How messages about behavioral genetics research can impact on genetic attribution beliefs

Introduction

How do laypeople—as opposed to experts—understand human genetics? This subject has been approached from a diversity of angles. For example, one stream of research focuses on clinical contexts to assess public understanding in situations where, for example, patients are informed about their own genetic predispositions or the predispositions of their relatives^1–3. Another stream of research examines how attributing certain conditions such as schizophrenia, obesity or sexual orientation to genetic influence correlates with or impacts on stereotyping attitudes^4–8. Moreover, a group of studies tests how different types of claims about the existence or non-existence of significant genetic differences between genders or ethnic groups impact on individual perceptions and intergroup relations^9–11.

My research contributes to closely-related literature on how people develop, reinforce or update their beliefs about the influence of genetics on various human traits. What is it that makes people think that some traits are more or less affected by genetic predispositions? The existing literature points to different sources of influence. Research in psychology indicates that basic intuition about the influence of biology on physical and social traits is rooted in childhood^12,13. Some research also verifies whether participation in an introductory biology class or social science programs influences students’ views on the role of genetics^14–16. The present study speaks to a third branch of research that investigates the roles played by the news media.

The contribution of this paper is threefold. It begins with a review of fascinating literature on how human genetics is disseminated in the media and how this type of information could influence beliefs and world views. This literature is produced by intellectuals and social scientists from a variety of disciplines such as sociology, ethics, psychology, and communication studies. My review highlights the milestones in the evolution of this branch of research. With this map in hand, I then present the theoretical framework for my research program, providing a logical basis for how messages about human genetics may influence people’s beliefs about the role of genes in causing human traits. The central argument is that people’s reactions vary depending on whether the presented information is from medical/physical genetics or behavioral genetics research findings. The main hypothesis that can be derived from this framework is the genetic interpolation hypothesis, which predicts that messages about how genes impact on specific behavior can lead the public to infer greater genetic causation for social traits that are not even mentioned in the content of the message.

The main strength of my framework is that it offers clear, testable predictions. However, it is not without limitations and some questions remain unanswered. One of these questions is: what kind of message formats can succeed at altering people’s views about the influence of genes on human beings? The third contribution of this paper is to begin to address this question empirically. I present the results of a survey experiment that was designed to test whether a simple, short paragraph about behavioral genetics is a powerful enough stimuli to cause the genetic interpolation effect.

Genetics in news media: A literature review

During the ’90s and early 2000s, an important group of intellectuals expressed concern about how genetics research is communicated to the public^17–21. The blame was divided between three culprits: overenthusiastic scientists, fantasizing pop culture, and misreporting news media. The news media were also charged for reporting inflated optimism about genetics research, over-emphasizing benefits and ignoring or downplaying risks. But the media was mainly criticized for how they described the influence of genetics as deterministic, that is, as if genetics were the only, or at least, the most important factor at play. It was argued that, while this depiction might be accurate for a few distinct physical traits or rare mono-genetic diseases, such a deterministic outlook is largely misleading for describing the influence of genetics on most human traits, especially social traits like sexual orientation, intelligence, personality traits, or other complex social behaviors.

Among academics, this critique found both supporters and skeptics. The debate was partly fueled by the fact that most of the evidence gathered to support these claims rested on anecdotes and case studies. A number of social scientists decided to contribute to this debate by systematically analyzing the contents of media coverage. In a paper published in 1998, Condit, Ofulue and Sheedy²² test whether, as argued by critics, the advent of molecular genetics is associated with an increase in genetic determinism in news coverage. Their results refuted expectations, indicating that there had been a small decrease in deterministic framing during the 1985–1995 decade compared to previous periods. However, the study also reveals that in approximately a third of news article, causation is attributed solely to genes. Regarding this finding, the authors preferred not to adopt a normative stance: “Whether this is an appropriate balance or whether it instead represents excessive determinism will depend on one’s own view of the role of genes.” (p.981)

This quote emphasizes the intrinsic difficulty of evaluating the quality of news coverage: that of choosing which standards to use. Many other studies have assessed how genetics research is covered in the media. Often, the focus of these analyses varies according to the type of research presented in the news: medical genetics^23–25, psychiatric genetics²⁶, genetics of addiction^27,28, or behavioral genetics²⁹. In line with critiques’ claims, optimism is found to be prevalent, and the language used to describe the effect of genes on human traits often suggests that the role of genetics is predominant. (For an interesting exception, see 24.) However, each of these works sets its own standard for evaluating whether a claim is too optimistic or too deterministic. The challenge is to find a discriminant benchmark that can be replicated.

