In the United States, climate change is a highly politicized issue; beliefs about the proper response to—and even the reality of—human-caused climate change fall largely along partisan lines (Guber, 2013). In this highly politicized environment, motivated reasoning, the tendency to see the world through one’s own lens of preconceptions and affiliations (Chen et al., 1999; Klaczynski, 2000), can significantly impact learning (Nisbet et al., 2013; Somerville, 2011). Our research asks whether motivated reasoning around this politicized topic changes after an introductory-level nonmajor undergraduate course.

A significant body of research shows that perceptions and interpretations of the world are shaped by psychological needs and goals (e.g., Jost et al., 2009; Jost, 2017), social identities (e.g., Bolsen et al., 2014; Fielding & Hornsey, 2016; Hornsey et al., 2016; Klaczynski, 2000), and the need to resolve dissonant ideas (e.g., Feygina et al., 2010; MacCoun, 1998). The need to conform to the beliefs of one’s identity group is particularly powerful for college-age students (Walker et al., 2017). This leads to a subconscious bias toward information that conforms to preconceptions and skepticism of information that contradicts existing beliefs (Taber and Lodge, 2006). Motivated reasoning is particularly persistent for highly politicized topics such as climate change (Lewandowsky et al., 2013; Walker et al. 2017). However, semester-long college courses that contain scientific content can increase students’ belief in the reality of climate change and willingness to take some action to address it (Maier et al., 2018; Walker et al., 2017). Such classes may change student perceptions by exposing them to persuasive data and arguments, but are only successful with students who show characteristics of open-mindedness and well-developed critical-thinking skills when they enter the course (Sinatra et al., 2012). These courses are less effective for students whose identity group advocates climate change denial (Walker et al., 2017). Additionally, motivated reasoning may lead learners to focus on irrelevant facts or misconceptions that justify pre-existing beliefs (Li et al., 2011).

College-level science classes aim to teach students to evaluate data and make evidence-based interpretations of the natural world, not simply to memorize facts or accept a certain point of view. Some work hints that such an approach may relax motivated reasoning (Lombardi et al., 2013), but the idea has not been directly tested. The science education literature explores when and how college classes can change students’ attitudes about scientific topics (e.g., Campisi & Finn, 2011; Ding & Mollohan, 2015; Smith & Williams, 2013). A focus on motivated reasoning looks for factors underlying those attitudes (Walker et al., 2017); we consider the motivating context of attitudes rather than the attitudes themselves. To the extent that attitudes are based in knowledge, changes in knowledge about climate should precipitate changes in attitudes. In contrast, if motivated reasoning reigns, attitudes may be rooted instead in psychological predispositions. Moreover, motivated reasoning can influence what an individual knows because it leads people to selectively learn and retain information that supports preexisting beliefs and dismiss evidence that contradicts them. We ask: Does an introductory science class affect whether and how much students engage in motivated reasoning about climate change?

A quasi-experimental design compared undergraduate nonmajor science classes, one which focused on climate change, with others that made only passing reference to the topic. All classes were taught over a 14-week semester and were of similar enrollment (40–50 students per class). All classes were taught using active-learning methods (Freeman et al., 2014; Michael & Modell, 2003) that combined written reading response, lecture, and in-class, hands-on activities that engaged students in gathering and interpreting real-world data. Students were awarded extra credit for their voluntary participation in the research project.

Students in the climate change group were enrolled in Earth System Science at a small liberal arts college in the northeastern United States during fall 2016, spring 2017, or fall 2017. One geoscience professor taught the classes for two semesters, and a different geoscience professor taught the third semester. The course introduces the “spheres” of the Earth system: the geosphere (the rocky part of the Earth), the hydrosphere (the mass of liquid water across the Earth), the atmosphere, the biosphere, and finally, the cryosphere (glaciers and sea ice). Students explore each sphere’s individual function, interconnections, and how all these spheres interact to produce climate over geological time. The course concludes with a two-week discussion of modern climate that focuses on data and modeling to demonstrate climate change and its causes.

The comparison group comprised two nonmajors’ science courses that do not focus specifically on climate change. Earth and Life Through Time (fall 2016), taught by a geoscience professor, examines the major events in the history of life on Earth, including the formation of the planet, origin of life, transformations of the atmosphere by photosynthesis, evolution of major groups, extinctions, and the role of contingency in shaping Earth’s current biota. Natural climate change is mentioned at two points, but is not a major theme of the course. Astrobiology (fall 2017) also served as a comparison. This course is co-taught by a geoscience and a physics professor. Astrobiology investigates the conditions necessary for life and how they are explored in our solar system and beyond. Climate and climate change are not discussed in this course.

Students generally enrolled in these classes to meet a general curriculum requirement. Most students selected courses based on offering time rather than specific interest in the topic. Students were invited to participate in this research project during the first week of their class. Two hundred and twenty eight students opted to participate: 134 from the climate change classes and 94 from the comparison classes. See Table 1 for demographic information.

Attitude about climate change. Attitude is defined as how concerned one is about climate change and how much one believes that it is a problem. We used two validated scales to measure attitude: the Plausibility Perceptions Measure (PPM) (Lombardi & Sinatra, 2012) and the New Ecological Paradigm (NEP) (Dunlap et al., 2000). These measures indicate the degree to which people believe that human-caused climate change is happening and how concerned they are about it.

Knowledge about climate change. In contrast to attitude, we defined knowledge as objective facts, such as the identity of carbon dioxide as a greenhouse gas and the role of the ocean in distributing heat around the globe. The knowledge survey also evaluated critical reasoning skills that are part of the process of climate science (e.g., interpreting data visualizations).To assess participants’ understanding of current scientific consensus about climate and climate change, the authors wrote a test bank of questions, based on the information covered in the syllabus. Each participant answered a different set of nine of these questions at each of the three test points in the semester.

