Communicating Scientific Consensus: John Cook responds

Many thanks to Dan Kahan for the opportunity to discuss this important (and fascinating) issue of communicating the scientific consensus. I fully concur with Dan’s assertion that we need to be evidence-based in how we approach science communication. Indeed, my PhD research is focused on the very issue of attitude polarization and the psychology of consensus. The Cultural Cognition project, particularly the paperCultural Cognition of Scientific Consensus, has influenced my experiment design. I’m in the process of analysing data that I hope will guide us towards effective climate communication.

Dan initially posted 4 points although point 2 is essentially a lead-in to points 3 and 4. So I’ve divided my response into three sections – the scholarly value of Cook et al., the knowledge deficit issue and attitude polarization. 

1. What scholarly knowledge does Cook et al. contribute?

While the media coverage of Cook et al. focused on the simple message of an overwhelming consensus among climate papers expressing a position on anthropogenic global warming (AGW), there are a number of significant, nuanced results coming out of our research. For example:

  • The significantly larger sample size (11,944 abstracts over 21 years) allowed us to determine the evolution of the scientific consensus – which hadn’t been done in earlier analyses. We found that the consensus has continued to strengthen over time
  • We found that overwhelming consensus was already in place from 1991. This demonstrates that the scientific consensus was not manufactured in 1995 by the IPCC. Instead, the IPCC was formalising the pre-existing scientific consensus in a synthesised report.
  • We did address a number of criticisms of earlier works. One criticism of Oreskes 2004 was that the small sample size (although 928 is not that small) meant a number of rejection papers weren’t included in the sample. Oreskes’ response was that the analysis demonstrated that rejection of AGW held a negligible presence in the literature and while an increased sample size might include some rejection papers, the same conclusion should result. We confirmed this.
  • Oreskes 2007predicted that as a consensus strengthened, the number of papersrestating the consensus position should diminish. An elegant confirmation of this prediction that came out of our data is that as the consensus strengthened (e.g., the percentage of endorsements among position papers), we also observed an increase in the percentage of “no position” abstracts.
  • One of the strengths of this study is that it lays the foundation for a number of follow-up analyses. For example,Boaz Miller lays out three conditions determining when a consensus is knowledge based (social calibration, consilience of evidence, social diversity). The Cook et al. data is ideally suited to examine this. Citations are a commonly used metric for research impact so it will be interesting to examine the relative impact of endorsement versus rejection papers and analyse citation networks. For example, to what extent do rejection papers cite other rejection papers? We really only scratched the surface in Cook et al. and there are a number of further research questions.

2. The importance of two channel science communication

Two paradigms in the world of science communication are the information deficit model (the public will get with the program if we can just get them enough information) and cultural cognition (people assimilate information in a biased manner based on cultural values). Neither paradigm on their own can fully explain the public’s climate beliefs.

I’ll demonstrate this with observed data conducted as part of my PhD research into belief polarization (with a particular focus on the psychology of consensus). The following graph plots perceived scientific consensus from a survey of a US representative sample versus support for free markets (-1 is equivalent to extreme liberal and 1 is equivalent to extreme conservative). I’ve included a green dotted line indicating the 97% consensus (I can never resist adding some bling to my graphs).

Several features jump out from this data. We confirm recent studies and public surveys that find a significant relationship between perceived consensus and free market ideology (p < .00001). Also, it’s hard not to notice the sharp drop in perceived consensus for participants with very strong support for free markets (e.g., extremely conservative).

However, the key feature for the purpose of this discussion is that even for liberals, there is still a large gap between perceived consensus and the 97% reality. This indicates that there is still a significant lack of awareness of consensus, even among those whose cultural values are predisposed to agree with AGW. This misperception is crucial given the link between perceived consensus and public support for climate policy (more on that later).

What is required is two-channel science communication – information content with cultural meanings. The consensus gap across the whole ideological spectrum means that we need to increase overall public awareness of consensus. The ideological bias necessitatesframing scientific messages inways that don’t threaten worldview. An important element to worldview-friendly framing is using messengers who share the values of the audience.

The two approaches are not mutually exclusive. Rather, both approaches should be integrated and complement each other. To quote, well, Dan Kahan: “to be effective, science communication must successfully negotiate both channels.”