In their article published in 2004, Bubela and Caulfield³⁰ present an innovative approach. Their clever study design consists in gathering 627 news stories from 26 newspapers published in 4 countries. Their work compares each of these news stories with the content of the original research article on which it is based. This comparison allows the authors to assess the prevalence of exaggerated claims, and to explore which factors are associated with these claims. The authors summarize their results in a few words:

The literature studying how genetics is presented in the media is far too broad for a fair review [see also 31,32]. But the last sentence from Bubela and Caulfield’s quote points us to the second and key component of this debate: whether or not media coverage of genetics research has a significant impact on public beliefs. Indeed, from a professional perspective, science journalists’ tendency to inflate the influence of genes is a questionable practice. However, this exaggeration would be of less concern if it was found that public beliefs remained largely unaffected by the message.

This leads us to a closer review of the empirical literature on media effects, to which the present paper intends to contribute. The predominant approach has thus far consisted in testing specific hypotheses using psychology experiments where different groups of participants are randomly exposed to one of several messages about the influence of genes before being asked to report their beliefs about genetics. Provided a rigorous implementation and sufficient statistical power, this research design allows researchers to be confident that the differences that may emerge between the experimental groups following message exposure result from the influence of these messages on participants’ beliefs^33,34.

Most of the empirical work produced thus far tests whether messages about medical genetics have an impact on public beliefs about the influence of genes on human traits. The evidence gathered in the literature indicates that this type of information has limited influence. In their 2001 study involving approximately 100 college students, Condit and colleagues test whether using or not using a deterministic headline influences participants’ reading of a 500-word, non-deterministic news article about the influence of genetics on diabetes³⁵. Following message exposure, participants were asked: "How significant a role do you believe that genes play in human health generally?" The study finds no significant post-exposure difference between the groups. In their 2008 paper, Lynch and colleagues describe a study where they recruited a similar sample of participants (n=104) and conducted a quasiexperiment to test whether repeated exposure to news headlines, news summaries and video advertising about medical genetics can, among other things, increase genetic determinism as measured using a 7-item battery of questions³⁶. The results show no statistically significant increase.

These two studies converge in suggesting that participants do not mechanically endorse genetic determinism in response to messages about medical genetics. Still, it would be premature to conclude that the public remains unaffected by this type of message. Indeed, these experiments included only a small number of participants, thus offering low statistical power. Additionally, the outcome measures were designed to test if message exposure causes an overall change in views about the influence of genetics. One might expect that these two studies would fail to detect more nuanced or subtle effects.

Lynch and colleagues conclude their paper by recognizing that people likely "assign different levels of [genetic] causation to different characteristics" (p.51) and therefore, they suggest that future research should measure perceptions of genetic influence with regard to specific traits. Smerecnik’s paper, published in 2010, follows this recommendation³⁷. In the literature, the term "genetic attribution" refers to the level of importance people attribute to genetics in explaining a particular trait or group difference³⁸. Smerecnik’s study tests if messages about health genetics impact on genetic attribution. The 235 students and members of the public who participated in the experiment were randomly exposed to one of two short—approximately 140-word—health messages. One was a paragraph about how salt intake can impact blood pressure; the other was the same paragraph complemented with an indication about a gene-environment interaction: "People without a genetic predisposition to salt-sensitive blood pressure may consume salt as usual." Following message exposure, participants were asked to use a 7-point scale to report how much they believe genes impact on each of 11 health conditions (e.g., obesity, lung cancer, hypertension). Results reveal that the group exposed to the GxE interaction version of the stimuli showed a statistically significant increase in the influence they attributed to genetics in causing salt sensitivity, while the other group did not. But more crucially, participants "did not extrapolate these implications to other diseases." (p.390)

This experiment offers more than a conceptual replication of previous studies. Indeed, the results reinforce the conclusion that messages about medical genetics do not cause genetic determinism. But more importantly, Smerecnik’s research design also offers a novel, quantifiable benchmark to evaluate one potential type of interpretation bias: the risk that a message about the role of genetics in causing one characteristic could lead the audience to infer greater genetic causation for other traits that are not the object of the message. His work inspired me to apply the same benchmark in order to address a related, much discussed, but until then unexplored research question: How do members of the public react when exposed to news about behavioral genetics?

In my study, published in 2014³⁹, I divided a large convenience sample of approximately 1400 respondents into three groups. The control group, used as a baseline for comparison, was exposed to a news article about medical genetics, more precisely about the recent discovery of a key gene involved in the development of breast cancer. The second group was exposed to a news article claiming that scientists have found a gene that causes liberal ideology. The third group was presented with a news story claiming that credit card debt is linked to a particular gene. All three stimuli were real news articles published in well-known newspapers or magazines from the United States, Canada, and the United Kingdom. Immediately following news exposure, participants were asked to report how much influence they attribute to genetics in explaining 14 body, medical, psychological and behavioral traits on an 11-point scale ranging from 0% to 100%.

