Concealing research outcomes: Are there times when it is actually justified?

The recently published informative and well-written editorial from Dr. Bhaskar[1] addressing the reasons why as much as 60% of research results go unpublished provides several plausible and well-documented reasons including overt publication bias[2] and the “file drawer effect.” Although I agree that these effects could, in part, be related to researchers’ inherent biases not to report the results of negative studies (thinking that journals’ past tendencies that favour publishing positive over negative study results will reduce their chances of manuscript acceptance), other factors might also be at play. Indeed, although this type of publication bias may have been common in the past,[3] and is hopefully on the decline, there is also a large amount of research that is undertaken that involves poor study design, including studies that are either too small (and underpowered) or include an overly optimistic effect size in their power calculation. Thus, many "negative" studies are only negative (and thus “filed away” by investigators) because they were underpowered from the start and proceeding with publishing an a priori – designed underpowered trial can contribute to some distortion of the literature. So, whereas I am a firm believer that all well-conducted research should find a public forum for dissemination (i.e., through print publication), the risk of publishing negative underpowered studies cannot be overlooked. Although 60% of research is arguably far too large an amount of unpublished data, some research should arguably never be published for the single reason that it likely should not have been undertaken in the first place.