Abstract
The appearance of being scientific can increase persuasiveness. Even trivial cues can create such an appearance of a scientific basis. In our studies, including simple elements, such as graphs (Studies 1–2) or a chemical formula (Study 3), increased belief in a medication’s efficacy. This appears to be due to the association of such elements with science, rather than increased comprehensibility, use of visuals, or recall. Belief in science moderates the persuasive effect of graphs, such that people who have a greater belief in science are more affected by the presence of graphs (Study 2). Overall, the studies contribute to past research by demonstrating that even trivial elements can increase public persuasion despite their not truly indicating scientific expertise or objective support.
Keywords
Science communication in the popular media is a prevalent aspect of the modern age. Scientific findings are widely communicated by the media and by universities, companies, and public relations agencies by using graphs, statistics, formulas, and narratives (Dahlstrom, 2010; Gastel, 1983; Haard et al., 2004; Tufte, 2001). These elements can play a crucial role in the believability of their findings (Abelson, 1995; Fahnestock, 1998; Gross, 1990). This research investigates how nonverbal elements such as trivial statistics, formulas, or graphs might make an advertisement, product package, brochure, or news release more persuasive.
The credibility of a communication source can powerfully influence persuasiveness (Wu and Shaffer, 1987). Given the high standing of science in society, merely implying legitimate scientific standing can enhance persuasion due to the enhanced credibility of scientific sources (Miller, 2001; Weisberg et al., 2008). If a claim “looks and smells” scientific, a person may be inclined to believe it. In other words, communications may be made more convincing without any alteration in content, simply by virtue of being presented with elements associated with science. For instance, if a medical company presents claims regarding its medication, consumers may be more persuaded of the medication’s efficacy if a graph was presented along with the claim. That is, saying a medicine was twice as effective in curing a disease may be less persuasive than showing a bar chart with one bar twice as big as the other. Similarly, showing a medicine’s chemical formula might be more persuasive than simply saying the formula’s name, even to people who know nothing about chemistry.
The believability of information that appears to have a scientific basis partly emanates from the ethos of science (Segal and Richardson, 2003). For Aristotle, Ethos referred to the persuasion emanating from the character of the source of communication (Aristotle, 1926; Segal and Richardson, 2003). It encompasses (1) practical wisdom, intelligence, or expertise; (2) virtue or moral character; and (3) good will (Constantinides, 2001; Prelli, 1989). Together, these properties can have a persuasive effect. The ethos of science refers to “How an audience perceives the professional character of a scientific rhetor or group of rhetors” (Prelli, 1989).
Ethos is particularly important when people lack relevant knowledge (Miller, 2003). In cases where people cannot assess claims’ veracity, they are forced to rely on the character of the speaker (Prelli, 1989). The ethos of science, given the current prestige scientists hold, can in those cases serve as a basis for belief (Segal and Richardson, 2003). If a claim appears to be scientific, scientific credibility can be conferred on it by the perceived character of science as a conveyor of truth.
In its sociological usage, scientific ethos refers to the social norms of the scientific community (Merton, 1942; Prelli, 1989; Segal and Richardson, 2003). These norms can attest to the character of science, and confer credibility on what is perceived to be scientific. Giving the appearance of correct scientific style can signal that the source of claims is scientific, and therefore credible (Segal and Richardson, 2003). Thus, displaying characteristics of presentation style that evoke a scientific image can serve as a means of persuasion through the ethos of science.
The current research examines whether the mere presence of elements associated with science—graphs and formulas—can indeed convince people of medicinal efficacy. If graphs are present with a product claim, people would tend to believe the claim to a greater extent. This is because graphs are associated with science, and so grant a halo of scientific validity to the claims due to scientific ethos. For example, if a graph is shown along with a claim of medication efficacy, consumers would see the medication as more effective than they would in the absence of a graph. This should happen even if the graph added no information to the textual claims. Formally,
Hypothesis. People who read claims about medication accompanied by a graph would rate the medication as more effective than people who read the information without the accompanying graph.
