Affordable Access

Are the studies (tests) assessing the efficacy of the treatments of the alcoholdeprivation effect underpowered?

Authors
  • Léonard, François
  • Tirelli, Ezio
Publication Date
May 28, 2021
Source
ORBi
Keywords
Language
English
License
Green
External links

Abstract

While adequate power and sample sizes are indispensable for the detection (by the statistical test) of a meaningful effect size (ES), many published studies in psychology do not describe sample-size calculation (SSC), which weakens the study methodological quality. Omission of SSC is often associated with a lack of prospective power, which exaggerates observed ES and increases the False Report Probability (FRP), thereby jeopardizing results reproducibility. Our purpose is to investigate in which extent this practice concerns the literature assessing the efficacy of the psychological and pharmacological treatments of the alcohol-deprivation effect (operationalized in an animal model). We will firstly select articles published from 1993 to 2020 using the database PubMed and check whether they mention a SSC. We will then classify the articles mentioning a SSC according to five components of a complete description of SSC (ex. power analysis with or without details). We will also check whether the hypothetical ES (used to determine sample sizes) is justified and the observed ES interpreted (discussion). Thereupon, in order to assess a possible ES overestimation in the selected literature, we will examine the relationship between the observed ES and the sample sizes. We will compute the “power-to-detect” of each relevant statistical test using small, medium and large ES (classifications). Finally, we will compute the FRP and the True Discovery Rate using a risk alpha of 1, 2.5 or 5%, the median “power-to-detect” and a representative range of pre-study odds (from 0.01 to 0.99), according to Ioannidis (2005) and Szucs & Ioannidis (2017).

Report this publication

Statistics

Seen <100 times