.png){kind=icon}
.png){kind=icon}
Return
A Cautionary Note on Estimating Effect Size
.png){kind=icon}
OA
AI
DOI：10.1177/2515245921992035.png)
.png){kind=icon}
.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
An increasingly popular approach to statistical inference is to focus on the estimation of effect size. Yet this approach is implicitly based on the assumption that there is an effect while ignoring the null hypothesis that the effect is absent. We demonstrate how this common null-hypothesis neglect may result in effect size estimates that are overly optimistic. As an alternative to the current approach, a spike-and-slab model explicitly incorporates the plausibility of the null hypothesis into the estimation process. We illustrate the implications of this approach and provide an empirical example.
Keywords:
effect size
Bayesian estimation
modeling
shrinkage
open materials
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
A
IF:
13.4
Papers:
311
Citations:
3.7K