The climate change consensus extends beyond climate scientists

The existence of anthropogenic climate change remains a public controversy despite the consensus among climate scientists. The controversy may be fed by the existence of scientists from other disciplines publicly casting doubt on the validity of climate science. The extent to which non-climate scientists are skeptical of climate science has not been studied via direct survey. Here we report on a survey of biophysical scientists across disciplines at universities in the Big 10 Conference. Most respondents (93.6%) believe that mean temperatures have risen and most (91.9%) believe in an anthropogenic contribution to rising temperatures. Respondents strongly believe that climate science is credible (mean credibility score 6.67/7). Those who disagree about climate change disagree over basic facts (e.g., the effects of CO2 on climate) and have different cultural and political values. These results suggest that scientists who are climate change skeptics are outliers and that the majority of scientists surveyed believe in anthropogenic climate change and that climate science is credible and mature.


Introduction
One vexing and ongoing concern about climate change is the discrepancy between the public view about climate change and the view of climate scientists. Approximately 97% of active, publishing climate scientists believe 4 in anthropogenic climate change (Doran andZimmerman 2009, Anderegg et al 2010) and about 97% of papers published about climate change with an explicit position on the existence of climate change affirm that it is occurring (Cook et al 2013). However, only approximately half of the American public believes in anthropogenic climate change (Leiserowitz et al 2010, Weber andStern 2011). There are many reasons for public doubt in climate change, including cognitive and affective factors (Gifford 2011, McCright andDunlap 2011) and media coverage that gives undue weight to the climate change controversy (Boykoff andBoykoff 2004, Boykoff 2007) or is influenced by political ideology (Dotson et al 2012). Scientists who are publicly skeptical about climate science and anthropogenic climate change may also contribute to public disbelief in climate change. These scientists often specialize in disciplines other than climate science and typically are not currently researching climate issues (Jacques et al 2008, Lahsen 2008, Anderegg et al 2010. Concern has been expressed Dunlap 2003, Jacques et al 2008) that these scientific skeptics have added false scientific credibility to skepticism. But is it false? Are non-climate scientist skeptics representative of the broader scientific community? What influences scientists' belief in climate change? Though many scientific organizations have produced climate change position statements affirming the existence of anthropogenic climate change (Oreskes 2004), the drafting of these statements is at times controversial and is often done without directly polling membership (Stenhouse et al 2014). This question is especially significant in the United States, which is the second-largest global emitter of carbon dioxide (Boden et al 2013). Here, we report on a survey of biophysical scientists at universities in the Big 10 Conference, a group of large, research-oriented universities in the United States (see Methods section for more details on the sample). The results show that scientists across disciplines nearly unanimously believe in anthropogenic climate change, are highly certain that climate change is happening, and find climate science to be trustworthy and credible.
1.1. Values, knowledge, and climate change beliefs Disagreement about climate change is rarely a simple dispute about facts. Indeed, people's interpretation of climate change information appears to be influenced by cognitive factors and motivated reasoning (Kellstedt et al 2008, Gifford 2011, Hart and Nisbet 2012, Kahan et al 2012, Carlton and Jacobson 2013, Carlton and Jacobson 2015. Notably, perceptions about risks such as climate change have been shown to be affected by where people fit on two cultural values scales: individualism versus communitarianism and hierarchicalism versus egalitarianism (Kahan et al 2008, Kahan et al 2012. People with more individualistic than communitarian values tend to believe that individuals (as opposed to communities) are responsible for their own well being without help or interference from society. People who hold more hierarchical than egalitarian values tend to believe that authority comes from strict social order based on characteristics such as race, gender, and class. These cultural values influence how people interpret information about risks. According to the cultural cognition hypothesis, hierarchical individualists are more skeptical of environmental risks, including climate change, because accepting these risks would undermine hierarchical individualists' belief in commerce and industry. Those who hold more egalitarian and communitarian values tend to perceive environmental risks more acutely because they feel that commerce and industry (the drivers of many environmental risks) promote individuals over the community (Kahan et al 2011, Kahan et al 2012. Cultural cognition appears to influence climate change beliefs among everyone, not just those who are uninformed about climate change or who have insufficient reasoning skills to properly evaluate the evidence for anthropogenic climate change. In fact, the effects of cultural cognition may actually increase with higher levels of science literacy, greater technical reasoning skills, and higher numeracy (Kahan et al 2012). Additionally, climate change beliefs may be influenced by sources of climate change information (Trumbo 1996, Boykoff and Boykoff 2004, Antilla 2005, Boykoff 2007) and mediated by trust in scientists (Hmielowski et al 2014).
Are scientists different? One might expect all scientists to agree with the consensus on climate change.
However, evidence suggests that cognitive factors may influence scientists' climate change beliefs, as well. A qualitative study of three physicists who were prominent climate change skeptics suggested that their beliefs were influenced about the role of science in society and the elite nature of physics compared to other sciences (Lahsen 2008). However, the role of cognition and knowledge in scientists' climate change beliefs is unexplored. In this study, we examine scientists' beliefs about climate change and climate science to determine (1) whether or not scientists agree with the climate science consensus about climate change and (2) whether cognitive factors and trust in climate science influence scientists' beliefs.

