Mellman: Seeing the seduction of science

I probably shouldn’t tell you this: I deal partly in science, and the more science-y I make it look and sound, the more persuasive I am.

I don’t know that from my own experience — though political consulting is built on anecdotes, they can be totally misleading. I know it from solid experimental evidence. 

{mosads}Two Cornell University scientists gave a sample of highly educated Americans information about a new medication that ostensibly prevents colds, and then divided them at random into two groups. Half were shown a graph that provided no new information but charted the “fact” that 87 percent of those who did not take the medicine caught a cold, compared to only 47 percent of those who did ingest the medicine. Those who did not see the graph were given exactly the same the information in the text itself, just the way I am here.

Nearly all those who saw the graph (97 percent) thought the medicine really reduced illness, compared to 68 percent of those who only read the text and were not exposed to the graph.

While graphs help provide a persuasive patina of scientific credibility, so do other modes of communication sporting the look and feel of science. 

The same Cornell scientists gave participants information about an anti-inflammatory drug. Everyone was also told that the “drug’s chemical was carbon-oxygen-helium and fluorine based.” Half the sample was also given the chemical formula (C21H29FO5). Both groups were then asked how long they thought the drug would work. 

Those who saw the text without the chemical notation estimated it would last 3.77 hours on average. Those given the formula predicted it would keep reducing inflammation for a whopping 7.17 hours, almost twice as long.

Making the description appear slightly more scientific, merely by inserting a chemical formula, significantly increased the perceived efficacy of the drug.

Mathematical formulas can have the same kind of impact — even when they are nonsensical. 

A Swedish professor exposed a sample of people with master’s degrees and Ph.D.s to abstracts of research papers and asked them to grade the quality of the research. The only difference between them was the sentence, “A mathematical model (T_pp=T₀-∫T_0d2f-–fTpdf) is developed to describe sequential effects,” which was appended to half the abstracts. These symbols had absolutely no meaning and no referents in the papers.

Abstracts that contained the “formula” were seen as better research by some two-thirds of respondents. Reassuringly, mathematicians and those in hard sciences were far less likely to fall for the trick, but even some of the real experts succumbed to the siren song of something that looks like science.

Two other researchers, Brendan Nyhan and Jason Reifler, recently documented a similar effect in the political realm. They gave some participants in the experiment graphical information about global temperature rise, while others received the same data in text form. Republicans were more likely to believe global warming was a reality if they saw the graph instead of just reading the text.

Similarly, a group of scholars at New York University looked at the relative ability of text and charts to persuade on political issues — corporate income taxes, incarceration and violent video games. Those with weaker initial views were more likely to change their views by more than 20 points when the evidence was presented in charts rather than tables. 

Charts and formulas look like science; they give off the air of “truth” and are therefore more persuasive than “mere” explanation.

Human beings have spent a lot more evolutionary time developing our visual skills than we have learning to analyze language. Our brains devote vastly more resources to visual processing as well. 

There’s a good lesson in these studies for persuasive communicators.

There is also a strong warning to the public: Don’t be seduced merely by the sound and look of science.

Mellman is president of The Mellman Group and has worked for Democratic candidates and causes since 1982. Current clients include the minority leader of the Senate and the Democratic whip in the House.