Many scientists are wary of discussing unpublished results with journalists, suggests a meta-analysis of survey data (H. P. Peters Proc. Natl Acad. Sci. USA http://doi.org/nhz; 2013). Hans Peter Peters, a social scientist at the Jülich Research Centre in Germany, examined published and unpublished data that he and his colleagues collected in surveys of US neuroscientists and German scientists. A little more than half of US and German neuroscientists, and 44% of German scientists across all fields, agreed that acceptance of a paper by a journal is threatened if the findings have already been revealed in the mass media. Peters notes that most scientists make a sharp distinction between how they communicate among themselves and how they talk to the media and the public.