Responses from the control group reveal that the average genetic attribution for political ideology and credit card debt is low, 10.6% and 7.2% respectively. Comparison with the two other groups shows statistically significant differences in genetic attribution for the specific trait described in the news article. For those exposed to the story about a liberal gene, genetic attribution for ideology doubled to 21.5%; for those exposed to the story about the debt gene, genetic attribution for credit card debt showed a threefold increase, 24.2%.

Therefore, as was the case for salt sensitivity in Smerecnik’s experiment, results show that participants tend to accept the specific genetic argument presented to them, and to adjust their genetic attributions accordingly. However, in contrast to the salt experiment, my study reveals the presence of a side effect. Groups exposed to behavioral genetics report higher genetic attribution for other traits not mentioned in the new articles. The size of these increases in genetic attribution may seem relatively small, ranging from approximately 3 to 8 percentage points on the 0% to 100% response scale. But what is more striking is that traces of this side effect were found in genetic attribution for most, if not every other social trait: intelligence, obesity, gambling addiction, violence, alcoholism, mathematical ability, and sexual orientation. Interestingly, exposure to news about behavioral genetics has no significant additional effect on genetic attribution for height, skin color and skin cancer.

This study offers support for some of the concerns expressed by intellectuals during the ’90s and early 2000s: news about human genetics can impact on people’s beliefs. Yet, this media effect is much more subtle than anticipated. Firstly, interpreting my research in light of the literature suggests that news articles impact views about the influence of genetics when they cover research on behavioral genetics, but not when they cover research on medical genetics. And secondly, this impact is not an instant endorsement of genetic determinism, but a marginal upward adjustment in genetic attribution for social traits.

While my 2014 paper offers an important empirical contribution, the issue of why people react the way they do was only superficially addressed, due to space limitations. The next section presents a theoretical framework for cognitive mechanisms that could account for why people generalize particular findings from behavioral genetics to other social traits.

The genetic interpolation hypothesis: A theoretical framework

Psychological essentialism can be observed when people tend to associate the characteristics of an individual, group or phenomenon to its "deep-seated hidden essence" [40, p.4]. There exist different types of essentialism. For example, in many religions or sects, prophets are believed to be the embodied manifestation of divine intervention. This belief is a manifestation of spiritual essentialism. Another example is nationalist essentialism, where a particular nation is, in its history, ethnicity and culture, seen as fundamentally unique and therefore different from other nations.

The present framework is concerned with a particular type of essentialism. Genetic essentialism is a world view that holds a particular set of assumptions about the origins and characteristics of human traits⁴¹. In its most drastic form, often referred to as genetic determinism, this world view perceives most of the features that define human beings and their interactions as innate, shared by family members, immutable, predetermined and therefore independent of one’s own will. While very few people adhere to this extreme view, research shows that there is substantial disagreement among members of the public regarding the extent of genes’ impact on humankind⁴². From this perspective, genetic essentialism should be perceived as a continuous dimension bounded by a bottom limit – genes have no influence whatsoever – and a top limit – genes explain everything – with most people positioning themselves somewhere in between. Research also indicates that genetic essentialism may coexist with some world views or ideologies (e.g., creationism) and conflict with others (e.g., constructionism)⁴².

While the social implications of genetic essentialism have been the object of much scholarly attention^43–45, relatively little is known about the reasons why some people come to endorse or reject this world view, partially or completely. Of course, some dimensions of genetic essentialism refer to properties that are partly observable in the real world (e.g., whether a trait is detectable in infancy, shared among relatives, and stable over time). It seems plausible that people use these features as proxies to assess the influence of biology and genetics¹². Members of the public may also form their beliefs by relying on knowledge acquired in college as part of the biology or social science curriculum¹⁶. The theoretical framework described in the paragraphs below leads me to predict that another aspect of their environment, namely media coverage of behavioral genetics, can also affect people’s beliefs about the influence of genetics on human beings.

Defining whether and to what extent human traits are caused by genetics requires rigorous scientific investigation, and involves significant expertise and resources. Very few people will ever have the opportunity to empirically measure the real influence of genes. Still, this limitation does not prevent members of the public from forming their own beliefs and impressions on the matter. In many studies, participants were asked to estimate the extent of the impact of genetics on various human traits^39,46–48. A consistent pattern emerged. People tend to consider genetics to be a dominant influence on lasting and stable traits like physical characteristics, chronic diseases or persistent health conditions. Genes are perceived to play a weaker but still significant role in explaining various psychological traits like mental abilities and personality traits. Participants also believe genetics play a moderate or weak role in explaining impulsive or addictive behavior such as violence or gambling addiction. Finally, genetics attribution is very small or null for complex social preferences or behavior such as voter turnout or brand preferences.

