The Science of Science Communication II: Summary of a Colloquium (2014)

1

The Sciences of Communication

Several years ago, a committee appointed by the National Academy of Sciences began working on the third edition of the book Science and Creationism. The first two editions of the book had been widely cited in legal cases, said Barbara Kline Pope, executive director for communications at the National Academies and director of the National Academies Press, in her introductory remarks. But the committee and its staff wanted the third edition to have a bigger impact. At the time, intelligent design creationism was a relatively new concept, and it was being pushed into science classrooms by vocal and well-financed groups. The committee decided that formal audience research was warranted—despite the time and cost it would add to the project—to gauge what people believe about evolution, intelligent design, and creationism.

Just as the facilitators’ guide for the focus groups was being written, Judge John Jones issued his decision in the case Kitzmiller v. Dover Area School District, ruling that intelligent design “cannot uncouple itself from its creationist, and thus religious, antecedents.” The staff working on the book thought that this message would be a “slam dunk,” said Pope. They felt that the message would resonate strongly with audiences and directed the focus group facilitators to point out that a judge had decided that intelligent design is a form of creationism and thus religion and that teaching it in science classrooms is therefore unconstitutional and illegal.

Pope was watching the live focus groups on her computer while eating dinner. “I slowly lowered my fork to my plate, and my jaw dropped with it. I saw backs seizing up and eyes getting squinty, and one guy said,

‘No judge is going to tell me how to run our schools.’ I saw the rest of the participants clenching their jaws and nodding enthusiastically.”

The staff was shocked at how wrong they had been, especially when a quantitative survey uncovered the same attitudes. Relying on their intuition about effective messages would have been “a very bad idea,” said Pope, and the third edition of Science, Evolution, and Creationism (NAS/IOM, 2008) is a very different document than it would have been had audience research not been done.

The Science of Science Communication II colloquium was similarly devoted to using the best available evidence to guide science communication. The colloquium was built on the first Science of Science Communication colloquium,¹ but it sought to dig deeper into the methodologies, analyses, and findings of science communication research. It also featured, on the third day of the colloquium, concurrent workshops on four pressing topics—evolution, climate change, nanotechnology, and nutrition and obesity—where researchers and practitioners could develop research-based insights on communication strategies that would have immediate application.

LAY NARRATIVES AND EPISTEMOLOGIES

Science communication occurs through artifacts, including language, diagrams, and other representations. These artifacts both reflect the cultural assumptions of their creators and reinforce different ways of seeing the world, said Douglas Medin, the Louis W. Menk Professor of Psychology at Northwestern University. Science communication, therefore, needs to pay attention both to the artifacts with which it is conducted and to the different ways people have of looking at the world.

As an example of cultural differences in perspectives, Medin cited cognitive research comparing East Asians, typically Chinese, Japanese, and Koreans, with westerners, typically people from the United States. East Asians tend to pay more attention to background information, while westerners attend more to focal objects. For example, when shown successive pictures that look very similar, East Asians are much better at detecting background changes, while westerners are better at detecting foreground changes. Another study found that western paintings have three to four times as much representation devoted to faces, while East Asian portraits include more background information. The same difference was reflected in the aesthetic preferences of East Asians and westerners.

____________

¹ See http://www.nasonline.org/programs/sackler-colloquia/completed_colloquia/science-communication.html.

Native Versus European American Perspectives

Medin and his colleagues at Northwestern have been involved in a collaborative research partnership with the American Indian Center of Chicago and the Menominee Nation of Wisconsin. Four thousand to five thousand Menominee, who are the oldest continuous residents of Wisconsin and are well known for their sustainable forestry practices, live on tribal lands in and around three small communities. Interviews with Menominee and European American parents and grandparents revealed large differences in distancing discourse. When European American parents and grandparents were asked about the five things they would like their children or grandchildren to learn or know about the biological world, they talked about nature as an externality. They wanted their children to respect nature and know they have a responsibility to take care of it. Native American parents and grandparents were much more likely to say that they wanted their children to understand that they are a part of nature.

Another example of distancing discourse comes from depictions of ecosystems in publications by westerners. Virtually none include humans as part of the representation, suggesting that westerners generally think of themselves as outside of ecosystems.

Another demonstration of differing perspectives comes from an analysis of children’s books written by Native American and European American authors. The illustrations by Native Americans tend to have closer, more personal, and more wide-angle representations. As a result, they provide more alternative perspectives. The books by European American authors were more likely to have straight-ahead perspectives at eye level. The Native American books were more likely to provide the perspective of an actor in the scene by using an over-the-shoulder or embodied representation.

The texts of the books also differed. Native American–authored books were more likely to mention seasonal cycles, native animals, and objects that, in a western perspective, would be part of the background.

Conceptions of Nature

These results parallel those from cognitive experiments on conceptions of nature. For example, when Native American and European American adults in rural Wisconsin were asked to describe the last time they went fishing, the median point at which European Americans used the word “fish” was the 27th word, whereas the median for the Native American Menominees was the 83rd word. The Native Americans were much more likely to supply context and background information—so much so that some never mentioned fish at all.

In another study, about 40 percent of Menominee children spontaneously imitated or took the perspective of the animals when shown pictures containing those animals, while no European American children did so.

In a set of related studies, fishing experts were asked to sort 44 local species of fish into categories that made sense to them. European American experts tended to sort the species either in terms of taxonomic relations, such as the bass family, or by goals, such as large and prestigious game fish. Menominee experts were much more likely to sort ecologically or by habitat, such as fish found in fast-moving water. However, when the groups were asked to sort the fish by habitat, the differences disappeared, indicating that the intergroup differences involve the organization of knowledge rather than knowledge per se. In a follow-up study, Menominee experts were more likely to recognize positive reciprocal relationships among fish species, such as the reciprocal eating of spawn, fry, and small fish, while European American experts mentioned fewer relations, and those that they did mention primarily involved adult fish.

Implications for Science Communication

These findings have some important implications for science communication, Medin concluded. Distancing and outsider perspectives can undermine engagement with science. The use by researchers of terms such as “the public” also can be distancing and homogenizing.

In addition, community members bring many skills to an exchange with experts. In work the collaboration has done on community-based citizen science, participants have had strong backgrounds in chemistry, hydrology, and forestry.

Finally, mismatches between lay epistemologies and orientations implicit in communication may be a source of alienation, Medin said.

Artifacts That Shape Perspectives

The research described by Medin raises intriguing questions about the everyday artifacts that shape views of humans in nature, said Ann Bostrom, the Weyerhaeuser Endowed Professor of Environmental Policy at the University of Washington. For example, how are children’s perspectives shaped by what they see? And how readily can these perspectives be changed?

Bostrom addressed the first question by considering several video games popular among children. In the game Knytt Underground by Nicklas “Nifflas” Nygren, children explore a natural landscape from a third-person perspective. In the game Minecraft, on the other hand, children play from a first-person perspective. Some games, such as Kerbal

Space Program, can be played from either a first-person or third-person perspective, though one perspective may be easier to play than the other, depending on the game.

Other visual representations also can be either first or third person. For example, popular earthquake films can leave the viewer outside the scene by taking a third-person perspective or draw a viewer into the scene with a first-person perspective.