Methods
We surveyed the biophysical science faculty of the Big Ten universities in the US to ascertain (1) their beliefs about climate change, (2) their beliefs about climate science, (3) where they get their scientific information, and (4) their cultural and political values. Questions to evaluate climate change perceptions were adapted from earlier climate change surveys (Zimmerman 2008, Doran and Zimmerman 2009, Leiserowitz et al 2013, Prokopy et al 2013. Cultural values questions were adapted from those used in a study on the cultural cognition of scientific consensus (Kahan et al 2011). The specific question text can be found in the appendix.
The questionnaire was pilot-tested with a diverse sample of over 200 biophysical scientists from universities not included in the actual study sample. Because of concerns about the potential negative impact of the cultural values questions on the rate of survey completion, we randomly distributed two versions of the questionnaire to equal portions of the final sample. The first version included questions regarding field of study, climate change perceptions, beliefs about climate science, cultural values, political identification, and other demographics. The second version was identical with the exception that it omitted the cultural values questions (i.e., all items included in Q27 and Q28 in the appendix). physical sciences, earth sciences, agriculture, environmental sciences, natural resources, and other geosciences were selected for the study. Colleges based around engineering, architecture, liberal arts, technology, policy, law, business, education, fine and performing arts, health sciences, and animal sciences were excluded. Within each college, departments were identified that fell under the categories of biology, chemistry, physical sciences, environmental sciences, or geosciences. These departments included forestry and natural resources, fish and wildlife sciences, soil sciences, plant sciences, crop sciences, horticulture, atmospheric sciences, meteorology, geography, geology, entomology, biology, chemistry, physics, and astronomy. Climate scientists were not excluded from the sample. Some engineers who were not housed in engineering departments were in the final sample. Contact information for faculty members was located on departmental webpages. In order to standardize across universities, tenured, tenure track, visiting, and emeritus faculty members were included in the sampling frame. Emeritus faculty were included in the sample because prior work on scientists who were climate change skeptics tended to be from older generations (Lahsen 2008). Research and adjunct faculty were excluded from data collection because their listing on websites was inconsistent. The name, university, department, email, and phone number of each faculty member were recorded. Because email addresses were required to administer the questionnaire, faculty members without a listed email were not included. The final selection frame included 4816 names.
To create the sample, 2000 names were randomly selected from the list of scientists. An equal number of recipients were then randomly assigned to either group A, who received the questionnaire that included cultural values questions, or group B, who received the questionnaire excluding the cultural values questions. The survey was administered online using Qualtrics Survey Software. Links to the questionnaire were distributed through email in February and March of 2014. Based on survey administration best practices (Dillman et al 2008), up to three contacts (the initial email letter, a reminder email, and a final reminder) were made with recipients. Emails that bounced were removed from the sample and were not replaced with new addresses because of the relatively large sample size.

