The existence of debate and disagreement between scientists is a fundamental and necessary part of science itself. However, this does not mean that all disagreement is necessarily productive. Excessively rancorous forms of conflict can negatively affect the ability of the scientific community to engage and deliberate effectively. For this reason, it is important to understand—and find ways to manage—undesirable forms of disagreement among scientists.

One specific disagreement offers the opportunity to help understand unproductive conflict within a scientific organization. Many members of the American Meteorological Society (AMS) feel that members’ differing views on global warming have created unproductive conflict within their society. The AMS Committee to Improve Climate Change Communication (CICCC) was created in 2011 to foster dialogue between members about this issue. To better understand members’ views on global warming and perceptions of climate-related conflict within AMS, in 2011 and 2012 the CICCC surveyed AMS members in conjunction with George Mason University. Results of this survey have been published in a report of initial findings (Maibach et al. 2012) and in this journal (Stenhouse et al. 2014). This article builds on the prior papers by analyzing responses to open-ended questions that allowed AMS members to describe the conflict in their own words. Our aim was to reveal what AMS members see as the most important features of the conflict, without restricting them to closed-ended survey questions.

While most scientists would agree that some types of conflict are unproductive, opinions differ on what precisely makes a disagreement “good” or “bad.” This conceptual difficulty parallels scholarly debate on what kind of disagreement is worthy of the term “conflict.” The preferred definition of conflict for many communication scholars is “the interaction of interdependent people who perceive the opposition of goals, aims, and or values, and who see the other party as potentially interfering with the realization of these goals (aims, or values)” (Putnam and Poole 1987, p. 552).

In investigating conflict within the AMS, our aim was not to determine whether conflict that meets any particular definition was taking place. Instead, our main goal was to understand the views of AMS members who said they perceived conflict, leaving them free to interpret the term for themselves. Therefore, for this paper, the term “conflict” refers to whatever each survey respondent had in mind when they gave their answers. Many (47% of all survey respondents) did not perceive conflict on global warming among AMS members; these respondents were only able to answer the last of the four open-ended survey questions discussed below.

METHODS.

A web-based survey was sent to all professional (i.e., nonstudent) AMS members in December 2011. The survey was sent to 7,062 valid e-mail addresses, and 1,854 people completed some portion of the survey, yielding a minimum response rate of 26.3% (assuming all nonrespondents were eligible to participate). This is a reasonably typical response rate for web-based surveys (Callegaro and DiSogra 2008; Shih and Fan 2008). The open-ended questions were as follows:

• Q1: Briefly, why do you think the conflict among AMS members over global warming is productive?

  [Asked if participants had previously reported seeing the conflict as “highly productive” (n = 104), “somewhat productive” (n = 246), or “both productive and unproductive” (n = 260).]

• Q2: Briefly, why do you think the conflict among AMS members over global warming is unproductive?

  [Asked if participants reported seeing the conflict as “highly unproductive” (n = 88), “somewhat unproductive” (n = 165), or “both productive and unproductive” (n = 260).]

• Q3: Why are you reluctant to bring up the topic of global warming [in AMS meetings or forums]?

  [Asked if participants reported being reluctant to bring up global warming “in AMS meetings or other AMS forums” (n = 282).]

• Q4: What message, if any, would you like to convey to AMS leadership about global warming-related conflicts within AMS?

  [Asked of all respondents who reached this point in the survey (n = 1,826).]

Participants typed their answers into text boxes on their web browser. Although these questions are referred to here as Q1, Q2, Q3, and Q4, they were actually presented to participants in the middle of the survey, which consisted of approximately 20–30 questions in total (number of questions varied, as some questions were not asked of all participants). For full details of questions, see Maibach et al. (2012).

Responses were analyzed using quantitative content analysis. We created a set of content categories, with clear and precise rules for what belonged in each category. As is standard practice in content analysis, the coding and category creation process was iterated several times before arriving at the final results described below (Krippendorff 2004). Two members of the research team, acting as coders, first read responses and created an initial set of coding categories. They stopped once they perceived that they reached the saturation point for unique categories. Then they attempted to code a subset of the responses. They subsequently compared the results of this coding, examined instances where they had disagreed, and discussed how the coding rules could be refined to reduce disagreement in the next round of coding. The coders then repeated the process with new, refined coding rules, until adequate intercoder reliability coefficients were achieved.