This research extends past findings in this area by using simple information to show that the persuasive impact of peripheral cues associated with science (such as graphs) goes beyond that which could be explained by the mystique of incomprehensible information, or by the apparent but opaque informational value of such information. Both use of “Scientese” (scientific sounding words) or neuroimagery can increase belief in claim credibility (Haard et al., 2004; McCabe and Castel, 2008; Weisberg et al., 2008). However, in those cases, enhanced credibility may have emanated from inferred informativeness due to the association of such elements with science, coupled with an inability to understand them.
For instance, brain images might be incomprehensible to most people. However, people would still know that brain-scans are scientific. Accordingly, the presence of brain imagery would signal the existence of scientific support. Due to the esteem with which science is held, this inference may lead people to greater belief in the claims. In other words, brain images can be taken as a signal for a scientific basis for claims. A general belief that “science equals truth” may then lead people who see a claim accompanied by a brain image as more true than they would without the accompanying brain image. Since most people would not be able to understand the images, they would not be able to discount them as providing no real additional support to the claim.
The graphs in the current study are different from brain images or “scientese” on several fronts. They are easily comprehensible, so that their rhetoric effect cannot be the result of opaqueness which leads people to deduce credible scientific evidence. Faced with a brain-scan that they cannot understand, people may infer the presence of validating scientific evidence. Faced with a simple graph, people should be able to tell that it provides no additional convincing information.
Graphs may still increase persuasion due to their signaling value. Even understanding them and knowing that they provide no additional information, people may take their presence as a cue to the scientific nature of claims, leading to increases in persuasiveness due to scientific ethos. This might happen despite the fact that unlike brain images or obtuse scientific jargon, graphs do not in fact signal a credible, knowledgeable scientific source.
1. Study 1. Blinded me with science: Simple graphs bias drug effectiveness ratings
Methods
Participants (N = 61, 51.7% male, mean age 34.87, United States) were recruited online to complete the study in exchange for payment. Among participants who reported their level of education, 1.7% did not complete high-school, 86.69% had some college education and 60.65% had a 4-year college education or higher. Participants signed up for the study on Amazon Mechanical Turk. They completed the study for payment as part of a longer session containing multiple unrelated studies. The study was approved by the university’s institutional review board.
Participants read information about new medication which ostensibly enhances immune function and reduces the occurrence of the common cold. Participants were randomly assigned to one of two conditions. Half the participants were shown a graph (see Figure 1), and half were not. The graph provided no new information over that contained in the text. It was titled “Illness occurrence” and was sparse in information, showing separate columns for drug and control groups, with a reduction of 40% in incidence of illness between those two groups.
Graph displayed with Study 1.
Participants read,
A large pharmaceutical company has recently developed a new drug to boost peoples’ immune function. It reports that trials it conducted demonstrated a drop of forty percent (from eighty seven to forty seven percent) in occurrence of the common cold. It intends to market the new drug as soon as next winter, following FDA approval.
There was no additional information regarding the medication.
After reading the instructions, participants answered the question “How effective is the medication.” Ratings were given on a 9-point scale, anchored by not at all effective (1) and very effective (9). In other words, the left end of the scale was labeled with the words “not at all effective” next to the number 1, and the right end of the scale was labeled with the words “very effective” next to the number 9. There was no verbal anchor for the mid-point of the scale. We also asked participants to answer: “Does the medication really reduce illness?” with either a yes or a no.
Participants given graphs expressed greater belief in the claims, rating the medication as more effective (6.83 of 9) than did participants given verbal description only (6.12 of 9): t(59) = −2.1, p = .04. A chi-square test indicated that a higher percentage believed the medication would truly reduce illness for the graphs group (96.55%) than for the control group (67.74%): chi-square = 8.3, p = .004. In other words, while only two thirds of the people believed the medication would reduce illness without the graph, all but one participant in the graphs condition believed this. This provides initial evidence that graphs increase persuasion.