The basic metaphor for these representations, as Stanford psychologist Barbara Tversky has pointed out, is proximity, and proximity has an influence on cognitive progresses. Because humans have embodied minds and points of view, our spatial orientation affects the speed with which we process information. For example, from top to bottom is easier for us to discriminate than from side to side, because our bodily axes and asymmetry affect how we process information.

With regard to artifacts, readers form mental images from texts. These images can have a variety of perspectives due largely to the abstractness of texts. Translations offer a good example. One such translation is of an ancient poem by Wang Wei entitled “Deer Park”:

There seems to be no one on the empty mountain…

And yet I think I hear a voice,

Where sunlight, entering a grove,

Shines back to me from the green moss.

A more recent translation is by the poet Gary Snyder:

Empty mountains: no one to be seen.

Yet—hear—human sounds and echoes.

Returning sunlight enters the dark woods;

Again shining on the green moss, above.

Changes in perspectives also characterize mental models of hazardous processes, Bostrom observed. For example, about 10 percent of people asked in one study about climate change said that they have direct experience with climate change—for instance, through changes in the seasons. In other cases, people recruit mental models that do not depend on personal experience.

Stories create meaning, Bostrom concluded. They make causality concrete and close. But can stories alone enable readers to shift fluently between points of view? The broad environmental expertise of the Menominee may enable them to shift fluently between perspectives, as well as to better distinguish between correlation and causation. But the question of which artifacts matter for science communication remains largely unanswered.

Multiple Representations of Knowledge

Multiple representations of knowledge are also an important consideration in science education, said Kevin Dunbar, professor of human development and quantitative methodology at the University of Maryland in College Park. In the past, science education has been oriented toward filling students with facts that they can repeat back on tests. More recently, education has emphasized the construction of knowledge at a social as well as an individual level. By having students pose questions on nature, they are expected to learn through discussion. Students also learn by confronting their naïve conceptions with the results of classroom experiments.

However, what if multiple valid epistemologies exist, Dunbar asked. Similarly, if different cultures have different views of science, is the result different sciences? For example, do different Native American nations have different epistemologies, and if they do, should the differences be reconciled or should those differences be used to communicate science?

Dunbar has worked as a consultant on the Trail of Time in the Grand Canyon. At least 12 American Indian tribes think of the Grand Canyon as their homeland, and these tribes can have different epistemologies related to time. How can these differences be presented in positive way rather than one group being wrong and one being right? This is a major goal of science communication, said Dunbar—to deal with differences in a constructive way.

Finally, Dunbar mentioned ongoing research into how culture changes the brain. For example, are epigenetic changes a mechanism for the embodiment of cultural knowledge? Studies looking at epigenetic changes following educational interventions have been “very suggestive,” but much more work needs to be done to know whether biology can inform science education.

Multiple Cultures in Science

A prominent topic of discussion during the question-and-answer period was the influence of multiple cultures on science. As Medin pointed out, the way that science gets done depends on the cultures of the people who are doing it, and multiple approaches to science make for strong science. This is a strong argument, he said, for diversity in research teams and in science education. For example, Medin’s collaboration had a diverse research team to approach problems from multiple perspectives. However, attracting diverse groups to science can be difficult, Medin added. Unlike medicine, which often reflects the deepest values of medical students, the study of science often does not always allow students to express their deepest values.

As another example of the value of multiple perspectives, Medin cited research in primatology, which has made progress both from a western

orientation that sees researchers as distant from nature and from an eastern perspective that sees researchers as part of nature. Dunbar pointed to the differences in culture between U.S. molecular biology laboratories, which tend to be inductive, and Italian laboratories, which are much more deductive, even though the laboratories produce papers that are very similar and are published in the same journals.

The presenters also discussed alternatives to the term “the public,” given the diversity of audiences that science communicators would like to reach. Bostrom suggested using the names of professions, the groups with which people are involved, the communities of which they are members, the roles they play in relation to the use of science, or the technologies they use.

Discussion also revolved around the extent to which perceptions shaped by culture can be changed. Medin pointed out that cultural artifacts can have causal force. For example, presenting Americans with typical Japanese scenes can lead the American participants to become relatively better at detecting background changes, though perceptions are probably both chronic and flexible, he said. Bostrom observed that perceptions are reinforced by the roles people take in the world, which tend to reinforce both their cultural positions and their ways of thinking about the world. Yet psychology research shows that people’s views often depend on context as well.

Medin pointed out that there are hundreds of federally recognized Native American tribes, and so results from one tribe cannot generalize to all, and great diversity exists even within a single culture. As Bostrom added, even with professional groups such as hurricane forecasters and emergency responders, differences in the mental models within a group are larger within the differences among groups.

MOTIVATED AUDIENCES: BELIEF AND ATTITUDE FORMATION ABOUT SCIENCE TOPICS

Audiences’ motivations as human beings affect how they interpret science communications, observed Susan Fiske, Eugene Higgins Professor, Psychology and Public Affairs, at Princeton University. With climate change communications, for example, people who pay more attention to politics in the news have perceptions polarized away from those of people who pay more attention to scientific and environmental stories in the news. Furthermore, over time the amount of skepticism about climate change has increased, despite the increasing scientific consensus on the role of human beings in the changing climate. And to the extent that information about climate change is getting through, members of the

public are getting more alarmed, which is not necessarily the message that policy makers or scientists wish to convey.

The first step in reconciling differences between scientific and public perspectives is to recognize that scientists are not the sole source of valid information, Fiske said. The second step is to recognize that both cognition and emotions, or affect, influence perceptions and interpretations. Together, cognition and emotions create motivation, and persuasion works best when both factors are taken into account.

People are not idiots, Fiske said. The public now knows more about climate change than in the past, and it generally can distinguish science from nonscience. Some have a classical view that science yields a single true picture of the world, while some have a more modern view that science can produce multiple answers that have to be negotiated and debated.

The Credibility of Communicators

To be credible, communicators need both expertise and trust, Fiske said. If someone is seen as an expert on a topic, other people tend either to agree with that person or at least think about a message. But communicators also need to be trusted to be effective, which means that they need to be seen as having a motivation to be truthful and accurate.

Trust has been a largely neglected topic in the science of science communication. In general, people trust those who they think are like themselves. People who belong to a group have a shared reality and a motivation to share understandings. “This is human nature,” explained Fiske. “People trust people who they think share their values [and] goals. … This is a core insight within social and behavioral science.” Group membership provides a sense of control over one’s environment and circumstances. It also enhances feelings of self-worth. Thus, both cognitive and affective factors affect trust.

Warmth and Competence

Fiske and her colleagues have developed a framework for understanding the social and cultural landscape of groups. The first question people ask, in identifying whom to trust, is whether a person is friend or foe. If a person is seen as being on the same side or sharing the same values, they are seen as trustworthy and warm.

The second question people ask is whether the other can act on their own intentions. In other words, is the other person competent so that their acting on those intentions will produce a desired (or undesired) outcome?

The combination of warmth and competence produces a two-dimensional space that people use to interpret communications. For example, according to research by Fiske and colleagues that was under review at the time of the colloquium, polling data from more than 30 countries demonstrate that the middle class, white people, and blue-collar people are seen as high in both warmth and competence. Poor people and teens are seen as being neither warm nor competent—and the results for poor people are true all over the world, said Fiske. Children and old people are seen as being well intentioned but not very competent, while rich people are seen as competent but not warm, again all over the world.