General scientific knowledge. For comparison, we also selected a test bank of general science questions adapted from national surveys measuring public knowledge of science, including those from Pew Research (Pew Research, 2015), a U.S.–UK Civic Scientific Literacy project (Miller, 1998), and the Washburn Education Test of Scientific Literacy Skills (Gormally, et al. 2012). These questions were comparable in difficulty to the climate change knowledge items and also included science process questions (e.g., questions about experimental design). All general science items focused on nonpoliticized topics (e.g., we avoided questions about evolution or the age of Earth). Each participant received nine questions from this test bank with each of the three surveys.

Motivated reasoning. We selected three psychological predispositions that previous research has shown bias reasoning: (1) Need for Closure (NFC), (2) Belief in a Just World (BJW), and (3) authoritarianism. A 15-item NFC scale (Roets & Van Hiel, 2011) measures preferences for definitive, final decisions versus tolerance for ambiguity. A six-item BJW scale (Rubin and Peplau, 1975) measures the belief that things ought to happen in a fair manner versus the acknowledgement that they often do not. A four-item authoritarianism scale (Stenner, 2005) measures preferences for society to be organized in a way that follows traditional authority structures versus questioning such authorities. Social psychologists have studied these variables extensively and found that they all serve as motivations in reasoning (e.g., Jost, 2017; Jost et al., 2009). People have stable levels of each predisposition, and these predispositions filter how individuals process information and construct attitudes. For each of these variables, scores were calculated by taking the mean of the items on the scale.

We surveyed participants three times during the semester: the first week of the class, after the midpoint of the class, and just before the final exam. All three surveys included measures of knowledge and attitudes about climate change, as well as general scientific knowledge. The first survey also included measures of psychological predispositions (NFC, authoritarianism, and BJW) and demographic items. Because prior research (e.g., Caspi et al., 2005) shows that such psychological constructs are stable over a few months, a single survey point will be representative for the semester. Students who completed all surveys received extra credit toward their class grade.

To test for the presence of motivated reasoning and whether it was affected by the courses, we compared bivariate correlations between each of the three psychological predispositions (NFC, BJW, and authoritarianism) and each of the four knowledge/attitude variables (NEP, PPM, knowledge about climate change, and general scientific knowledge) at both the beginning and end of the semester. (The mid-semester survey was included to explore hypotheses not discussed in this paper. However, the pattern of correlations among these variables at mid-semester was similar to that at the end of the semester.) If motivated reasoning was occurring, we expect to see significant correlations between psychological factors and the information and attitudes individuals hold. In contrast, if correlations weaken over time, the data indicate that motivated reasoning has diminished.

At the beginning of the semester, we observed several significant correlations between psychological variables and both knowledge about and attitudes toward climate change (Table 2). Such significant correlations suggest motivated reasoning was present. In contrast, there was only one significant correlation between psychological variables and general scientific literacy, which suggests that motivated reasoning is more common for the politicized topic—climate change—and less so for general science (Table 2). These correlations were negative, suggesting that people who were more motivated by authoritarianism and BJW scored lower on the knowledge measures and were less concerned about climate change as revealed by attitude measures. (The positive correlation between NFC and PPM was the only exception to this pattern—it suggests that people more motivated by NFC believe human-caused climate change is more plausible.) When comparing across all courses, there were more significant correlations in the climate change classes than in the comparison classes (Tables 3 and 4), indicating that students entered the climate change class with more motivated reasoning.

At the end of the semester, there were no longer any significant correlations between psychological variables and either type of scientific knowledge. The remaining significant correlations were between attitudes about climate change and psychological variables; these remaining correlations were mostly reduced compared to values measured at the beginning of the semester (Tables 3 and 4). Motivated reasoning was reduced—but not entirely eliminated. The lingering role of motivated reasoning on attitudes mirrors prior work (Walker et al., 2017) that showed attitudes about climate change to be more stable than those of other politically charged scientific topics such as evolution and vaccination.

At the beginning of the semester, authoritarianism and BJW predict climate change knowledge across all of the students who participated (Table 2), but these same psychological factors do not predict climate change knowledge at the end of the courses. This finding suggests that knowledge was less related to psychological variables at the end of the semester than at the beginning—which is consistent with the hypothesis that reductions in motivated reasoning occur over the course of the semester. Psychological research suggests that motivated reasoning would be stronger for attitudes and beliefs about climate change, and weaker for factual knowledge (e.g., Bolsen et al., 2014; Taber & Lodge, 2006). Our data were consistent with this result. Some of the relationships between psychological preferences and attitudes changed between the beginning and end of the semester, but less so than the relationships between psychological preferences and scientific knowledge. This suggests that there may be different routes to changing information-based beliefs compared to attitudes about politicized scientific topics.

The relationships we observed were consistent in both the climate change and comparison classes. This suggests that direct instruction about climate change may not be the primary factor behind changes in motivated reasoning. We speculate that the active learning pedagogy shared by all classes may play a role. Instructors of all classes in this study gave students the opportunity to interpret data, consider what constitutes good evidence for a scientific claim, and practice such skills with peers. Future research should test these ideas in different styles of science classes, with different teaching strategies, to better identify which elements of a class most effectively reduce motivated reasoning.

Although motivated reasoning is a persistent feature of human psychology, we find evidence of reduced motivated reasoning about climate change after a one-semester college science class. This is promising information for promoting scientific literacy and evidence-based civic engagement.

Emily Fisher (fisher@hws.edu) is associate professor in the Department of Psychological Science and Nan Crystal Arens is a professor in the Department of Geosciences at Hobart and William Smith Colleges in Geneva, New York.