To interpret this relationship, I mobilize the notion of schema used in cognitive and educational psychology. This notion was initially developed in the first half of the 20th century^49,50 and later stimulated major theoretical innovations^51,52. However, the literature it inspired has been criticized for its contradictions, ambiguity and overstatements. It is therefore crucial to clarify the meaning I intend to invoke.

Schemata–plural for schema–are a form of knowledge that involves the organization of related mental objects into a coherent structure. The logical structure underlying this organization may take various forms, such as categories, geographical locations, hierarchy, chronology or associations, to name a few. The logic of a schema offers a representation that connects mental objects together and helps to make sense of these objects. One example of a schema is a timeline representing various events organized logically in chronological order of occurrence. This information can be presented visually using a material or numeric support. But people may also store the chronological order that connects events in their memory. In this way, information about the chronology of events (mental objects) is organized in their mind and remains accessible even when the visual representation is not available. This coherent structure of mental objects is an example of a schema. We use schemata in everyday life. It has been argued that organizing mental objects into schemata facilitates the encoding of information in memory during learning, and eases memory recollection afterwards⁵³.

The schema theory posits that preexisting schemata are consciously or unconsciously activated when people try to interpret new pieces of information. Rumelhart compares the role of schemata in everyday life with the role of theories in science⁵²:

People’s assessments of genetic influence are not precise estimates but informed guesses. In my view, the concept of schema is reasonably well-suited to the genetic attribution pattern described earlier. The structure of this schema is organized such that the more a trait is believed to be influenced by biology—as deduced from experience, observable features or as learned from other sources—the more genetics plays a predominant role. I use the term genetic attribution schema to refer to this belief structure. I further argue that people are more confident about their guesses when it comes to assessing the influence of genetics on traits that are either very close to biology or very far from it. This assumption is partly derived from the fact that one of the features of genetic attribution is that it is constrained by a ceiling value (completely genetic) and a floor value (not genetic at all). Figure 1, Panel A, visually illustrates the genetic attribution schema.

Figure 1. Visual representation of the theoretical framework.

What happens if a person comes across information about a case or event that presents characteristics that are inconsistent with the schema in which it should normally fit? The exact answer varies quite a bit from one person to another, and it also varies depending on the particular schema at stake. Rumelhart discusses three broad possible categories of reactions⁵². In a first scenario, the individual interprets this new information as an exception, a rare case that deviates from the general rule. Here, incoherent information is discarded and original schema remains intact. In a second scenario, the individual recognizes the new information as true or at least plausible, and adapts the original schema at its margins. This amended schema offers a way to interpret the new information without compromising the schema’s fundamental logic. In a third scenario, the original schema is discredited. The individual either finds an alternative schema compatible with the new information, or accepts to live without a coherent schema to make sense of the phenomenon at stake.

I argue that some of the findings derived from behavioral genetics research and disseminated in the news media are inconsistent with the genetic attribution schema. This is especially the case when research suggests that genes play a non-negligible role in causing complex social behavior such as credit card debt, generosity, friendship or voting preference^54–56. Indeed, the genetic attribution schema posits that genetics has very little to no influence on these kinds of characteristics. Some people likely consider these findings to be exceptions to the rule, or may simply discredit the information presented to them. However, provided a credible source of information, such as a scientific study or a respected news outlet, other people will accept the scientific argument presented to them. Consequently, they will try to adapt their genetic attribution schema so that the new piece of evidence can fit into it. One way to do so is to raise the floor level of genetic attribution and to conclude that genetics has at least a small but non-null influence on almost every human trait. This change at the margin allows the new information to fit into the schema without compromising its basic structure: the negative association between genetic attribution and perceived distance from biology.

From this perspective, one can anticipate four consequences at the aggregate level. First, exposure to a specific finding from behavioral genetics will cause an increase in average genetic attribution for the particular complex social trait on which the message is focused. I refer to this as the persuasion hypothesis, since this effect reflects an acceptance of the specific claim presented in the message (see Panel B from Figure 1).

Two other hypotheses can be derived by mobilizing evidence from well-established works in the field of cognitive psychology. These works investigate the strategies and shortcuts humans beings employ in everyday life when they have to make guesses in situations of imperfect information^57,58. As Epley and Gilovich write, “one way to make judgments under uncertainty is to anchor on information that comes to mind and adjust until a plausible estimate is reached.” [59, p. 311] The anchoring and adjustment heuristic is commonly used in tasks that require guessing the characteristics of an element. It consists in searching in memory or nearby environment for other elements that share similarities with the element at the center of the task, and to use them as anchors for comparison. The individual then proceeds by estimating how and to what extent the element under study differs from or is comparable to the anchor elements, and adjusts their final estimate accordingly.