Fiske noted that responses to people in the four quadrants of this two-dimensional graph fall into four emotional categories. Cold and incompetent people tend to be treated with disgust, warm and incompetent people with pity, competent and cold people with envy, and competent and warm people with pride.

According to pilot data collected online, when professionals are assessed against the dimensions of warmth and competence, researchers and scientists are seen as competent but cold, while professors and teachers are seen as both competent and warm—though not as competent and warm as doctors and nurses. When asked about the emotions felt toward these groups, researchers and scientists, in keeping with their position in the two-dimensional space, were more often the subjects of envy. People cooperate with envied groups because they have needed resources, including knowledge. But these groups can be attacked in times of instability, which creates a dangerous ambivalence. Envy implies that “you have things that I respect and I’d like to have, and I’d like to take them away from you,” Fiske said.

Cold competence also can create resentment. For example, envied groups can be the object of Schadenfreude—the sense of pleasure at someone else’s misfortune. When electrodes are connected to people’s facial muscles, images of someone from an envied group getting his or her comeuppance often generate smiles. “When a guy in an Armani suit gets splashed by a taxicab or sits in gum on a park bench, people smile. They can’t help it.”

Increasing Warmth

People tend to believe that scientists and researchers are competent but do not trust their intentions. For example, when asked about the intentions of climate scientists, some answered that scientists might lie with statistics, complicate simple stories, feel themselves superior to nonscientists, pursue a liberal agenda, or provoke and hurt big corporations.

The most common answer is that scientists might slant research to get research funding. “That’s our Achilles heel,” said Fiske.

Public perceptions of science and scientists are more polarized today than they have been in the past. For example, even the proposal to name a science laureate for the United States has encountered political resistance. Opponents felt that a science laureate would have a pulpit to talk about values rather than science and disseminate a political agenda.

The news is not all bad, Fiske continued. Scientists gain a measure of trust because they are seen as interested in educating the public, preserving the environment, and saving humanity. People respect the educational mission of scientists and researchers in the same way they do those of teachers and professors. People also tend to trust an impartial agenda and not trust a persuasive agenda, which argues for separating science communications from the policy implications of those communications.

Scientists would be more trusted if they emphasized deliberation rather than persuasion, Fiske said. Scientists need to respect the intelligence of their audiences. They need to convey information and resist issuing policy conclusions unless they clearly label such conclusions as their own opinions.

One way to warm up scientists would be to emphasize their service to the public through forums such as the National Academy of Sciences, Fiske emphasized. The teaching role of scientists also generates positive reactions among members of the public. Similarly, letting people know why someone went into science, and having a diversity of people working in the sciences, can increase trust. By clearly expressing motivation, scientists can establish credibility and not be treated simply through stereotypes. Though the incentive structure in U.S. universities remains oriented toward research, teaching in universities, through its influence on the next generation of managers and policy makers, can improve the public’s trust of the scientific community.

Influences on Perception

Craig Fox, Ho-Su Term Chair in Management at the UCLA Anderson School of Management, elaborated on the unconscious allure of in-group positions and the polarized political environment that tends to drive people’s perceptions apart. In one experiment, people were asked what their political affiliation was, after which they were asked whether they wanted to invest in a conservative, moderately conservative, moderately risk-tolerant, or risk-tolerant investment portfolio. People who identified as Republicans were attracted to the conservative option, while Democrats were attracted to the more risk-tolerant options. However, when they were asked about the investment decision first, Republicans and

Democrats tended to choose the low-risk options with relatively equal frequencies.

Choices also tend to be supported by an illusion of understanding, Fox observed. People are overconfident in how well they understand how everyday objects, such as toilets or ballpoint pens, work. But when they are asked to explain in detail how such an object works, they realize that the mechanisms are more complicated and lower their self-assessed understanding of an object.

The same effect applies with public policies. For instance, when the Supreme Court upheld most of the provisions of the Affordable Care Act in 2012, more than three-quarters of Americans in a Pew poll expressed a perspective on whether they supported or opposed the ruling. However, barely half of them could correctly identify what that decision was. In another experiment, Fox and his colleagues presented individuals in an online sample with several policy issues—for instance, a cap-and-trade policy to curtail carbon dioxide emissions or a national flat tax. Respondents gave their positions and their level of understanding of the issue, after which they were asked either to give the reasons for their beliefs or to explain how the policy would have the desired effect. People who had to explain how a policy works subsequently rated their understanding of the issue as lower. They also described themselves as less likely to contribute to advocacy causes, especially those who were most extreme in their views.

Self-assessed understanding of scientific issues displays the same pattern. When people were asked to explain how carbon emissions affect climate change, they later rated themselves as having less understanding of the issue and more moderate positions—an effect not seen when they were asked just for the reasons for their beliefs. However, people whose illusion of understanding had been punctured were also less willing to support further research on the topic. “That’s something we need to look at,” said Fox.

Overweighting Marginal Views

Even on scientific topics surrounded by considerable consensus, such as climate change, low-probability events tend to be overweighted in making decisions, Fox observed. For example, if people are told that 10 percent of scientists believe x, many interpret this statement as meaning that x could be true or false, even though the position is held by only 10 percent of scientists. However, if people are led through the response of each scientist one by one—so this scientists believes x, this scientists believes x, this scientist does not believe x, this scientists believes x, and so on—they become more sensitive to the actual probabilities.

This finding has implications for how the media report on scientific results. A story that gives roughly equal emphasis to both sides of an issue may accentuate the public’s bias to give undue credence to remote views.

People are undersensitive to probabilities in general, Fox said. For example, even as more scientists have become confident about the human contribution to climate change, public beliefs have not changed accordingly. In particular, media stories that devote time to marginal views have an exaggerated impact on the public’s perceptions. The media need to think creatively about ways to communicate relative proportions in a way that people can absorb, he concluded.

What People Want to Know

As a further example of the ways many people are concerned about the motivations of scientists, Bill Hallman, professor and chair of the Department of Human Ecology at Rutgers University, described a set of open-ended interviews done with approximately 30 people on their perceptions of animal cloning. At the end of the interviews, the people were asked what else they would like to know about animal cloning research. Here are the top 10 questions in the order of how often they were asked:

 1.  Who is doing it?

 2.  Where are they doing it?

 3.  Why are they doing it?

 4.  What are the goals of this?

 5.  What is the status of the research?

 6.  What are the risks and benefits of the research, both to consumers and to the animals?

 7.  Who is monitoring and regulating the research?

 8.  How does it work?

 9.  What happened to Dolly?

10. Will we eat the food from cloned animals, and is it safe?

The most asked question involves who is doing the research. The third to the last question is how cloning works. Yet one of the first things science communicators want to do is explain the science, Hallman observed, even though this question ranks relatively low on the list of questions people ask. And if people do not get answers to the questions they ask most often, they are unlikely to absorb the answers to questions that are of lower concern.

Hallman also emphasized that people create mental models based on the information that is available to them. For example, science often works on objects and issues that are largely invisible to people, such as nanotech-

nology, climate change, or food safety risks. Science communicators need to understand the mental models people create of invisible things and speak in ways that connect with what people think they already know.

Improving Perceptions of Science

On the issue of warmth and competence, Hallman pointed to data suggesting that the perceived mismatch in a person who is coldly competent can exacerbate mistrust of that person. These data point to the need for scientists to develop greater social and emotional intelligence. Too often, said Hallman, scientists talk as if they are trying to impress other scientists, whereas if they put themselves in the position of a young student or a member of the general public, they could connect with what their audiences already know.