Data analysis
Differences among groups were determined using a t-test and were considered significant at a=0.05. Following Kahan et al (2011), median splits were used to group respondents by cultural values. Logistic and multiple regressions were used to model climate change belief and certainty, respectively, using trust in climate science, proportion of climate change information coming from scientific literature, political orientation (1='Very conservative', 5='Very Liberal') as predictors and age, gender, amount of the respondent's research that concerns climate change, and number of courses taken as graduate or undergraduate in the following fields: chemistry, physics, earth/ocean/atmospheric sciences, math, biology, and engineering.
Data were analyzed using Stata version 12.1. Dot plots were created using Stata and the bean plots were created using the beanplot package in R version 3.0.

Results and discussion
After excluding invalid addresses (i.e., emails that bounced back), we surveyed a sample of 1868 scientists and received 698 responses (37.4% response rate). This response rate is slightly better than the prior work on climate scientists and climate change (30.7%, Doran and Zimmerman 2009). There were no significant differences in response rate between the survey version with the cultural values questions and the survey version without the cultural values questions. The results suggest a broad consensus that climate change is occurring: when asked 'When compared with pre-1800's levels, do you think that mean global temperatures have generally risen, fallen, or remained relatively constant?', 93.6% of respondents across all disciplines indicated that they thought temperatures have risen, 2.1% thought temperatures had remained relatively constant, 0.6% thought temperatures had fallen, and 3.7% indicated they had no opinion or did not know. Belief in climate change was relatively consistent across disciplines (range: 91.2%-100%, figure 1).
Most respondents believed that humans are contributing to the rise in temperatures. Of those who indicated that they believed temperatures have risen, 98.2% indicated they believe that 'human activity is a significant contributing factor in changing mean global temperatures'. Together, these two facts reveal that 91.9% of scientists surveyed believed in anthropogenic climate change. This number is slightly lower than the 96.2% of actively publishing climate scientists that believe that mean temperatures have risen and the 97.4% who believe that humans have a role in changing mean global temperatures (Doran and Zimmerman 2009).
Those who said that temperatures have risen were significantly more certain in their beliefs than those who did not (3.41 versus 2.40 on a 4-point scale, t=−5.08, p<0.001). Scientists who believed in anthropogenic contribution to temperature rises and scientists who did not believe in an anthropogenic contribution were similarly certain in their beliefs (3.40 versus 3.3, t=−0.57, p=0.57). Respondents who disagreed about whether temperatures had risen tended to have different beliefs about the relationship between greenhouse gases, solar activity, climate change, and climate change models, as well (figure 2), indicating that a disagreement about the 'facts' of climate change was at least part of the difference between those who did and did not believe in climate change.
The cognitive and demographic data are presented in figure 3. Those who believed that mean temperatures had risen had significantly higher levels of trust in climate science, were significantly more egalitarian, more communitarian, and more liberal than those who did not believe temperatures had risen. The respondents were divided in cultural values: 35.1% hierarchical individualist, 31.3% egalitarian communitarians, and 33.6% were either egalitarian-individualist or hierarchical-communitarian. A significantly smaller proportion of hierarchical individualists (e.g., those who scored above the median on both the hierarchical and individualism scales, Kahan et al 2011) believed in climate change than non-hierarchical individualists (94.3% versus 98.8%; t=2.11, p=0.02). The difference is greater when comparing egalitarian communitarians to hierarchical individualists (leaving out those in the middle): 100% of egalitarian communitarians believed in climate change compared to 94.3% of hierarchical individualists (t=2.21, p=0.01). Cultural values were also significantly associated with certainty that climate change is occurring: hierarchical individualists were significantly less certain when compared to the rest of the sample (mean certainty of 3.14 versus 3.63, t=5.43, p<0.001) and when compared to egalitarian communitarians (3.14 versus 3.65, t=4.69, p<0.001).
Cultural values were associated with belief in human contribution to climate change. Significantly fewer hierarchical individualists believed in climate change than others (90.3% versus 99.4%, t=3.72, p<0.001). Again, the difference was stronger at the cultural value extremes: 100% of egalitarian communitarians believed in a human cause compared to the 90.3% of hierarchical individualists (t=2.94, p=0.002). Compared to other respondents, hierarchical individualists who believed in anthropogenic climate change were significantly less certain of a human contribution to climate change (mean certainty 3.12 for hierarchical individualists versus 3.54 for others, t=4.46, p<0.001). As expected, this difference was even greater when comparing hierarchical individualists to egalitarian communitarians (3.12 versus 3.64, t=4.69, p<0.001). In all, these results affirm prior findings that cultural values are a significant determinant of climate change beliefs (Kahan et al 2011), though the effects may be smaller among scientists than the general public. Respondents generally found climate science to be credible. The average response to 'Climate science is a credible science' was 6.67 out of 7, indicating strong agreement. The average response to 'Compared to my field, climate science is a mature science' was 4.78 out of 7, indicating slight agreement. Respondents also rated the trustworthiness of climate science compared to their field, from 1 ('Much less trustworthy') to 5 ('Much more trustworthy') with a middle point of 3 ('About equally trustworthy'). The average response was 2.69, indicating that respondents thought climate science was slightly less trustworthy than their field. The data, grouped by respondents' disciplines, are presented in figure 4.
As with the general public, the mass media's tendency to give undue weight to climate skepticism (Boykoff andBoykoff 2004, Boykoff 2007) appears to have influenced scientists. Though amount of climate change information received from scientific literature compared to mass media was not significantly correlated with belief in climate change or belief in human contribution, there was a slight, significant correlation between proportion of climate change information received from scientific literature and certainty that climate change was occurring (r=0.15, p<0.001) and a moderate, significant correlation between proportion of climate change information from scientific literature and certainty of a human contribution (r=0.23, p<0.001). In other words, those who received more climate change information from mass media were less certain of the existence of and human contribution to climate change.
The results of the regression models for climate change beliefs and certainty are presented in tables 1 and 2. The only predictor that was significant in all models was trust in climate science: respondents who trusted climate science more were more likely to believe in climate change, were more likely to believe in a human contribution to climate change, and were more certain in these beliefs. These results suggest that, when it comes to climate change, scientists are people, too: prior research shows that trust in scientists mediates climate change perceptions (Hmielowski et al 2014). Political values and sources of media influence climate change beliefs in several of the models, though the effect is attenuated compared to the general public (McCright and Dunlap 2011).