For example, a question in a history exam may ask what year the Geneva Conventions were ratified. A student may recall that these international treaties were written to define the basic rights of wartime prisoners in response to abuses witnessed during the Second World War. The student may infer that the convention was signed after the end of the war (anchoring), following lengthly diplomatic negotiations (adjustment), and write 1951 as a best guess (actual answer: 1949). The broad literature on anchoring and adjustment heuristics has shown that people use this shortcut in a variety of everyday situations⁶⁰, but it can also be deployed when making guesses about scientific phenomena (e.g., estimating the freezing point of vodka⁶¹).

The task of estimating the influence of genetics involves a great degree of uncertainty. In this context, people will use the anchoring adjustment heuristic to reduce their efforts. I argue that the information presented in conventional news stories about how genes influence a particular complex social behavior provides people with an anchor information. This anchor indicates that the influence of genetics on this complex behavior is not null, and in fact is significant enough for scientists to investigate it. This anchor can later be used when estimating the influence of genetics on other complex social traits that are perceived as being at a similar distance from biology as the particular social trait covered in the message. Accordingly, exposure to the news story will cause an increase in average genetic attribution for other complex social traits. I call this second prediction the similitude hypothesis (see Panel C in Figure 1).

Finally, the anchoring and adjustment heuristic can also be used when estimating genetic attribution for traits that are seen as moderately distant from biology, such as mental abilities, sexual orientation, or vulnerability to addiction. As Figure 1, Panel A illustrates, I argue that people are even more ambivalent about the extent to which genetics impact on these traits than they are for complex social traits, on one hand, or medical and physical traits, on the other. When estimating the influence of genetics on traits that are moderately distant from biology, people thus think of traits about which they are more confident. They then use these traits as anchor points, and adjust their estimate depending on how close to or far from biology they believe the trait is.

Yet, suppose that some of these people are exposed to information about behavioral genetics research and, as a result, now believe that genetics does have a non-negligible impact on complex social traits. Implementing this calibration at the margin of the genetic attribution schema implies increasing genetic attribution values used as anchor points for estimating the influence of genetics on other traits. As illustrated in Figure 1, Panel D, this will cause people to increase their genetic attribution for social traits that are moderately distant from biology. I refer to this collateral impact as the genetic interpolation hypothesis, in reference to the mathematical operation which consists in estimating unknown mid-range values that are located between known values.

To summarize, some people react to news about behavioral genetics by thinking that "if genetics is strong enough to have a significant influence on complex social trait X or Y presented in the news story, then the general role of genes in causing other social traits must be stronger than I had originally imagined". This reasoning leads them to infer greater genetic influence for various social characteristics. Therefore, compared to the original belief system, the resulting one presumes that genetics has, on average, a greater influence on human beings. This reasoning provides a basis for one last hypothesis: the dissemination of news about behavioral genetics can cause greater endorsement of genetic essentialism as a world view. The logic behind this hypothesis is the same as for the genetic interpolation hypothesis. People will read a message about behavioral genetics and think to themselves: "If genetics is strong enough to have a significant influence on X or Y, then its the general role in causing racial differences, gender differences or family similarities must be stronger than I thought." To be clear, we should not expect that exposure to one message or news story will lead people to instantly embrace genetic determinism. But this stimulus will effectively leave the individual with the impression that genetics is more powerful than they initially thought.

In addition to offering a rationale for predicting how people react to behavioral genetics, this theoretical framework accounts for the null finding revealed in previous studies on how people react to medical genetics. Messages claiming that genetic predispositions influence a medical or biological condition, such a salt sensitivity in Smerecnik’s study, convey information that easily fits into people’s preexisting belief structure. Indeed, the genetic attribution schema already presumes that these traits are significantly influenced by genetics. While people may adjust their specific genetic attribution for salt sensitivity upward as a response to message exposure, this change does not challenge the general logic of this schema, which therefore remains unaffected.

Charting the limits of the genetic interpolation effect

There are many ways in which the public can learn about behavioral genetics, and it seems reasonable to expect that some stimuli will be more persuasive and will have a stronger influence than others. For example, when reading a whole book about behavioral genetics or participating in a university lecture focused on this topic, people may become aware of many research studies supporting the argument that genetics has a non-negligible impact on social characteristics. Also, such activities take hours to complete and require a significant level of intellectual involvement. For these reasons, books and lessons about behavioral genetics can be considered strong stimuli. In this context, we could expect that a great number of people will accept the behavioral genetics argument and its implications, and will thus perform genetic interpolation. Moreover, as I have shown in my previous study³⁹, news about behavioral genetics can cause genetic interpolation. However, compared to strong treatments like lessons or books, a news story or a magazine article requires less time and presents less evidence. From this perspective, this type of media material appears more like moderate stimuli.