If the public have heard ‘over & over & over that there is “overwhelming scientific consensus” on AGW’, then why is public perception of consensus so low, even among liberals? A significant contributor is a focused and persistent misinformation campaign designed tomanufacture doubt. Science communicators are not operating in a vacuum but competing in a marketplace of ideas.

As early as 1991, Western Fuels Association spent $510,000 on a campaign to “reposition global warming as theory (not fact)”. The infamous Frank Luntz memo urged Republican politicians to “…make the lack of scientific certainty a primary issue in the debate.”  In an analysis of syndicated conservative opinion pieces, the most popular myth was “there is no consensus”. Lobbyists sought to undermine Obama’s climate policy by attacking the consensus, arguing “If we win the science argument, it’s game, set, and match.”

It’s becoming increasingly apparent that attacking the consensus has been a central strategy of opponents of climate action. This strategy employs dissenting scientists as spokesmen toconstruct the impression of ongoing scientific debate.Adam Corner’s paper on attitude polarisationfound that when presented with opposing views on climate change, both conservatives and liberals become more sceptical. Based on Lewandowsky’s research, I would guess that what is happening in Corner’s experiment was participants came away with the impression of a 50:50 debate, a lowered perception of consensus and hence lowered belief in climate change.

Lewandowsky’s study demonstrated a causal link between perception of consensus and climate belief. When people were presented with information about the scientific consensus, their belief in AGW increased significantly. But the most interesting result that came out of this research, in my opinion, pertains to Dan’s last point on polarization.

3. Consensus: polarizing or neutralising ideology?

The intriguing result coming out of Lewandowsky’s studywasthat the consensus message partiallyneutralisedthe biasing influence of ideology. The graph below shows belief in AGW for a control group who received no information (red) and a group that received consensus information (green). The x-axis is support for free market where similar to my first graph, negative values equate to liberals and positive values equate to conservatives.

The change in belief is greatest among conservatives. The result among those receiving the consensus information is less influence of ideology on climate belief. This research doesn’t remove the danger of a consensus message backfiring among certain cultural groups. Lewandowsky’s study was conducted with an Australian sample where extreme free market support was weakly represented. My recent data indicates that at the extreme end of the free market scale, consensus messages are received poorly. Nevertheless, my data also confirms Lewandowsky’s result. Overall, belief in AGW significantly increases in response to information about consensus.

A key point that mustn’t be overlooked is that two studies (Ding et al 2011 and McCright et al 2013) have shown that perception of consensus is linked to support for climate policy. When people understand that climate scientists agree on human-caused global warming, they are more likely to support climate action. Both studies stress the importance of campaigns to correct the public misperception about consensus.

These two points taken together –consensus messages increase belief in AGW and perception of consensus is linked to support for climate policy – underscore the importance of consensus in climate communication. This is not to say that consensus is a magic bullet that will solve all our problems and usher society into an environmental utopia. If only things were that simple! But by reducing the consensus gap, we are removing a significant roadblock to public support for climate action.

I don’t explore these issues in Cook et al. because the nuances of science communication and cognitive psychology are beyond the scope of the paper (but are the focus of my PhD research). McCright and Dunlap recommend increasing awareness of the consensus without triggering ideological polarization although they don’t explore how to achieve that. With typical peer-review understatement, they describe this as “a major challenge”.I’m not saying I have all the answers regarding how to craft a consensus message that avoids backfiring among strong free market supporters. This is very much an open research question and something I’m keen to explore further.

Nevertheless, to fail to promote scientific consensus because of possible backfires is to deny ourselves a powerful message that overall increases climate belief and public support for climate policy.Given the misinformation campaign focused on casting doubt on consensus, it also cedes the territory to climate misinformers which would only exacerbate the situation. The question isn’t whether consensus is an important climate message – the question is how best to deliver that message.

Thanks again to Dan Kahan for the opportunity to discuss this important issue.  It would be great if the authors of Ding et al. or McCright & Dunlap, who recommend campaigns to correct the public misperception about consensus, could weigh in with their thoughts.I’ll be presenting the results of my PhD research at the AGU Chapman Conference on Climate Communication in Colorado next month so I look forward to further discussion on the topic there.

Leave a Comment