For example, the code “conflict improves research” was used for responses to Q1 that mentioned conflict improving the quality or quantity of research, improving scientists’ practices generally speaking, or increasing the quality or quantity of knowledge available. Responses that only mentioned conflict resulting in sharing or discussion of existing knowledge were explicitly excluded from this category. Coding categories were not mutually exclusive—each response could be coded as fitting into multiple categories. Ten percent of responses were coded by both coders to check for intercoder reliability. Only coding categories with Krippendorf’s alpha reliability coefficients of 0.667 or higher are reported below. This is a level of reliability deemed adequate for initial conclusions in areas of study that are little understood (Krippendorff 2004).

An earlier question in the survey asked participants about their views on the causation of global warming. Responses to this were used to identify four subgroups for analysis:

1. Members who said the global warming of the last 150 years was mostly caused by human activity (53% of full sample).

2. Members who said global warming is caused more or less equally by human activity and natural events (10% of full sample).

3. Members who said global warming is happening, but that there is insufficient evidence to determine the degree of human causation, and members who said global warming is happening, but they “don’t know” what the cause is (in total, 21% of the sample).

4. Members who said global warming is mostly caused by natural events, members who said they do not know whether global warming is happening, and members who said they do not believe that global warming is happening (16% of the sample).

Analysis of the data revealed only minor differences between groups 1 and 2, and between groups 3 and 4, regarding their views of the AMS climate conflict. We therefore collapsed groups 1 and 2 into a single group, referred to as “Convinced” members (owing to being convinced of substantial human causation of global warming). We collapsed groups 3 and 4 into another group, referred to as “Unconvinced” members.

Convinced members made up 62% of the full sample of 1,854 respondents; Unconvinced members made up the other 38%. However, only respondents who answered at least one open-ended question, as well as the questions about the causes of global warming, were included in our analysis. Nearly half of the survey participants met these criteria (47%; n = 879). Thus, the participants whose statements were used to create the coding categories constituted 12% of all 7,062 professional members of the AMS.

RESULTS.

To calculate percentages of Convinced and Unconvinced participants who gave each type of response, we used the number of participants who were asked that question as the denominator. This was done so that everyone who had an opportunity to respond to the question was included in the percentages. The choice not to write any answer is also a form of response. Therefore, it is important to include these responses when calculating percentages.

Additionally, for all questions, percentages do not add to 100%. As noted above, coding categories were not mutually exclusive, meaning that it was possible for responses to be coded as fitting in multiple categories. Also, we did not achieve sufficient intercoder reliability for several coding categories; percentages for these unreliable categories of response are not reported.

In total, 252 Convinced and 358 Unconvinced respondents were asked Q1, which assessed why they think the conflict has been productive. Members of the Convinced and Unconvinced groups shared largely similar views: both were roughly equally likely to say that conflict improves research (40% of Convinced, 43% of Unconvinced), that debate is good (36% of Convinced, 39% of Unconvinced), and that having a diversity of perspectives is helpful (29% of Convinced, 28% of Unconvinced).

A total of 309 Convinced and 204 Unconvinced respondents were asked Q2, which assessed why they think the conflict is unproductive. Members of the two groups had somewhat differing views. The most common view—held by a large minority of both groups (28% of Convinced, 31% of Unconvinced)—was that the positions of members on the opposing side of the conflict were based not on scientific evidence, but rather on other factors such as political biases.

Other views on why the conflict was unproductive were more likely to be voiced by one group or the other. Convinced participants were more likely to mention that the conflict confuses the public (30% of Convinced, 8% of Unconvinced) or that the conflict might lead to the public not taking action on climate change (9% of Convinced, 3% of Unconvinced). Unconvinced participants, on the other hand, were more likely to express the view that members on the opposing side of the conflict are being closed-minded (8% of Convinced, 18% of Unconvinced) or are suppressing unpopular views (2% of Convinced, 14% of Unconvinced).

A total of 65 Convinced and 217 Unconvinced respondents were asked Q3, which assessed why they were reluctant to bring up global warming in AMS meetings. There was some convergence in the two groups’ answers. The most common response for both groups was a wish to avoid conflict or heated discussion (46% of Convinced, 25% of Unconvinced). A smaller set of participants also mentioned concerns about the effects that bringing up global warming might have on their career; Unconvinced participants were more likely to report this (8% of Convinced, 15% of Unconvinced).