But there also exists other arguably weaker stimuli. In journalism, a "nutshell paragraph," sometimes called "nutgraph," refers to a short paragraph that summarizes the main point of a news story and explains why it is newsworthy. Traditionally, the nutgraph appears early in a news article, after the lead⁶². Yet, in some contexts, this type of news summary is the only information the public is exposed to. This may be the case for users of news feed applications, such as those designed by many news providers (e.g., BBC, CNN, Al Jazeera), where users can see lists presenting article titles along with a summary sentence and a URL link leading to full-length news articles. Additionally, many websites and tabloids now present daily news using a "news in brief" format, basically publishing nutgraphs without offering access to additional information. This short format fits with the time constraints of the free metropolitan newspapers’ target audience, where the common reader only takes a few minutes to browse through the news while using public transportation. Also, since an increasing number of customers read news using smart phones or tablets, it should be no surprise to see the emergence of applications specializing in this kind of news format (see for instance brief.news or newser.com).

Many findings derived from behavioral genetics research are eye-catching, and their main conclusions are often oversimplified though exaggerated claims. For these reasons, it would not be surprising to see these findings being summarized and disseminated to the public using the nutgraph news format. These short stimuli offer a treatment that is likely even softer than conventional news articles. However, tabloids and news apps have the potential to reach a large audience, certainly much larger than the audience of books, academic lessons, specialized magazines or broadsheet newspapers. For this reason, it is crucial to assess how non-experts react when exposed to this information. The third contribution of this article is to test whether a short paragraph about behavioral genetics research suffices to cause the genetic interpolation effect.

I designed my experiment with three goals in mind. As seen earlier in the literature review, Smerecnik’s study³⁷ suggests that a soft stimulus pointing to the influence of genetics in causing a medical condition does not lead people to increase genetic attribution for other medical traits. Firstly, my experiment attempts to replicate this null-finding, this time deploying greater statistical power and using a paragraph about how genes impact on physical characteristics. Secondly, the experiment verifies if being exposed to a short paragraph about how genes can influence a complex social trait is sufficient to cause the genetic interpolation effect. Finally, one interesting aspect of the news coverage is that some news stories are clear about the fact that the role of genetics is not deterministic, while others remain ambivalent. The third objective of my study is to explore whether clarifying that genes have a probabilistic influence impacts on the way people react to the stimulus.

Discovering that a single paragraph about behavioral genetics can cause genetic interpolation would have at least two important implications. Firstly, this finding would reinforce the position held by many intellectuals and social scientists who are critical about media coverage of human genetics, and express concern about how this coverage might impact on public understanding of genetics. Finding that even a soft stimulus on behavioral genetics conveys an image that is strong enough to reshape public understanding of genetics would offer a reason to question the relevance of disseminating behavioral genetics using such short and overly simplified communication devices. Moreover, from a psychological perspective, this conclusion would suggest that the genetic attribution schema is a belief structure that is less rigid than one might initially expect. If, however, no side effects emerge, the results would encourage future work studying the genetic interpolation effect to narrow the scope of investigation and focus efforts on testing the influence of stronger stimuli.

Methods

This study was conducted as part of the Short Study Program of Time-Sharing Experiments for the Social Sciences (TESS). TESS invites social scientists to submit their experimental protocol. If the protocol receives a positive evaluation, the experiment is funded and fielded inside a web survey on a sample of approximately 2000 American subjects along with other research projects from other researchers. To be eligible for the Short Study Program, projects must involve stimuli that take no more than 90 seconds to administer and require no more than three question items. Researchers are provided with basic sociodemographic information on each individual subject. My experiment was fielded by the firm Government for Knowledge during the month of August 2013. Further details about the sampling methodology can be found on the Open Science Framework page where the data are archived⁶³. This study leans on a final sample of 2080 subjects. No participants were excluded from the data provided by the survey firm.

