# 40. Discard unfavourable data ## 40.5. Methodology/Refinements/Sub-species ### 40.5.2. File-drawer effect This is the idea of burying ineffectual data. Often this happens accidentally, but sometimes it occurs for manipulative reasons because the data doesn't appear to contribute anything to some pre-determined script. When a study gives rise to results which find nothing of statistical significance or causal consequence, the researcher often tends to "file and lose" the offending study report, so that, effectively, it ceases to exist. The "file-drawer" effect refers to the practice of researchers in filing away and forgetting such apparently ineffectual study results. However, this is not a scientifically valid practice and has some unfortunate consequences on a line of research. The practice of reporting and publishing only positive-outcome research creates a misrepresentation of the subject under investigation, especially if a meta-analysis of the entire data set is ever conducted on some future date. Even non-results are results and need to be added to the sum total of our dataset and knowledge of a subject. A classic case of this phenomenon is the losing of "study non-results" in the study of parapsychology. Many studies are conducted, but only a fraction of the results ever see the light of day, the rest being considered to be statistically neutral, negative to parapsychology or of no interest. At the time of writing, the Swiss multinational Roche has been accused by the British Medical Journal of sitting on trial data for its flu treatment, Tamiflu. Governments around the world have stockpiled Tamiflu (at a great profit to the producers) against a possible pandemic, but at a cost to many governments of billions of pounds, euros and dollars.