SAMUEL NEFF, M.D.
To the editor: Readers of Douglas Hofstadter (1), logicians, and others will be amused by the publication of an analysis of meta-analysis, a meta-meta-analysis (2, 3). This letter, a commentary on those papers, is therefore an example of meta-meta-meta-analysis. As the previous authors point out, at each higher level new biases are introduced: publication bias and selection bias for the source studies, and publication bias for the meta-analysis itself (positive or "interesting" meta-analyses may be preferentially published).
It seems agreed that the need for meta-analysis has been created by the proliferation of studies that disagree, by sample sizes that are
NEFF S. Meta-Analysis. Ann Intern Med. 1987;107:791. doi: 10.7326/0003-4819-107-5-791_1
Download citation file:
Published: Ann Intern Med. 1987;107(5):791.
Copyright © 2018 American College of Physicians. All Rights Reserved.
Print ISSN: 0003-4819 | Online ISSN: 1539-3704
Conditions of Use