2. Study 2. Belief in science moderates graphs’ effectiveness
The second study aimed to replicate the effects of the first study, while shedding more light on the reasons behind the effects uncovered in Study 1. First, the study was designed to rule out the effects being due to mere repetition of information in graphic form. The effects uncovered in Study 1 could have been due to the fact that in the graphs condition information was given twice: in both verbal and graphic form. The effects found could have been due to this repetition in the graphs condition, rather than due to the presence of graphs.
Second, the study aimed to examine whether the effects of graphs were merely due to increased retention of information that is due to resulting from their visual nature. Images can constitute an effective way of communication, aiding information processing and retention (MacInnis and Price, 1987). People differ in information processing styles, and for some, images may be particularly helpful in information acquisition (Childers et al., 1985). They can serve to enhance the way meaning is conveyed (Scott, 1994). Accordingly, pictures have been shown to increase recall of persuasive information (Edell and Staelin, 1983), and support persuasion (Mitchell and Olson, 1981). The presence of graphs, then, might have aided comprehension and retention of information. If so, it is this, rather than the signaling of a scientific source, that may have enhanced belief in medication efficacy.
Finally, the study aimed to provide support for our claim that it is the perceived scientific nature of graphs that enhances persuasion. The study was meant to achieve this by showing that the effects of graphs are moderated by a general belief in science. Specifically, we meant to show that a higher belief in the validity of science enhances the persuasive effects of graphs. If the persuasiveness of graphs comes from their being a signal toward the scientific backing for a claim, and so lend credibility to a claim through their association with science, people who put more faith in science should display a stronger effect of graphs on persuasion. In other words, graphs would have a greater effect for people for whom the ethos of science confers credibility. Accordingly, we anticipated an interaction of belief in science (as a conveyor of truth) and the presence of a graph, such that the stronger one’s belief in science, the more the presence of a graph should enhance persuasion.
Methods
Participants for Study 2 (N = 56, 53% male, mean age 19.08) were recruited from the participant pool of a behavioral lab in a large Northeastern University. Participants in this study were all college students. Participants signed up for the study on an online system (Experimetrix) containing a brief description of the study session. They completed the study for payment as part of a longer session containing multiple unrelated studies. After giving informed consent for their participation in the study, participants were handed paper packets containing questionnaires for the different studies in the session. The position of the questionnaire in the packets was similar for both conditions.
The study was similar to Study 1, with the main difference being that in this case the control condition contained a repetition of the information given in graphic form in the experimental (graph) condition. For the control condition, after reading the previously described paragraph about the medication, participants read “Incidence of illness drops from 83% to 63% with the medication,” paralleling the information given in the graph condition. Participants were randomly assigned to one of two conditions. Experimental condition participants read the same scenario used in the previous study. Control participants were given an additional sentence, stating “Incidence of illness drops from 83% to 63% with the medication.” This sentence was given right after that specifying a 20% reduction in illness. Experimental condition participants were split into two groups, half with the graph cutoff at 0% incidence of disease, as before, and half with the graph cutoff at 50% incidence of disease.
In this case, the dependent variable was perceived medication efficacy. Participants were asked to rate how effective the medication was. Specifically, participants were asked to rate their agreement to the statement “I believe the new drug is effective.” Ratings were given on the same 1–9 scale as before, anchored by “strongly disagree” (1) and “strongly agree” (9).
To test whether it is enhanced retention of information that was responsible for increased belief in product efficacy, participants were asked to report the percent by which the medication reduced illness. We asked this toward the end of the study session, approximately 30 minutes after participants completed the medication questionnaire. If it was increased retention of information due to graphs that underlied our effects, participants in the graph condition should display higher recall.
In support of the effect’s reliance on the scientific aura of graphs, we also wished to see whether participants’ belief in science moderated the effect. Participants rated their agreement to the statement “I believe in science” on a scale of 1 (strongly disagree) to 9 (strongly agree). This information was gathered in conjunction with the recall question given at the end of the study session.