Conclusions
Though public awareness of the scientific consensus on climate change may be insufficient to spur largescale adaptive or mitigative measures (Kellstedt et al 2008), prior work has suggested that it may be necessary. For example, public support for climate policies is affected by incorrect perceptions that the existence of anthropogenic climate change is . Big 10 scientists' perceptions of climate science as measured by overall credibility of climate science, the maturity of climate science relative to respondents' field, and the trustworthiness of climate science relative to respondents' field. Relative trustworthiness was re-scaled from 5-points to 7-points for this figure.  (1) climate change is occurring, (2) humans are contributing to it, and (3) climate science is a trustworthy, mature, and credible discipline. Scientists who continue to claim otherwise are operating outside of the consensus, not just of climate scientists, but also of scientists as a whole. However, the fact that cultural values and political ideology appeared to influence the scientists' beliefs underscores the difficulty of climate change as a public issue. There is a temptation to think of those who don't believe in climate change as uninformed or irrational. However, studies are increasingly showing that knowledge and rationality are just one piece of the complicated climate puzzle. Values and identity matter, among the general public (e.g., Kahan et al 2011) and, as this research shows, among putatively rational scientists. It is becoming increasingly apparent that effective climate change outreach, communication, and policy must account not just for the 'facts' of climate change, but for the 'feel' of it, as well.

Acknowledgments
This project was funded by the Purdue Climate Change Research Center and the Purdue Department of Forestry and Natural Resources. The authors would like to thank J Reiman and Z Modeliste for assistance with the project.

Appendix. Survey questions and results
Q1 Which of the following best describes your primary field of study:  Q30 Thinking about the environmental movement, do you think of yourself as: 1. An active participant in the environmental movement 21.55% 2.
Unsympathetic towards the environmental movement 1.58%