The media, which relentlessly stereotype scientists as cold, are part of the problem. But scientists also need to work against the stereotype. “We need, as part of our science education, to teach people how to tell a story, with a beginning, a middle, and an end, to tell a joke that’s actually funny, and to take a joke when it’s warranted.” Standard resumés contain plenty of information on expertise and competence but very little on social abilities, despite the importance of these attributes.

The Problem of Persistent Minorities

One of the issues discussed during the question-and-answer period was the general reaction to individuals or small groups who insist that they are right and everyone else is wrong—what Fiske termed “persistent minorities.” People often ascribe credibility to such individuals because they are standing up to everyone else and not caving in. “It’s like the holdout in a jury,” Fiske said. “That person has to be really motivated to resist all the other people on the jury.” Because of this perception, people may favor a minority perspective even if it lacks credible evidence.

Hallman also brought up the costs to scientists when they claim to know what is going on and subsequently are shown to be wrong. A better strategy is to say what is known now and how likely that observation is while also describing what is being done to reduce uncertainty. In that case, if a scientist turns out to be wrong, people are more forgiving.

Fiske agreed that scientists do not have the right to tell people what to do. They have to provide information and talk about consequences in the most accessible way possible. Also, people need solutions, not just problems. If a scientist simply makes people afraid, they will avoid the topic, since they will assume that they cannot do anything about it.

Finally, Fox noted that stories can be more successful with a lay audience than data, since people’s emotions are driven more by stories than by statistics. “A lot of science can be communicated in stories,” he said.

COMMUNICATING UNCERTAINTY

To provide answers to the questions that decision makers ask, science communicators need to be able to communicate uncertainty, said Baruch Fischhoff, the Howard Heinz University Professor in the Departments of Social and Decision Sciences and Engineering and Public Policy at Carnegie Mellon University. Often that requires greater precision than scientists and communicators naturally provide. Behavioral decision science shows the problems caused by such ambiguity and ways to do better. It involves both analytical research, determining what people need to know, and descriptive research, making that information comprehensible. For example, studies of intelligence analysts have revealed the confusion potentially caused by verbal quantifiers such as that describing a Soviet attack on Yugoslavia as “a serious possibility.” The U.S. intelligence community’s current attempt to deal with this problem is seen in the standard explanation used in the National Intelligence Estimate: Prospects for Iraq’s Future: A Challenging Road Ahead (FAS, 2007):

When we use words such as “we judge” or “we assess”—terms we use synonymously—as well as “we estimate,” “likely” or “indicate,” we are trying to convey an analytical assessment or judgment. These assessments, which are based on incomplete or at times fragmentary information, are not a fact, proof, or knowledge. Some analytical judgments are based directly on collected information; others rest on previous judgments, which serve as building blocks. In either type of judgment, we do not have “evidence” that shows something to be a factor that definitively links two items or issues.

Although he appreciates the motivation underlying this clarification, Fischhoff said that “you would be hard pressed to figure out what they meant if you were actually a consumer of these documents.”

Similarly, a recent analysis of 12 extreme climate events during 2012 stated that

Approximately half the analyses [19 analyses by 18 different research groups on 12 extreme events during 2012] found some evidence that anthropogenically caused climate change was a contributing factor to the extreme event examined, though the effects of natural fluctuations of weather and climate on the evolution of many of the extreme events played key roles as well.

As Fischhoff observed, “You’d have to be a very select user audience to get the information you needed to make any decisions” from this analysis.

Questions Decision Makers Ask

The questions that decision makers ask of scientists can be divided into three broad categories, Fischhoff continued:

1.  Whether to act,

2.  What to choose, and

3.  Whether and how to create options.

For each of these categories, communicators face both analytic challenges in extracting the information experts know and communication challenges in conveying that information with the precision that decision makers need.

Decisions about whether to act often involve determining whether a signal has passed a threshold. In this case, the analytical challenge is determining how much scientific information is available and translating this knowledge into an action threshold. Signal detection theory studies how well experts can discern a signal and how their reporting balances the costs and benefits of correct and incorrect responses. For example, Mohan et al. (2012) have looked at the decisions that physicians at regional or primary care hospitals make about transferring a patient to the emergency room at a tertiary (major) medical center hospital. Their analysis found great variability among physicians in determining when a patient meets the guidelines for transfer. The communication challenge in such a situation entails describing the decision rule that physicians use and how good physicians are at following it. Such communications reveal how much these experts know and how appropriately they apply that knowledge.

The next category of questions involves deciding between two or more alternatives. Here, the analytical challenge can be expressed by a decision tree, which, again, considers the uncertainties and expected consequences of each choice option. For example, the Food and Drug Administration uses a structured approach to summarize the expected benefits and risks in its drug regulatory decision making (FDA, 2013). FDA’s approach considers the evidence, uncertainties, conclusions, and reasons associated with five decision factors. Those uncertainties may arise from variability in the observations, the internal validity of studies, their external generalizability, and the quality of the underlying science. Patients and physicians then can use this information in making their own decisions about a treatment.

The communication challenge in conveying such information includes explaining studies’ methodological flaws and hidden values. All analyses embody values that favor some interests. When transparent, those assumptions can be controversial. However, they are often obscured in the measures that scientists and analysts use.

The third class of decisions involves whether and how to create options. The analytical challenges in making these decisions include identifying the relevant expertise and assessing the uncertainties created by omissions in the analysis. For example, an analysis by Casman et al. (2000) looked at options for dealing with drinking-water contamination, such as sending out notices to boil drinking water. The system for preventing health effects is complicated, involving detection, management policy, the public health system, how the messages go out, whether people know how to boil water well enough to eliminate contaminants, and other factors. If the analysis left out any of these factors, it could be missing an important part of the problem. Without such a comprehensive view, decision makers cannot understand the uncertainties underlying their choices. One particular challenge for this class of decisions is conveying the implications of excluded information.

A Reluctance to Express Uncertainties

Experts are often reluctant to express their uncertainties, Fischhoff observed. Sometimes they see such efforts as misplaced imprecision. Sometimes they think they will be misunderstood. Sometimes they fear being punished by their organizations for their candor. And sometimes they are uncomfortable with the elicitation method, not knowing quite how to express themselves in the required form.

Fischhoff had three proposals for dealing with the reluctance to grapple with uncertainties. One is to create standard procedures for making and communicating decisions. With such procedures, people get used to thinking in a particular way, organizing their evidence, expressing their thoughts, and getting feedback on how well they have done. For example, standardization has helped FDA not only with its external communications but also with its internal communications among its own personnel.

The second proposal is to create a resource center to provide experts with publication-quality support in eliciting and communicating uncertainty. Such a center could provide quality assurance, take advantage of economies of scope, anticipate common problems, form trusted personal relationships, and stimulate basic applied research into the challenges associated with analyzing and communicating uncertainty.

The third proposal is to create shared understanding of the analytical approaches needed to characterize uncertainty. All communication begins

with an analysis of what people need to know and of the associated uncertainties. If scientists understood analytical procedures, they could work more easily with those who do communication science research.

An orderly treatment of uncertainty would produce more useful science by addressing decisions makers’ needs, Fischhoff concluded. At the same time, it would produce better science by encouraging disciplined reflection on the uncertainties in scientific knowledge.