Subjects were randomly assigned to one of four experimental conditions without their knowledge. Table 1 reports the stimulus for each experimental condition. Group 1 was exposed to no information and is used as our control group. Group 2 was presented with a short paragraph about the genetics of body traits. This paragraph begins by reminding subjects about the advent of the Human Genome Project (HGP), before pointing their attention to a general conclusion about the strong influence of genetics on physical conditions. "Coding the chemical components" may not be the most appropriate expression to accurately describe what researchers involved in the HGP were actually doing. This ambiguity—for which I am the only person to blame—may have caused confusion among some of the participants. Luckily, this mistake does not blur the key argument of this stimulus: that some "physical traits are strongly influenced by genetics." In contrast, the stimulus to which Group 3 was exposed argues that genes can impact on voter turnout. This influence is possible because genes impact on some of the personality traits that motivate some citizens to vote. The content of this paragraph reflects the causal framework that emerges from a number of studies in genopolitics, an interdisciplinary field that applies behavioral genetics research methods to study political behavior and orientation⁶⁴. Finally, Group 4 was given the same material as Group 3, plus a sentence presenting one of the genes involved and clarifying that its influence is probabilistic and rather small.

Following treatment exposure, subjects were asked how much they believe genetics (as opposed to the environment) impacts on three human traits: “Turning out to vote,” “intelligence,” and “natural hairstyle (curly or straight).” These items were presented in random order. Table 2 shows the 11-point response scale. This experiment included no other question items and no other treatment groups.

Figure 2 presents a Consort Diagram showing how many participants where initially contacted, how many participated in the study, and how many were assigned to each experimental group. Table 2 reports a randomization check. All four groups are very similar in terms of sociodemographic characteristics, suggesting that randomization succeeded in producing comparable groups.

Figure 2. Consort Diagram.

The final sample size for this experiment was imposed by TESS design constraints for the Short Study Program. However, the four experimental groups (see Table 3, bottom row) are similar or greater in size to those I recruited for my previous study during which subjects were exposed to real news articles. It follows that this experiment would have sufficient statistical power to detect the treatment effects of similar sizes than those I found in this earlier study (between 3 to 8 percentage points on the 0% to 100% response scale).

The analyses that follow are divided into three steps. Firstly, for each dependent variable, a one-way ANOVA will assess if there are significant differences between experimental groups. Secondly, those mean differences which are directly related to the hypotheses will be calculated using Dunnett’s test which corrects for the greater error risk induced from multiple simultaneous comparisons. These tests will exclude data from Group 4 since the aim of this group is not to test a particular hypothesis but to explore if participants react differently than those in Group 3. The third step of the analysis will explore the differences between Group 3 and Group 4 using tests that do not correct for multiple comparisons. For all the analyses presented in this paper, the statistical threshold is set to p<0.05, two-tailed.

Results

Table 4 reports the number of observations, the mean value and the standard deviation for the three dependent variables in each group. In line with the genetic attribution schema, respondents from Group 1 (the control group, without stimulus) report, on average, a higher genetic attribution for "natural hair style" than for "intelligence," and higher genetic attribution for "intelligence" than for "turning out to vote." Next, three ANOVAs are performed, one for each variable, in order to test for whether there exist significant group differences. For the item "turning out to vote," the ANOVA analysis shows that there are statistically significant differences between groups (F(3,2012) = 6.77, p < .001, η² = .010). In contrast, no group differences are found for "intelligence," (F(3,2015) = 0.68, p = .565, η² = .001). While the theoretical framework predicts no effects for the item "hair style," the ANOVA does indicate the presence of significant group differences (F(3, 2024) = 5.50, p < .001, η² = .008).

Following each ANOVA, means for Group 2 and 3 are compared to Group 1 using Dunning’s method for multiple comparisons. Table 5 organizes results by group to facilitate discussion. Hypothesis 1 predicts that subjects exposed to a paragraph about how genes can influence physical traits (Group 2) will report genetic attribution levels that are similar to those found in the control group (Group 1). The results displayed in the three lines at the top of Table 5 show support for this hypothesis. The observed group differences are small, and none of them reach statistical significance.

Hypothesis 2 focuses on the persuasion effect, and predicts that exposure to a paragraph about genopolitics (Group 3) will cause respondents to increase their genetic attribution for "turning out to vote" compared with the control group (Group 1). The first line of the bottom part of Table 5 confirms this hypothesis. In Group 3, the average genetic attribution for voting is 4.1 percentage points higher than in Group 1. While statistically significant, this persuasion effect is much weaker than the one found in my previous study, where respondents were exposed to a real article about genopolitics or genoeconomics (between 11 and 17 percentage points).

Hypothesis 3 focuses on genetic interpolation. It predicts that, in response to the paragraph about genopolitics, subjects from Group 3 will infer greater genetic causation for intelligence. The theoretical framework also predicts that the genetic interpolation effect will be smaller in size than the persuasion effect. Table 5 shows that the mean genetic attribution for intelligence in Group 3 is not statistically different from the one found in Group 1.

