Media representations of genetic discoveries: hype in the headlines?

• Author: Caulfield, Timothy

Many commentators have expressed concern regarding the sensationalistic reporting of biomedical stories by the popular press. (1) It has been suggested that inaccurate or exaggerated reporting can have an adverse impact on public understanding, creating unwarranted hope or fears, and the development of informed policies. (2)

Readers get their first or only impressions from headlines. Unfortunately, there are reasons to believe that headlines may be particularly inaccurate or "hyped." Science and medical stories have to compete with other news stories and, as such, headlines must be constructed to catch the attention of both the potential reader and editors who make publishing decisions. (3) As a result, even when a media report is circumspect, headlines may be sensationalized. (4)

"Headline sensationalism" has been associated with a variety of specific social concerns. For example, in the context of genetic discoveries, repeated exposure of the lay public to such headlines may lead to heightened genetic determinism. (5) That is, the public will come to develop an inaccurate belief that there is a tight causal linkage between a gene and a given human trait or disease. It has also been noted that headlines can influence how the reader interprets the information presented within the body of the full article. (6)

Sensationalized headlines that bear little resemblance to the article may generate antipathy or disappointment among some readers, (7) creating a degree of bad will with a portion of the potential audience that should give editors pause. (8) They may also alienate sources, including those in the scientific community.

While there is a growing body of research on the accuracy and nature of newspaper stories (9), there is little available data on the accuracy of headlines in the context of genetic research. This study builds on the results of a previous paper examining the accuracy of newspaper stories. (10) We examine the degree and nature of the "hype" present in newspaper headlines associated with stories on genetic discoveries.

Methods

We describe only those methods that are specific to this study on headlines. The selection of scientific papers and newspaper articles, the general coding frame and coding, and a detailed explanation of the statistical analysis using CART were presented in Bubela and Caulfield. (11)

Parts of the coding frame were specific to the headline study. The coders were asked a series of questions with standardized categorical responses on the theme; the source of information (voice) beside the scientific paper; assessment of risk, benefits, or controversy for headlines, newspaper articles, and the scientific paper that generated the press coverage (Table 1).

Three coders, who all had scientific backgrounds, were asked to subjectively assess the technical accuracy of the headline compared to the scientific paper. The coders also subjectively assessed whether the claims made in the headline were exaggerated (1) relative to the newspaper article and (2) relative to the scientific journal article. The coders assigned the headline to one of three categories: no exaggerated claims, slightly exaggerated claims, and moderately-highly exaggerated claims with respect to both the contents of the newspaper article and the scientific journal article. When compared to the newspaper and the scientific journal article, the headline was not considered to have been exaggerated if its claims had first been made there.

Statistical Analysis

All three coders assessed a random selection of 84/627 (13.7%) headlines, newspaper articles and their associated papers to assess inter-coder reliability. We calculated Intraclass Correlation Coefficients (Model 2) for questions on the coding frame for which there was only one answer, and in all cases the coefficient was greater than 0.75, indicating good agreement (Table 2). (12)

A classification tree analysis determined which variables from the coding frame contributed to the assignment of the headline to one of the three categories of exaggerated claims using CART 4.0 (Classification and Regression Tree) software when compared to the contents of the newspaper article and the scientific paper, respectively. (13) The relative contribution of variables as primary splitting variables in the classification tree are listed in Table 3. These indicate...