy (M = 0.07, SD = 0.69) and information condition (M = 0.03, SD = 0.83). For participants who rated their initial opinion below five and were closer to the climate change denial side of the scale, their change in opinion toward the majority opinion was larger in the intimacy (M = 5.00, SD = 1.58) than information condition (M = 2.50, SD = 4.95).
Next, we examined how the group decision differed from the pre-group individual decision between conditions. Analysis is at the group, rather than individual, level and the pre-discussion opinion is the mean of individuals in the group. We conducted a repeated measures ANCOVA with condition as a between-subjects’ variable and opinion on whether climate change is really happening (mean of before/after group discussion) as a within-subjects’ variable with group size as a covariate. The interaction between condition and opinion (before/after) was significant, F(1, 92) = 7.01, η2 = .071, p = .01. To understand the interaction, we tested change in opinion for before and group opinion in each condition separately in paired-samples t-test. In the intimacy condition, change from mean of individual pre-group opinion (M = 6.33, SD = 0.80) to group opinion (M = 6.81, SD = 0.42) was significant, t(46) = −4.04, p < .001. In the information condition, change from mean of pre-group opinion (M = 6.55, SD = 0.41) to group opinion (M = 6.59, SD = 0.76) was not significant, t(47) = −0.41, p = .68. This supports Hypothesis 4.
5. Discussion
Increased knowledge-sharing is not always associated with shrinking gaps between technical and public understanding of science (McFadden, 2016) or a catalyst toward action and consensus with climate change (Cagle and Tillery, 2015; Nolan, 2010). Hart and Nisbet (2012) noted that information can “amplify partisan differences” and increase resistance to new information and to climate mitigation policies (p. 702). A focus on fostering similarities and forming bonds and a group identity among people appear to be a fruitful avenue to increase consensus and persuasion and opportunities for positive discussion. According to the SIMPEA (Fritsche et al., 2018: 260), a social identity can shift “the focus from intergroup differences to group unity and common goals.” On scientific topics as polarized and complex as climate change, it is imperative to foster welcoming and intimate environments for discussion and consensus-building. While polarization and highlighting differences can be effective in some climate contexts (Schwarze, 2006), this study encourages climate communicators to acknowledge shared ground and to foster a common identity among audiences to encourage collaboration and decision-making.
By engaging in a conversation focused on affirming values and identity, participants in the intimacy condition were encouraged to develop a group identity. By identifying and feeling cohesion with their group members, they are more likely be open-minded to new information and viewpoints (Cohen et al., 2000) and may develop more respect for each other (Renger and Reese, 2017). Participants in the intimacy condition were more likely to report reaching consensus within their group and to change their opinion more in the direction espoused by the article that climate change is real, especially participants who initially rated their opinion toward the climate denial side of the scale. Instead of remaining in a state of gridlock that focuses on the facts of climate change, this study provides evidence that community building can be a strategy for moving conversations to policy and action. Lynda Walsh (2017) referred to this process of moving from fact and definition to values and action as “stasis shifting,” and other scholars have argued for its utility in climate communication (e.g. Bloomfield, 2019; Ceccarelli, 2011). Building group intimacy can thus be a way to encourage stasis shifting within climate communication.
In support of research on group cohesion and psychological safety, participants in the intimacy condition had higher ratings of social cohesion, task interdependence, collective engagement, and group attraction than participants in the information condition. Starting the conversation with self-disclosure may have created a group norm that encouraged open communication and psychological safety (Duhigg, 2016). In addition, participants had lower ostracism in the intimacy than information condition, and this mediated the effect of condition on their ability to reach consensus. The amount of positive emotion communication in the group also mediated the effect of condition on reaching consensus. Our measure of language tone positively correlated with social cohesion, task interdependence, collective engagement, and group attraction, and negatively correlated with ostracism. Thus, we were able to examine how the intimacy manipulation affected group processes and communication, which in turn facilitated the ability to reach consensus.
This study has important implications for the public understanding of scientific topics, such as the environment and climate change. Communication of information often is not enough to mobilize people to act. Instead, this study adds to existing research that offers that conversations between trusted and identified individuals may better motivate consensus and action on environmental topics. Goldberg et al. (2019) argued that discussing climate change with friends and family creates a “social feedback loop,” where those conversations spur both deeper understanding and a likelihood to have further discussions. Furthermore, interpersonal and group discussions center the importance of the people in the conversation: “when the message comes from close friends and family, people less engaged with the issue may be more receptive than when an identical message is communicated by someone not part of their close social network” (Goldberg et al., 2019: 14805). This study provides evidence that supports the claim that intimacy is an important factor in climate communication and additionally shows that intimacy does not have to only be defined by pre-existing social networks. Fostering a sense of intimacy can occur with relative strangers through an intimacy-building activity, or by acknowledging common, universal values, which itself leads to similar consensus-reaching effects. To mobilize people, there needs to be “some type and level of social engagement and action” (Moser, 2010: 38). Considering that solving climate change will require sustained collaboration, strategies that help to reduce ostracism, increase positive emotions and facilitate consensus that will be integral to future climate advocacy.
Limitations
There are several limitations to this study that point to avenues for future research. This study measured opinions right after discussion, and therefore, long-term effects on climate engagement are unknown. Ideally, attitudes could be measured 1–2 months after the manipulation or the manipulation could occur multiple times to show long-term effects. Second, the study used undergraduate students who often have a lot in common and who lean in a politically liberal direction. Although they are not monolithic in their beliefs and behaviors, undergraduate students have a large reservoir of potential commonalities to draw from, meaning that these results may not be generalizable. In addition, undergraduates may be more comfortable with assignment to ad hoc groups, so whether such a manipulation could work in the general population should be tested. Additional studies can test the intimacy-building activity and whether the fostering of intimacy in more varied group compositions leads to the same result. Third, many participants were already near the end point of the opinion scale before the discussion, so there may have been ceiling effects in how much they could change their opinion. Fourth, group discussions were online, so whether effects would generalize beyond an online chat is unknown. The hyperpersonal model (Walther, 1996) has theorized that online communication can improve interpersonal bonds due to lack of nonverbal cues and a focus on the message and self-presentation, and Jiang, Bazarova, and Hancock (2011) found that intimacy in online interactions can be stronger than face-to-face interaction. Thus, future research could replicate the study in an in-person environment. We further encourage replications of the study and similar designs to test the knowledge deficit model and particularly whether the manipulation of the information groups’ discussion would complicate or refine this study’s questioning of the knowledge deficit model. Finally, the effect size supporting Hypothesis 4 was small and should be replicated.
6. Conclusion
To address climate change, we will need large-scale collaboration and action. While information is the bedrock of climate change communication, evidence repeatedly shows that sustained action and consensus are unlikely to be reached with a sole focus on facts. Rather, people need to focus on collective identities, similarities, and intimacy to create consensus, cohesion, and engagement. When people do not feel ostracized in a conversation and, instead, are communicating positively and feel engagement and social cohesion, there is increased opportunity for consensus-building and progress. These results are particularly promising because the manipulation was relatively short, only 8 minutes, providing evidence that fostering intimacy and identification in group collaborations can be simple but still effective, even on complex and ideologically linked topics, such as climate change. In our current time, where the human impact on the environment is uncontested scientifically, there is still much work to be done by environmental communicators to bridge the gaps between technical and public audiences and mobilize actors for environmental progress.
Footnotes
Acknowledgements
The authors thank Pete Sengstock for technical support and also thank Mary Lu and Wei Xiao for help running the experiment.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was sponsored by the Wisconsin Alumni Research Foundation.