Uncertainties in Communicating Uncertainties

Communicating about uncertainties has its own uncertainties, said David Budescu, the Anne Anastasi Professor of Psychometrics and Quantitative Psychology at Fordham University, in his comments on Fischhoff’s presentation. Some communications inform multiple types of decisions. Audiences are highly heterogeneous in their knowledge bases and mental models. And audiences may lack sensible models of the sources, causes, types, and limits of uncertainties. Many people, for example, think all uncertainties are alike, which can be highly misleading.

Given these uncertainties, the communication of uncertainties should be judged not by an absolute threshold but by the demands of particular circumstances, said Budescu. Frameworks of communication need to be flexible enough to accommodate different kinds of decision cycles and various levels of knowledge about uncertainty.

As Fischhoff said, experts are sometimes afraid to be perceived as too imprecise. But imprecision is sometimes necessary. With Tom Wallsten, Budescu developed the principle that uncertainty should be communicated in a mode that matches the nature of the event and the sources of uncertainty. For example, it makes little sense to communicate precisely vague uncertainties about ambiguous events, such as “the chance of abrupt change in the climate in the near future is 0.128.” This would be an overly precise estimate for such an ill-defined and ambiguous outcome. By the same token, it is suboptimal to communicate imprecisely precise uncertainties about unambiguous events, such as the “chance of drawing the queen of hearts from a full deck of cards is quite low.” The uncertainty needs to match the nature of the event.

When publicly traded companies issue forecasts of earnings per share, they often use imprecise forecasts, Budescu noted. An analysis of more than 33,000 quarterly forecasts by almost 5,000 companies issued between 1996 and 2006 found that forecasts citing a range were more accurate than point forecasts (51 percent versus 24 percent). Furthermore, the receivers of such information tended to prefer broader estimates because they judged the forecasts to be more informative, more accurate, and more credible.

People often have clear expectations of the appropriate level of precision or imprecision to communicate, Budescu concluded, and they value communications that more nearly match their expectations.

Meeting the Needs of Decision Makers

Adam Finkel, executive director of the Penn Program on Regulation and a senior fellow at the University of Pennsylvania Law School, made four points in agreeing with Fischhoff’s analysis. First, one type of uncertainty is generated by interindividual variability. Unlike uncertainty, variation is irreducible, and unlike uncertainty, it forces decisions about risks and benefits to individuals.

Second, simple innumeracy can be the cause of mismatches between how experts offer information and how the public interprets it. As an example, Finkel cited a New York Times article that reported the failure rate of in vitro fertilization to be 77 percent while not mentioning that the failure rate of natural conception is also almost exactly 77 percent.

Third, instead of producing a weighted combination of mutually irreconcilable probabilities, a better option is a solution-focused process of decision that uses risks and benefits to discriminate among choices. For example, taking the midpoint of the probability distribution for hurricanes forming in the Gulf of Mexico is suboptimal because it does not consider the costs of error.

Finally, the most valuable information usually concerns the specific uncertainties that are plaguing decision makers and reducing the expected regret of the best possible decision.

The Responsibilities of Decision Makers

Given these four points, Finkel emphasized placing more of the onus on decision makers to demand better information. Decision makers need to act when they are not satisfied with the information they are getting because it allows them to make only rudimentary decisions.

He also emphasized that, in the work that he does on cost-benefit analysis for environmental and occupational health, the only reason to act is risk, and the only reason not to act is cost. However, the estimation of cost may be the broken link in cost-benefit analyses. Risk is what will happen if a policy is not created or implemented. Cost is what will happen if a policy is created and implemented. Yet economists have devoted relatively little attention to what policies cost. They also tend to minimize the uncertainties in estimations of cost, interpreting error bars as indicating an unfinished analysis, said Finkel.

Finally, Finkel cited two problems with estimates. First, they are often based on the mean, but the distribution of events around the mean is rarely distributed on a Gaussian curve. Rather, the mean depends on the characteristics of the tails of the distribution.

Second, the public and decision makers often are bound to what Finkel called “estimaticles”—estimates that, in the words of William Blake, act as “mind-forged manacles.” When fully informed, a decision maker may choose a different estimator than one imposed by an economist or may conclude that no single estimator tells the whole story.

The Meanings of Uncertainty

During the discussion session, one colloquium participant pointed to the difficulties surrounding the word “uncertainty,” which can lead decision makers to delay deciding until the uncertainty has been reduced. In such cases, it would be better to talk about confidence, best scientific opinion, or best estimate rather than uncertainty.

Fischhoff responded that these terms need to be tested, as do the other aspects of analyzing and communicating uncertainty. Uncertainty has a precision that works in the scientific community, but it may not work with a particular audience. Finkel added that deciding not to decide is still deciding, which essentially leaves a risk unchanged. Budescu agreed that uncertainty can be used as an excuse for inaction, as has often been the case with climate change. But with climate change, the uncertainties involve the rate of warming, not whether warming is occurring. Thus, the magnitudes and roles of uncertainty in a model need to be carefully specified and explained.

The presenters also examined several possible roles of scientists in communicating uncertainties. Finkel pointed out that the perception of being apolitical is crucial for scientists. For that reason, scientists should not overstep their bounds. They can provide information, including information about uncertainties, but decision makers should be the ones making the decisions.

However, in some cases, being apolitical may not be the best way of effectively communicating scientific information, he continued. It is more important to be transparent about one’s values. People need to be willing and able to explain what they are doing when asked why they are doing something.

In any domain, people have acquired knowledge in school, from the media, and from friends, said Fischhoff. The challenge for communicators is to find out what people already know and how they can be informed in useful directions. If people are motivated and respected,

they can understand a lot. “I think you’re often surprised how far you can take people.”

Policy arguments should be about thresholds for action, said Fischhoff, so as to identify value judgments, thereby maintaining the credibility of science by focusing it on the evidence. If pressed, scientists can show how evidence and values might be integrated in statements such as, “this is what I would do if I were in your situation.” In many cases, that may not even be necessary, added Budescu, if the information and uncertainties surrounding a decision are made clear.

People have different levels of risk aversion, Fischhoff noted in response to a question. But emphasizing these differences can move the onus from the expert who should be explaining the uncertainties or the alternative options to individuals who are not necessarily well enough informed to make decisions. Budescu added that risk aversion is more domain specific than many people assume, so it cannot necessarily be applied globally to perceptions of risk.

Examples of Success

When asked for specific examples of successful approaches to communicating uncertainty, Finkel cited the job the Environmental Protection Agency has been doing on the health effects of air pollution. The agency has been careful about trying to separate the statistical uncertainties from individual variability. It also has prepared documents for decision makers that express the center of gravity and where the extremes are rather than subsuming different types of uncertainty into a single measure.

Fischhoff cited good television weather forecasters. They know their audience, have gotten feedback on their presentations, and know how the weather plays out in people’s lives.

Budescu urged the use of multiple methods to convey uncertainty, including visual depictions and text. People should have a choice of methods or be able to experience multiple methods simultaneously.

SOCIAL NETWORKS

The analysis of social networks preceded the development of electronic social networking, said Noshir Contractor, Jane S. and William J. White Professor of Behavioral Sciences at Northwestern University. But several factors have come together that make it possible to understand and use networks in new ways. The social sciences have made substantial progress in understanding why people connect with others. New analytic methods for analyzing network dynamics and confirming hypotheses have become available. The use of “big data” makes it possible to analyze

very large networks connecting large numbers of people through different kinds of technologies and platforms. And the computational infrastructure now exists to develop hypotheses and conduct analyses.

