Powering Reproducible Research
Katherine S. Button, Marcus R. Munafò
Abstract
Open-access reader
Katherine S. Button, Marcus R. Munafò
Abstract
Open-access reader
The widely used null hypothesis significance testing (NHST) framework grew out of the distinct statistical theories of Fisher, and Neyman and Pearson. From Fisher, this chapter considers the concept of null hypothesis testing, and from Neyman-Pearson the concepts of Type I (α) and Type II error (β). Power is a concept arising from Neyman-Pearson theory, and reflects the likelihood of correctly rejecting the null hypothesis (i.e., 1 - β). However, the hybrid statistical theory typically used leans most heavily on Fisher's concept of null hypothesis testing. Within the hybrid NHST framework that continues to dominate within psychological science, the power of a statistical test is the probability that the test will correctly reject the null hypothesis when the null hypothesis is genuinely false. Low power may therefore contribute to the poor reproducibility of scientific findings, which continues to be a cause of concern.
OpenAlex reports 13 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
The widely used null hypothesis significance testing (NHST) framework grew out of the distinct statistical theories of Fisher, and Neyman and Pearson. From Fisher, this chapter considers the concept of null hypothesis testing, and from Neyman-Pearson the concepts of Type I (α) and Type II error (β). Power is a concept arising from Neyman-Pearson theory, and reflects the likelihood of correctly rejecting the null hypothesis (i.e., 1 - β). However, the hybrid statistical theory typically used leans most heavily on Fisher's concept of null hypothesis testing. Within the hybrid NHST framework that continues to dominate within psychological science, the power of a statistical test is the probability that the test will correctly reject the null hypothesis when the null hypothesis is genuinely false. Low power may therefore contribute to the poor reproducibility of scientific findings, which continues to be a cause of concern.
Key concepts: Null hypothesis, Statistical hypothesis testing, Type I and type II errors, Null (SQL), Statistical power, Scrutiny, Alternative hypothesis, p-value