A bias in what academic research is more likely to be published; e.g. findings with significant outcomes are more likely to be published than the rest.
The term publication bias is used to describe the situation where the chances of a study being published depend on its results. Statistical attempts to detect it usually rely on detecting small study bias where it is hypothesised that large studies are more likely to be published irrespective of outcome whereas small ones may be affected. Techniques used include funnel plots and various regression tests of which the best know are due to Egger and to Begg.
The wikipedia article https://en.wikipedia.org/wiki/Publication_bias has more detail and some references.