Finally, Hypothesis 4 predicts that exposure to the paragraph about genopolitics will not affect the average genetic attribution for natural hair style. However, the bottom line of Table 5 reveals a statistically significant negative difference for this mean comparison. Genetic attribution for natural hair style is 4.2 percentage points lower in Group 3 than in Group 1. Therefore, H4 is not supported by the data.

Group 4 was designed to explore whether clarifying that the effects of a single gene on political behavior is probabilistic in nature could change how people react to this information. The results, reported in Table 6, show that none of the genetic attribution mean values observed in Group 4 are significantly different from those of Group 3 Overall, this exploratory analysis offers no indication that this clarification changes how subjects interpret the message about genopolitics.

Discussion

One interdisciplinary branch of research investigates how people form their beliefs about the influence of genetics on human traits. A potential source of influence is the information disseminated in news media coverage of genetics research findings. In fact, many intellectuals have expressed concern about the risk that news about human genetics would lead the public to believe that genes play a dominant role in individual differences. My review of the literature shows that researchers have addressed this issue empirically by testing how people react to news about medical genetics. Results of various studies have largely failed to support original concerns and instead showed that people tend to interpret this information with caution.

While this conclusion seems to be robust, my own personal contribution to this literature, published in 2014, reanimates the debate. My study suggests that people react differently to behavioral genetics than they do to medical genetics. It showed that news stories about how genes impact on a specific behavior can lead people to generalize this finding to other social traits. The results further demonstrate that people’s reactions are not a sudden endorsement of genetic determinism. Instead, participants show a modest but systematic upward adjustment of their genetic attribution for other social orientations and behaviors. However, due to space limitations, this 2014 contribution did not elaborate extensively on the potential psychological mechanism that may account for this phenomena.

The present article is a direct follow-up to my previous study³⁹, and copes with some of its limitations. The core contribution of this article is to present an original theoretical framework that offers clear hypotheses predicting how people react when exposed to news about behavioral genetics research findings. This framework combines two theoretical elements that are well-established in the literature. The first is the concept of schema. This concept is mobilized to account for people’s pre-treatment genetic attribution belief system, and how they view a positive relationship between genetic causation in human traits, and the perceived distance of these traits from biological roots. The second concept is the anchoring and adjustment heuristics. This cognitive shortcut accounts for how individuals react when exposed to behavioral genetics.

This framework leads me to make some predictions about how people react when exposed to information suggesting that genes are powerful enough to have a non-negligible impact on complex social traits. The central prediction is the genetic interpolation hypothesis, which predicts that people will adapt their genetic attribution schema in an attempt to integrate this new piece of evidence, but will try do so without compromising the basic logic of their belief system. As a result, people will incrementally increase genetic attribution for various other social traits that are not mentioned in the information material. Furthermore, this reasoning leads me to predict that exposure to news about behavioral genetics will increase genetic essentialism as a world view.

In addition to this theoretical development, the present article also contributes to the literature by attempting to evaluate empirically whether exposure to a weak stimuli suffices to trigger the genetic interpolation effect. The relevance of this test should not be understated. Imagine that this experiment had found that even such a soft treatment could lead people to adjust their belief system. First, this finding would have suggested that the genetic attribution schema can be altered relatively easily. And second, one could likely infer that almost any other stimuli reporting about behavioral genetics would succeed at reshaping attribution beliefs.

In contrast, the results of this survey experiment suggest that a short paragraph, presented out of context, is not sufficient to cause the genetic interpolation effect. While participants do increase their genetic attribution for the particular behavior presented in the stimuli – turning out to vote – their genetic attribution for intelligence remains unaffected.

This experiment also presents a finding that my theoretical framework fails to account for. Exposure to behavioral genetics caused participants to modestly reassess downward their genetic attribution for a physical trait: natural hair style. The exact reasons for this unexpected finding remain open to speculation. One possibility, consistent with the patterns observed in Table 5, is that the treatment caused ambivalence or confusion about how genetics impact on human beings, leading some participants to select middle-range response values to express their uncertainty⁶⁵. This experiment did not include questions asking respondents how confident they are about their response, and therefore it is not possible to test whether this interpretation is accurate.

Altogether, the theoretical framework and the experiment presented in this article offer many interesting directions for future research. For instance, researchers could design new studies to address one or many of the following research questions:

Answering these questions will certainly require the mobilization of other theories relevant to the particular dimensions investigated. But in the end, we can hope that the collected evidence will help design new guidelines for communicators, be they science journalists, educators or behavioral geneticists worried about the implications of their works for society.

Data availability

The dataset and the pollster report for this study are hosted on the Open Science Framework: DOI: 10.17605/OSF.IO/2UBP2.