Editing Parapsychology (section)

===Selection bias and meta-analysis===
[[Selection bias|Selective reporting]] has been offered by critics as an explanation for the positive results reported by parapsychologists. Selective reporting is sometimes called a "file drawer" problem, which arises when only positive study results are made public, while studies with negative or null results are not made public.<ref name="pmid16822164"/> Selective reporting has a compounded effect on [[meta-analysis]], which is a statistical technique that aggregates the results of many studies to generate sufficient statistical [[power (statistics)|power]] to demonstrate a result that the individual studies themselves could not demonstrate at a [[statistical significance|statistically significant]] level. For example, a recent meta-analysis combined 380 studies on psychokinesis,<ref name="pmid16822162"/> including data from the PEAR lab. It concluded that, although there is a statistically significant overall effect, it is inconsistent, and relatively few negative studies would cancel it out. Consequently, [[Publication bias|biased publication]] of positive results could be the cause.<ref name="Smee"/>

Numerous researchers have criticized the popularity of meta-analysis in parapsychology,<ref name="UttsStatisticalScience" /> and is often seen as troublesome even within parapsychology.<ref name="UttsStatisticalScience">{{Cite journal|title=Replication and Meta-Analysis in Parapsychology |journal=Statistical Science |year=1991 |first=Jessica |last=Utts |author-link= Jessica Utts |volume=6 |issue=4 |pages=363–403|doi=10.1214/ss/1177011577|doi-access=free }}</ref> Critics have said that parapsychologists misuse meta-analysis to create the incorrect impression that statistically significant results have been obtained that indicate the existence of psi phenomena.<ref name=Stenger>{{cite web |last=Stenger |first=Victor J. |author-link=Victor J. Stenger |title=Meta-Analysis and the Filedrawer Effect |website=Committee for Skeptical Inquiry |year=2002 |url=http://www.csicop.org/sb/show/meta-analysis_and_the_filedrawer_effect |access-date=2007-07-30 |archive-url=https://web.archive.org/web/20180918091029/https://www.csicop.org/sb/show/meta-analysis_and_the_filedrawer_effect |archive-date=2018-09-18 |url-status=dead }}</ref> Physicist [[Robert L. Park|Robert Park]] states that parapsychology's reported positive results are problematic because most such findings are invariably at the margin of statistical significance and that might be explained by a number of confounding effects; Park states that such marginal results are a typical symptom of [[pathological science]] as described by [[Irving Langmuir]].<ref name="Park 2000"/>

Researcher J. E. Kennedy has said that concerns over meta-analysis in science and medicine also apply to problems present in parapsychological meta-analysis. As a [[post-hoc analysis]], critics emphasize the opportunity the method presents to produce biased outcomes via selecting cases chosen for study, methods employed, and other key criteria. Critics say that analogous problems with meta-analysis have been documented in medicine, where it has been shown different investigators performing meta-analyses of the same set of studies have reached contradictory conclusions.<ref>{{Cite journal|title=A Proposal and Challenge for Proponents and Skeptics of Psi |journal=Journal of Parapsychology |year=2005 |first=J.E. |last=Kennedy |volume=68 |pages=157–167 |url=http://jeksite.org/psi/jp04.htm |access-date=2007-07-29}}</ref>