Changing Behaviors

The challenge in science communication is not to find better ways of communicating facts to people, Contractor said. Overwhelming evidence indicates that people hold onto their attitudes and behaviors despite, not because of, facts. Other approaches, therefore, are needed to change behaviors.

One is simply to ask people to do something. Another is to tell people that they must do something or they will incur penalties. A third approach is to create incentives to do something. However, neither penalties nor rewards are guaranteed to create a system or culture where people routinely engage in a desired behavior, Contractor said.

The literature on social influence suggests more effective approaches to change behaviors. For example, the psychologist Robert Cialdini has laid out six key principles of social influence:

1.  Reciprocity: People tend to return a favor.

2.  Commitment and consistency: Once people have made a decision, they tend to stick with it.

3.  Social proof: People tend to conform and do what other people are doing.

4.  Authority: People tend to obey authority figures, regardless of the situation.

5.  Liking: People are easily persuaded by people that they like.

6.  Scarcity: Perceived scarcity generates demand.

All of these strategies can be helpful, said Contractor, but they are general strategies, and scaling them up can be a major challenge.

The Who of Social Influence

Scaling up science communication to reach large numbers of people requires leveraging three types of knowledge, Contractor said:

•  Science about how social influence strategies can be effective,

•  Science about who the touch points are in networks, and

•  Science about strategic choices involving social media.

Focusing on the “who” of social influence rather than the “how” provides particular promise for scaling up science communication. For example, researchers have shown that ideas can be introduced by the mass media but spread to larger publics via opinion leaders. In addition, people are socially influenced by the people they know and trust when forming an opinion or engaging in a behavior. These observations become even more crucial as new social media platforms add to the firehose of information people receive.

The field of social network analysis has sought to identify trusted opinion leaders who can help disseminate information. For example, one aphorism describing social networks, Contractor said, is “it’s not what you know, it’s who you know.” This description can seem disparaging, but success often depends on both what someone knows and who that person knows.

A second aphorism, which describes cognitive social networks, is “it’s not who you know, it’s who they think you know.” People act on the basis of their perceptions, so the perception of being part of a social network can spur action.

A third aphorism, which describes knowledge networks, is “it’s not who you know, it’s what they think you know.” People often act on the basis of stereotypes rather than factual knowledge about other people.

The final aphorism, which describes cognitive knowledge networks, is “it’s not who you know, it’s what who you know knows.” In this case, social networks and knowledge networks are merged.

People use all four of these ideas of networks every day, Contractor said, though some people leverage particular networks more than other people do. Furthermore, all of these networks are becoming more complex. The result is a multidimensional network where some of the nodes are people, some are concepts, some are keywords, and some are hashtags. Communicators need to see themselves as embedded within these different contexts if they are to understand how to leverage networks.

A Strategy for Leveraging Networks for Science Communication

Contractor proposed a strategy for leveraging networks for science communication that involves three elements:

•  Discovery: “If only we knew what we know.”

•  Diagnosis: Identify the touch points that can serve as a multiplier for scale-up of scientific communication.

•  Design: Recommend strategies for selection of social influence, touch points, and social media to scale up scientific communication.

With regard to discovery, people often do not know what they know. As a result, they spend time reworking issues because it seems less expensive to reacquire knowledge than to access already acquired knowledge.

Diagnosis involves finding the key people in a network and using the best strategies to influence those people. Naturally occurring networks are not always efficient or fully functional. For example, disparities in information sharing via networks can lead to pockets of information haves and have-nots, thereby increasing knowledge gaps. Also, not all nodes are created equal. Some nodes are particularly well positioned to be touch points or to serve as a network multiplier in a scale-up effort.

Design involves selection of the right social influence strategy, the right touch points, and the right social media channels to optimize the speed and coverage of communicating scientific information to publics or targeted audiences. In particular, the science of networks can help reveal which touch points are most likely to serve as multipliers for a scale-up effort.

Being Influenced by Others

In a famous experiment described in 1955, the psychologist Solomon Asch asked undergraduates to say which of three lines drawn on an index card was the same length as a single line drawn on a separate card. The task was so simple that no one got the answer wrong when they were asked the question at the beginning of the experiment. But if the students were first exposed to a series of other participants consistently choosing the wrong line, 37 percent chose that line as well.

This experiment is a classic demonstration of how people often conform to behaviors that they know are wrong, said Katherine Milkman, the James G. Campbell Assistant Professor of Operations and Information Management at the Wharton School of the University of Pennsylvania. Other research has shown that people conform even to behaviors that are only described to them. For example, if college students are told that 75 percent of their peers engage in a certain behavior, such as safe sex or moderate drinking, they are more likely to engage in that behavior themselves. As another example, messages about a household’s electricity usage that include information about the usage of neighboring households can reduce overall electricity usage over long periods of time. Even attributes as fundamental as obesity, smoking, happiness, and loneliness can be spread through social networks, Milkman said.

New social networking technologies raise intriguing questions about observance of social norms. For example, research has shown that Facebook users were more motivated to vote when they got messages

about friends of theirs who had voted. Leveraging such interactions poses both opportunities and risks for science communication, Milkman said.

Risks Posed by Social Media

As an example of the risks posed by social media, Milkman noted that social networking technologies can support herding, where people associate only with like-minded compatriots. On the flip side, herding may be used for constructive purposes. Technologies make it possible to track, for example, a snowball effect taking place on the web, so if a video has gone viral and is communicating misinformation, people can take countervailing steps to counter it.

As another example of possible risks, Milkman pointed out that efforts to help someone within a network can hurt someone who does not have as strong a connection to the network. For example, women and minorities who do not have the same strong network connections as men or majority populations are less likely to receive the same recommendations and favors.

The Future of Social Media

In the past, new communication media such as radio, the telephone, and television, each of which had sweeping effects on society, took a couple of decades to develop and deploy at scale, said Deb Roy, a tenured professor at the Massachusetts Institute of Technology and chief media scientist at Twitter. Today, the convergence of communication media and computers has made it possible to develop new communication mechanisms within weeks. Once an idea has been developed, software can quickly be written for hardware that already exists. The development of communication platforms has gone from what Roy called a solid phase to a liquid phase.

In such circumstances, envisioning the state of communication media 5 to 10 years from now is much more difficult than it was in the past. Cultural adaptation, as much as technological capability, is what determines which communication platforms spread and how fast they do so.

Changes in the Culture

In response to a question from a participant about the factors that keep scientists from using new media, Roy responded that the 140-character limit to tweets is not a serious limit, because people can always tweet again. But it is a format that works well for some kinds of communica-

tions and not for others, he said. For example, tweets can point to new scientific papers, intriguing observations, and other points of interest. As another example, the Sackler colloquium itself generated a healthy stream of tweets, Roy noted. When one of these is retweeted, it can end up in the timeline of someone unaware that the colloquium was occurring. The result is an ad hoc dynamic social network that can amplify messages and create long-lasting links.

New media may initially be seen as lightweight or trivial, but they evolve over time, Roy said. Today, scientists are using new media successfully, and new uses will continually be discovered. Contractor added that the culture has changed in institutions in the past few years, and people who know how to use these tools and use them effectively are now being rewarded. For example, more portfolios submitted to tenure and promotion committees include metrics involving new media. Among younger students and graduate students, many are adept users of Twitter and other new media.