All participants who did not drop out of the survey (1,133 Convinced and 693 Unconvinced) were asked Q4, which assessed what messages, if any, they wish to convey to AMS leadership about the conflict. Although most respondents did not answer this question (73% of Convinced, 46% of Unconvinced), those who did conveyed a wide range of messages. Because of the much larger denominators, each category of answer was given by a smaller proportion of respondents.

Although small, a noteworthy set of respondents commented on whether the AMS should take an official organizational stance on global warming. Both groups were about equally likely to say that the AMS should take an official stance (2% of Convinced, 1% of Unconvinced), while Unconvinced participants were more likely to say the AMS should not take a stance (0.1% of Convinced, 4% of Unconvinced). Another specific action—suggested by a small minority of participants from both groups—was that the AMS should provide better forums for discussion (3% of Convinced, 3% of Unconvinced).

Finally, one topic appeared frequently in the answers to all four questions—the influence of politics. The influence of politics was overwhelmingly mentioned as a negative influence, regardless of participants’ views on global warming. Of all individuals who responded to at least one open-ended question (424 Convinced and 455 Unconvinced participants), 20% of the Convinced and 28% of the Unconvinced mentioned politics in a negative light. Only 6% and 3% of each group, respectively, mentioned politics in a neutral or positive light.

DISCUSSION.

Regarding the conflict about climate change, members of both groups within AMS—the “Convinced” and the “Unconvinced”—see different actions as appropriate, based on how they perceive the weight of scientific evidence. The Convinced see the weight of evidence leaning in favor of substantial human causation. Accordingly, many of the Convinced see it as unproductive to have AMS members in conflict over this evidence in a way that might send an inaccurate message to the public. On the other hand, because Unconvinced respondents see the science as not settled, many of them see actions to marginalize the influence of dissenting voices as unfair and antiscientific.

The idea that all sides see themselves as acting in accordance with the weight of scientific evidence is noteworthy because it may make any kind of compromise, “agree to disagree”–type settlement more difficult. Acknowledging the legitimacy of the other side may be seen as a tacit endorsement of their views on the weight of scientific evidence for human-caused global warming. For many AMS members with strong views in either direction, any implication that the other side’s views are intellectually respectable may be seen as itself promoting an inaccurate view of the weight of the evidence.

One way to increase mutual understanding in this kind of situation might be to remind everyone that those on the other side of the debate are acting appropriately, given how those others perceive the weight of the evidence. As in other conflicts, dialogue about differences can provide a structured framework whereby disagreement does not destroy relationships or increase polarization (Bohm 1996). AMS members who requested better discussion forums would seem to agree with this idea.

The division within AMS over what message to send to the public also highlights a dilemma that may be faced by other scientific organizations in the future. Whenever there is strong scientific consensus, sending a clear message about the state of the science necessarily implies that dissenting views are, at least to some extent, less legitimate. Scientific organizations should be prepared to decide whether they will prioritize sending a clear message to the public or legitimizing dissenting voices, while acknowledging that prioritizing one of these goals may impair their ability to pursue the other.

AMS members frequently mentioned politics as a negative influence on the debate. Some scholars have suggested that because political differences underlie the debate, they should be discussed directly, so that the real source of the conflict can be addressed. On the one hand, it is reasonable to expect that interventions focused on political discussion and shaped by research on conflict resolution (e.g., Schweizer et al. 2014) could lead to productive dialogue. However, our results show that there may be resistance to dialogue that directly addresses the role of politics. If this strategy were to be tried with AMS members, substantial effort would need to be devoted to first overcoming their impression that politics should be kept out of the debate altogether.

This analysis was novel in several important ways. Other studies have looked in depth at how scientists perceive norms of conflict and disagreement within science. However, this analysis is one of the first to systematically analyze a large number of scientists’ views on a conflict—expressed in the respondents’ own words. Moreover, the conflict over global warming is not just a controversy internal to science, but is widely recognized as having external political influences and implications for large-scale societal action. Understanding whether scientists see conflict over this kind of topic differently from how they see other scientific controversies is important.

In conclusion, this article has shown that many within AMS see the appropriate solution to the conflict as being largely defined by the state of the evidence itself. A problem for those hoping to reduce tensions is that acknowledging the legitimacy of both sides may be seen by many as sending an inaccurate message about where the weight of evidence lies. Scientific organizations should prepare for future conflicts (1) by recognizing that asking for compromise may be seen as going against the norms of science, (2) by being prepared to choose between clear messaging and legitimizing of dissenting voices, and (3) by being prepared for difficult work if they see the solution as a more open discussion of how views on science might be informed by political orientations.