File drawer problem

From Wikipedia, the free encyclopedia

It has been suggested that this article or section be merged into Publication bias. (Discuss)

The file drawer problem is that many studies in a given area of research may be conducted but never reported, and those that are not reported may on average report different results from those that are reported. An extreme scenario is that a given null hypothesis of interest is in fact true, i.e. the association being studied does not exist, but the 5% of studies that by chance show a statistically significant result are published, while the remaining 95% where the null hypothesis was not rejected languish in researchers' file drawers.

The term was coined by the psychologist Robert Rosenthal in 1979.^[1] The term "file drawer problem" is often taken to be synonymous with publication bias. "Publication bias" is a more general term, as it may include differences in the availability or accessibility of published papers due to the language, format or journal of publication.

Contents 1 Example 2 See also 3 References 4 External links

[edit] Example

Suppose that 20 studies are done around the world to assess whether vitamin C prevents some disease. A result will be considered statistically significant if the probability that it (or a more extreme result) would occur by chance is less than 5%. If vitamin C has no such effect, then we would still expect that 1/20 of the studies (i.e., one study) will report a significant result. If only that one study is published, it looks as if vitamin C is effective. If all 20 are published, the opposite conclusion appears. Thus the file drawer problem biases the published conclusions against the null hypothesis that vitamin C is ineffective.

[edit] See also

• Meta-analysis
• Counternull

[edit] References

[edit] External links

• http://skepdic.com/filedrawer.html