Media Adaptation

Contractor also noted that when a new communication medium is developed, people tend to think that it will democratize access and bring back the public square. But history often indicates otherwise. For example, new media can help people connect more strongly with others like themselves, creating echo chambers for opinions. Even in science communication, new media can create an interest in talking with like-minded others. In such cases, research can point to people who can serve as bridges or brokers between groups.

Network dynamics also can vary from place to place, Contractor said. People in India use Twitter and Facebook frequently, but not for professional work. Most companies in China frown on their employees using LinkedIn because it indicates that an employee is looking for a job. Distinct norms emerge in different places and evolve over time.

Contractor added that when a new medium is developed, the existing media need to adapt if they are not to be displaced. Radio replaced the newspaper as the main place where people got their news, and radio in turn was replaced by television as the major means of news dissemination. As the communication ecosystem continues to explode, all media will need to adapt to continuing change. Furthermore, many communication media remain in what some have called permanent beta, where they continue to change and evolve.

Don’t Believe Everything You Read on the Web

During a follow-up conversation on the risks posed by new communication technologies, Contractor pointed out that echo chambers have a natural tendency to form in new networks. Research may suggest ways to open up the dialogue, even if doing so would push people out of their comfort zone.

Contractor also observed that new media make it possible to target specific audiences with specific messages to accomplish a common good, even though people are reached in different ways. Roy agreed that it is easier to deliver a specific message to targeted groups, but this can make it difficult to reach large numbers of people because it is hard to differentiate messages for so many different groups.

As another potential risk of new media, Contractor pointed to the “dark web,” where people can mobilize using new media without being publicly visible. Even Facebook has secret pages that invite large numbers of people by invitation only.

Roy acknowledged the risk of misinformation spreading through new technologies. Information on the web can be difficult to trace back to its source, reducing certainty about the veracity of that information. Better education and better tools are needed to help people make sense of what they encounter through social media. As Contractor reminded the participants, a useful reminder is the quotation “Don’t believe everything you read on the web.—Abraham Lincoln.”

Measures of Success

Regarding metrics of success, the particular people being reached may be more important than the absolute number, said Roy. In addition, the content of the messages that flow through a network can be used to measure the effectiveness of dissemination.

Tools are being built to assess who is reached, Contractor observed, but they are in their early stages. For example, two people with the same followers do not necessarily increase the distribution of a message as much as two people with different followers. Similarly, if a person’s followers have many followers themselves, that person may be a more effective disseminator of a message.

SCIENCE COMMUNICATION AS POLITICAL COMMUNICATION

Science is encountering politics more and more often, and the trend will not change anytime soon, said Dietram Scheufele, John E. Rose Professor in the College of Agricultural and Life Sciences at the University of Wisconsin-Madison. The societal applications of modern science

are inherently political issues, whether the issue is stem cells, climate change, obesity, or synthetic biology. This is the case despite the fact that many members of the public know relatively little about these issues. For example, when a sample of the public was asked whether it was true or false that the Obama administration had recently banned all research on synthetic biology, about a third were able to provide the correct answer, 12 percent thought it was true, and 55 percent did not know.

An idealized model of society holds that events occurring in the arenas of politics, science, economics, or other societal domains have a direct influence on perceptions of reality and public opinion. But very few people can observe these events directly, Scheufele noted. Most perceptions of reality are mediated, usually through the media. The information transmitted through the media is in turn selected through a process known as agenda building in which people negotiate the content of media messages. For example, corporations work to push certain content, while scientists and their institutions write press releases to get their stories covered. In this negotiation, science is only one of many voices in society.

Research has shown that media coverage of particular science issues increases when politics become involved in the issue, and media coverage spikes when an issue becomes controversial. Thus, science tends to get covered by the media when politics become involved. Furthermore, people tend to remember and use information that they get from the media—a phenomenon known as media priming. If an issue is ignored by the media or by the people the media is covering, it will not become salient in public perceptions.

Framing

Information derived from science is often ambiguous, Scheufele observed. Carbon nanotubes may cause cancer because they behave similarly to asbestos fibers, but carbon nanotubes are also important components of many types of materials and equipment that may allow for the early detection of cancer. For a lay audience with no training in nanotechnology, this is an ambiguous stimulus that could be interpreted one way or the other.

All perception depends on the context, especially for ambiguous stimuli. The framing of information therefore shapes how people think about that information. Sometimes framing has a partisan motivation, but in most cases it is simply a tool for information processing to help people determine why an issue is important and how to think about it. Framing reduces ambiguity by contextualizing information, and it is most successful if it resonates with an underlying schema.

There is no such thing as an unframed message, Scheufele said. Even in an entirely professional setting, such as a scientist’s grant proposal to the National Science Foundation (NSF), the use of framing is inevitable.

Making Sense of Science

Researchers have developed several models of how public opinion is formed, both at the individual level and at the societal level. For example, the expression of opinions can vary depending on individual characteristics. Males are more likely to speak out in controversial situations than females, and young people are more likely to speak out than older ones. But expressing opinions also depends on the social environment. If a particular opinion is not favored in a society, then people are less likely to express that opinion, which in turn leads other people not to express that opinion. The result is a spiraling process in which particular opinions can become dominant because people in the minority are much less likely to express their views.

Social norms campaigns use the same logic to shape opinions. For example, when hotel guests see a sign saying “75 percent of the guests who stayed in this room reused towels,” they are more likely to reuse their towels than if they see a sign saying simply “75 percent of guests reused towels” or “You can show your respect for nature and help save the environment by reusing your towels.”

Motivated Reasoning

People process information based on their beliefs, identities, and ideologies. Studies of this process of motivated reasoning are not new but have seen a recent renaissance, said Scheufele. Motivated reasoning functions both through selective exposure to information and through the interpretation of that information. When people firmly believe something, they are more likely to seek out new information that conforms with that belief. They also are more likely to question information that does not fit with that belief. However, today’s targeted media environment is making it more difficult to be exposed to debates and the other side of the issue. Even newspapers soon could be customized to give people only information they want.

Ambivalence about a topic makes if more likely that people will engage in both sides of an argument. Also, if people are in groups that disagree with them so that they have to justify their opinions, they are more likely to process information carefully.

As a final example of this research, Scheufele pointed to work in political communication on why having a heterogeneous network that exposes

a person to different opinions is correlated with having more factual political knowledge and greater public participation. “Disagreement is good for us for a whole variety of reasons and actually forces us to think through some of these things more carefully.”

Next Steps

Separating the social context from the science is critical for the scientific endeavor but is dysfunctional for science communication, Scheufele noted in closing. Whether scientists like it or not, new science will be debated in a complex media environment where individual predispositions and societal expectations loom large.

In such an environment, systematic efforts are needed to increase citizens’ ability to find scientific information in increasingly fragmented media environments, connect science to their daily lives, and process information accurately. It is not just about getting information to citizens but about helping them get it right, Scheufele said. There are no easy answers for how to do all of these things, but the questions can be addressed through empirical research.

Problems with Science Communication Research

Science communication research is not necessarily a cumulative scientific enterprise that gives practical guidance on difficult issues, said Patrick Sturgis, professor of research methodology from the University of Southampton. It is better characterized as a loose assemblage of interdisciplinary frameworks and approaches. As a result, it will not necessarily provide easy off-the-shelf solutions to the long-term problems of science communication.

