(2011) in slightly different terms: p-hacking, he holds, “refers to the practice of reanalyzing data in many different ways to yield a target result.” Meanwhile, biologists Head et al. (2011) is that p-hacking amounts to “flexibility in (a) choosing among dependent variables, (b) choosing sample size, (c) using covariates, and (d) reporting subsets of experimental conditions.” Thomas Insel, a former director of the National Institute for Mental Health, summarizes the definition in Simmons et al. An influential definition from psychologists Simmons et al. What exactly is p-hacking? To illustrate the range of methods described as p-hacking, consider the following recent claims about p-hacking from researchers in a variety of disciplines. P-hacking and the related issue of publication bias are frequently cited as twin evils at the core of the ongoing ’replication crisis’ in the social and medical sciences, and are the targets of proposals for policy reforms intended to rework the incentive structures of scientific publishing. Since statistically significant results are more likely to be published and promoted, researchers have incentives to engage in such activities. It has become the norm in scientific publishing to describe p-values below some threshold (often 0.05) as “statistically significant.” Researchers may thus “hack” their analyses and reports of analyses in order to produce p-values that qualify as statistically significant.
The phrase originates in the ubiquitous statistical method of reporting ‘p-values’ of statistical analyses of data. ‘P-hacking,’ a term used widely in contemporary scientific discourse, refers to a variety of practices.
0 kommentar(er)
