A team at Stanford University reports online this week in Science that scientists are unlikely to even write up an experiment that produces so-called null results. A study of 221 survey-based experiments funded by the TESS (Time-sharing Experiments for the Social Sciences) program at the National Science Foundation has found that almost two-thirds of the experiments yielding null findings are stuck in a file drawer rather than being submitted to a journal, and only 21% are published. In contrast, 96% of the experiments that yield strong results are written up, and 62% of them are published. Such practices by researchers can skew the literature and lead to wasteful duplication, the authors argue. To combat the problem, the authors call for a social science registry that would contain all such data, as well as descriptions of the methodology used to analyze the results.