Sturgis described several of the key problems associated with science communication. One problem is that small variations in the wording of questions make big differences in the answers obtained. The wording of questions can shift apparent public opinion from a minority to a majority position, which is “worrying if what we think we’re doing is measuring something real and concrete.”

Another problem is that people are willing to provide opinions on nonexistent issues. Polls demonstrate that many Americans are in favor of the monetary control bill, the agricultural trade act, and so on, but these pieces of legislation do not actually exist. In Britain, people were even more strongly in favor of nonexistent bits of legislation, Sturgis noted. If people are willing to offer opinions about nonexistent issues, it must lead to questions about the robustness of opinions measured on genuine questions of public policy.

There, problems are particularly acute in the area of science and technology because many people have heard little or nothing about key areas of research and application. A 2012 Wellcome Trust survey of adults in Britain, for example, found that over half of the population had never heard of the term human genome.

Science communication research has a valuable space in describing the shape of public understanding and preferences, but it is on shakier ground when it comes to explanatory accounts. Because most data are observational and nonexperimental, multiple accounts of how they arose are often equally plausible. For example, accounts of motivated reasoning can be equally well explained through an “enlightened preference” framework. A recent study found that greater knowledge about genetics does not necessarily lead to greater approval of the use of genetically modified crops. It does for people on the left of the political spectrum, but for people on the right side, greater knowledge leads to less approval for the use of such crops. This can be interpreted as a case of enlightened preference rather than motivated reasoning, in which an increase of knowledge enables people to connect their core values to their policy preferences. More generally, it points to the difficulty in making causal inferences from observational data.

Despite this somewhat pessimistic perspective for the ability of science communication to deliver, Sturgis concluded by pointing to the “huge amount of benefit” that can be derived from the interdisciplinary field of science communication. However, social scientists need to avoid overpromising what the field can deliver, a trap that bench scientists are often accused of falling into themselves. They also need to avoid the implicit promise that they can provide insights that will enable scientists to get the public on their side for the latest favored technology.

Microtargeting and Counterframing

Kathleen Hall Jamieson, Elizabeth Ware Packard Professor of Communication at the University of Pennsylvania’s Annenberg School for Communication, made four points in commenting on Scheufele’s presentation. First, microtargeting—repeating a tailored message through multiple channels, including social media channels, to a small target audience—is becoming increasingly common. For example, during the 2012 presidential campaign, Republicans targeted groups in coal states to receive targeted messages about “the Obama war on coal.” This approach can have a powerful influence on targeted groups by creating coherent arguments and supporting strands of evidence. At the same time, microtargeting works against the kinds of exchanges across ideology or perspectives that can expose people to new information. For example, microtargeting

strategies rarely raise the microtargeted message to a higher level because doing so would make it susceptible of being identified as inconsistent.

The second point Jamieson made is that presidential debates, in contrast to microtargeting, function well in American society by helping people learn about the candidates and issues. They feature both sides of the campaign and are watched by large numbers of people. As a result, they promote postdebate discussions that at least have the potential to intersect with non-like-minded networks. Even if they may not be persuaded, people are at least going to hear the other side. The success of presidential debates prompts the question of how to create more mass exposure experiences featuring exchanges about the things that matter to public policy.

Jamieson’s third point is that framing does not have much of an influence when counterframing also exists. Most experiments that show framing to be powerful do not include counterframing. Where counterframing exists, as in highly partisan media environments with high levels of microtargeting, framing’s influence is reduced.

Jamieson’s fourth point involved the credence given to minorities that resist majority opinions. Historical examples such as Galileo indicate that minority resistors can be correct. As a result, when people argue aggressively and persistently for a position, they tend to be taken seriously.

Changes Due to Social Media

The potential of social media to change science communication was a prominent topic of discussion during the question-and-answer period. Scheufele said that social media and other online media can give scientists a voice they have not had before by enabling them to present issues in the ways that they desire. But messages disseminated through social media can have a widespread effect only if they can take advantage of a multiplier effect.

Jamieson added that the linking and alerting functions of social media potentially can be used to motivate individuals to gain deeper knowledge about an issue, including greater ability to argue and counterargue an issue. This is a very rich area for scholarly pursuit, she said.

Sturgis, however, was pessimistic about social media’s ability to better connect scientists to parts of the public that are not usually reached. People select into social media channels, and if they are not interested in information they tend to be put off by it. Social media have extraordinary potential for communication among scientists, “but I very strongly doubt that it will open up a new channel of unmediated communication between scientists and lay public,” he said. Sturgis also noted that the entire public,

not just the users of social media, need a voice on matters that are consequential for public policy.

Social media opinions are not proxies for public opinion, Scheufele agreed. But social media can point to emerging themes, which in turn can suggest ways to start conversations and promote engagement.

From Mass to Micro Media

In many ways, said Scheufele, the idea of a mass medium is declining. The media environment is much more diverse than in the past, and people have much more access to this diverse environment. The problem is that people do not have a motivation to seek out information that does not accord with their preconceptions. Even on social networking sites, people are far more likely to have friends with similar views than different views.

But Jamieson countered that the mass media era is not yet over. Even if a transition is under way to other sources of information, much of the information on the Internet originates in a mass media channel of some sort.

With any medium, people need to be able to engage in a search for accuracy, Scheufele observed. Part of the motivation for this search comes from being socially accountable. Part comes from a willingness to acquire information from heterogeneous networks. Rather than providing more information to counter misinformation, can people be taught how to make a more effective cognitive investment in assessing the accuracy of information, so that they can make more enlightened choices even if they do not fully understand an issue?

The Limits of Public Opinion Polls

During the question-and-answer session, Jamieson also discussed some of the problems with polls of public opinion. They often force people to make choices when in fact their opinions are unformed. The survey research system needs to spend more time figuring out what people actually know, which would indicate what they need to learn to enter into a discussion.

Reporters used to report polls very uncritically, Jamieson pointed out. Now their stories often include such information as margins of error, response rates, and the existence of framing effects. This is an example of how the education of journalists can help protect the integrity of journalism.

LESSONS LEARNED

Fischhoff concluded the first day of the colloquium by identifying four points that he considered the highlights of the day.

First, the science of science communication has ways of better understanding other people. Social scientists have produced tools to learn about individual interactions. These tools make it possible to avoid thinking that public opinion is refractory to any sort of external criticism. As Susan Fiske said earlier in the day, Fischhoff noted, people are not idiots. The science of science communication can help figure out what they are thinking and doing. In the process, scientists and science communicators can reflect on themselves and broaden their own perspectives.

Second, science communications can be improved. Even though there are no magic bullets, a base of knowledge exists that scientists and science communicators can use to avoid things known not to work and to use approaches that have a better chance of working. Outcomes then can be monitored to increase the rate of success.

Third, changes in social and intellectual organizations are needed to improve science communications. For example, integrated multidisciplinary teams are needed so that people can work together on complex multidisciplinary problems, which will require that institutions support scientists to do this kind of work.

Finally, scientists and science communicators themselves need to be willing to create the new forms of interaction needed to solve the problems they face. They need to engage in the kinds of research that will bring disciplines together and avoid fragmented and potentially misleading efforts.

This page intentionally left blank.