Results and discussion
Planned contrast between the low and high graph cutoff produced no differences, such that we collapsed graph conditions. There was a significant effect of graph on medication effectiveness, such that participants shown a graph rated medication as more effective (5.75 vs 4.66), despite verbal repetition of information in the control/no-graph condition. We analyzed the data using a general linear model controlling for the presence of a graph, belief in science, and their interaction. Analysis revealed a significant effect of graphs: F(1, 51) = 8.18, p = .006. This change did not appear to be due to enhanced retention of information. The percentage of participants who correctly reported reduction in illness (within 3% error) was not greater for the graphs condition, with 78.95% (vs 70.27%) reporting correctly in the control (graphs) condition (p > .2).
In support of the effect’s reliance on the scientific aura of graphs, the study also showed that a belief in science moderates and enhances the effect of graphs on persuasion. Participants’ agreement to the statement “I believe in science” produced a significant interaction with the presence of graphs: F(1, 51) = 10.1, p = .0025. Participants who expressed higher agreement with this statement demonstrated a higher effect of graphs on perceived medication effectiveness, supporting the notion that it is the association of graphs with science which grants them persuasiveness. In other words, the higher participants’ self-reported belief in science, the higher the increase of persuasion due to the presence of graphs. For those who expressed lower belief in science, the effects seem to have been eliminated, since if you do not believe in science as an arbiter of truth, signaling a scientific basis for claims would not make them more credible. Supposedly, graphs increased persuasion to greater extent for people who expressed a strong belief in science because of their signaling a scientific basis for claims.
3. Study 3: Formulas for truth
The results of Studies 1 and 2 could have been due to the visual nature of graphs. Prior research supports the notion that visuals in particular can persuade, regardless of their connection with science (Tufte, 2001). People may have an easier time processing, and consequently shifting opinion, based on visual information (Childers et al., 1985). While the results of Studies 1 and 2 argue that there is no difference in the processing of information, Study 3 aimed to further rule out the possibility of results being specifically due to the visual nature of graphs and the increased persuasion of the visual modality by testing whether another element associated with science, chemical formulas, had similar effects.
The study aimed to expand the generalizability of the last two studies by two additional means. First, we examined persuasion on another dimension that more directly taps a medication’s effectiveness. Rather than answering about medication effectiveness in the abstract, we asked participants to evaluate a variable more directly and concretely reflecting the effectiveness of medicine: the length of its effects. Second, we conducted the study with a more general population than that of our previous two studies, by recruiting participants at a shopping mall.
Methods
Participants (N = 57, 56% male, mean age 31.24) were recruited at a shopping mall. Among participants who reported their level of education, 10.5% did not complete high-school, 61.4% had some college education, and 42% had a 4-year college education or higher. Participants completed the study in exchange for a payment of US$5 for 20 minutes of their time. They were recruited using signs advertising a paid research study. Some participants were recruited by experimenters asking passerby if they wished to participate in the study.
Participants completed informed consent, and were randomly assigned to a formula or non-formula condition. They read scenarios similar to those used in previous studies. A different scenario was used to extend the validity of our results to drugs that have a specific effect versus generally reducing illness. Specifically, they read, “a different company has developed an anti-inflammatory drug called Florinef. It is currently manufactured by Sigma. You can see a picture of the tablets below. The drug’s chemical is C21H29FO5, meaning it’s carbon-oxygen-Helium and-fluorine based.” Control condition participants received the same text, without the formula.
After reading the scenario, participants were asked to estimate the length of time for which the medication would work. This measure was used to extend generalizability from abstract judgment of effectiveness to actual metrics of effectiveness. Participants wrote the number of hours they thought the medication would work for on an open measure. Outliers over 3 standard deviations (SDs) above or below the mean (18 participants over the 55 reported) were eliminated from analysis. Note that inclusion of these outliers actually produced a much stronger pattern of results.
Results and discussion
Participants supplied with a chemical formula estimated the medication would work for 2.14 hours longer than participants shown the formula in words, 5.91 vs 3.77 hours, t(55) = −2.03, p = .05. When including outliers, formula participants anticipated 7.17 hours versus 3.77 hours for control participants (.07, marginal significance).
These results support the notion that elements that appear scientific—in this case a chemical notation—enhanced the persuasion of the message. This was robust and occurred regardless of whether people focused on a visual modality.
