- Published: January 2, 2022
- Updated: January 2, 2022
- University / College: Johns Hopkins University
- Level: Undergraduate
- Language: English
- Downloads: 19
Article Analysis Article Analysis Behavioral researchers and statisticians had been suspecting for a long time that the studies on behavioral sciences were biased. For any research ever done no one could tell how many studies conducted had never been reported. A view on the’ file drawer problem’ showed that journals were 5% constituted by studies showing the Type 1error while the file drawers were 95% constituted of the studies showing the non-significant results. This research, therefore, aimed to provide quantitative measures for computing the tolerance for filled and future null results and their inference.
The researchs first finding was that, the elemental idea of keeping up with the file drawer problem was to calculate the number of studies averaging the null results that must be in the file drawers before bringing the overall probability of a Type 1 error to any level of implication. This was despite the fact that the calculation of an overall probability becomes much easier if done on all the independent studies at the researchers disposal, the question in effect is not a Type 1 error (Rosenthal, 1979). The second finding was that, after evaluating the tolerance for future null results a determination on whether the tolerance level of the future null result was small enough to threaten the overall conclusion by the researcher. Should the overall significance of the research review be brought down to the level of just significant by the summation of just a few more results, then it was concluded that the finding was not resistant to the file drawer threat.
The author’s interpretation if the result is that the small numbers of studies that seemed insignificant could be misleading because only a few studies filed away could change a combined significant result to a non-significant one. An increase in the availability of studies could safely rule out the file drawer hypothesis as a credible opponent hypothesis.
In my view there is a great importance for the researcher to provide the number of researches are done and the tolerance for future null results associated with their overall significance level so as to help in the determination of the significance of the file drawer hypothesis.
Reference
Rosenthal, R. (1979). The “ File Drawer Problem” and Tolerance for Null Results. Psychological Bulletin, Vol. 8 (3), 638-641.