4. General discussion
In combination, these three studies demonstrate that the mere presentation of elements associated with science alongside claims about medication efficacy increases persuasion. People who were given graphs or formulas along with claims regarding medication efficacy displayed greater belief in medication effectiveness. Such effects occurred for both graphs (Studies 1–2) and chemical formulas (Study 3), and for different populations: an online panel (Study 1), a campus population (Study 2), and a general population (Study 3).
The prestige of science appears to grant persuasive power even to such trivial science-related elements as graphs. Ostensibly, graphs signal a scientific basis for claims, which grants them greater credibility. This does not seem to be because graphs help cognitive processing. The effects of graphs hold even when no additional information is supplied or even implied by the graphs, and it is not moderated by increased understanding or retention of information. The effects of graphs are also not due to their visual nature—similar non-visual scientific signals also increase persuasion, as seen in Study 3. It also appears that it is the general belief in science that is at least partly responsible for the persuasive power of graphs, as attested to by the moderating effect of such beliefs in Study 2. Given that they signal scientific credibility, graphs have a greater effect for those who have faith in science.
The effects of graphs on persuasion might exemplify a broader inferential process:
The information contains a graph (premise);
Graphs signal a scientific basis (premise);
Therefore, the information has a scientific basis (conclusion);
A scientific basis indicates truth (premise);
Therefore, the information is true (conclusion).
This inferential process would then lead to a greater belief in the efficacy and positive attributes of a marketed product, as reflected in our findings.
Unlike prior research addressing the effects of rhetoric devices on persuasion (McCabe and Castel, 2008; Weisberg et al., 2008), current results cannot be explained by a legitimate inference of expertise. Upon seeing a brain scan, people may infer a source expertise, since laymen may not be able to generate such images. The very fact that they cannot understand the images can lead them to infer that there may be a sophisticated scientific basis to claims. Graphs, on the other hand, are easily understandable and produced, and do not attest to scientific expertise that potentially grants legitimacy to their source. Consequently, the effects of inclusion of graphs in persuasive argument cannot be due to inferred content resulting from their opaqueness.
Finally, our research examines more direct indications of persuasion, assessing people’s belief in claims rather than merely rating scientific reasoning (McCabe and Castel, 2008), or satisfaction in explanations (Weisberg et al., 2008). This underscores the importance of such rhetoric devices, showing that they possess an effect on people’s beliefs, rather than just influencing their assessment of the claims.
Further research can investigate the links between the “science-signaling” of graphs and their effects on persuasion. In addition, research may extend to study what elements other than graphs and formulas increase persuasion through “science signaling.” Finally, research can also examine whether particular populations would be more or less vulnerable to these effects, and investigate methods of mitigating such effects.
On the practical front, the research demonstrates how easily companies can create a scientific appearance. Accordingly, it advises caution when encountering communications hinting at scientific credibility. Companies may create a scientific façade for claims that is not backed by solid scientific evidence using even simple means such as graphs or formulas. Consumers may wish to consider what the actual basis of claims is, and ignore spurious “cues to a scientific basis” that do not in fact bolster the evidential basis of claims.
The findings also emphasize the importance of assessing what constitutes legitimate communication. Though the use of elements hinting at a scientific basis may not constitute outright deception, it may at times be abused to create a scientific appearance which is not in fact justified by evidence.
5. Conclusion
Even easily produced, trivial elements that are associated with science, such as graphs, can enhance persuasion. These findings demonstrate that companies can easily abuse the prestige with which science is held. Adding even trivial or peripheral elements that are associated with scientific objectivity can help persuade people of product efficacy. This must be guarded against in a wide variety of different contexts, including advertising, product packaging, web-design, sales visits, and press releases. The fact that elements associated with science can so easily enhance persuasion urges caution in the communication of purportedly scientific claims, and a more critical eye when it comes to assessing claims that are given a scientific veneer. If in need of further convincing, please refer to our graph (Figure 2).
A highly convincing graph.
Footnotes
Acknowledgements
The authors are grateful to Julia Hastings-Black for editorial assistance and to the reviewers and editorial team for their contributions to this work.
Funding
This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.