On a generalizable approach for sample size determination in Bayesian t tests DOI Creative Commons
Tsz Keung Wong, Jorge N. Tendeiro

Behavior Research Methods, Journal Year: 2025, Volume and Issue: 57(5)

Published: March 31, 2025

Abstract The Bayes factor is often proposed as a superior replacement to p values in testing null hypotheses for various reasons, with the availability of many user-friendly and easily accessible statistical software tools facilitating use Bayesian tests. Meanwhile, design analysis (BFDA), counterpart power analysis, also ensure maximum efficiency informativeness study. Despite conducting BFDA being limited mostly relying on Monte Carlo methodology, methods based root-finding algorithms have been recently developed (e.g., Pawel Held, 2025), overcoming weaknesses simulation approaches. This paper builds these advancements by presenting method generalizing existing approach sample size determination t major advantage current that it does not assume normality effect estimate, allowing more flexibility specification priors. We showcase Shiny app BFDA, illustrated an empirical example. Furthermore, using our method, we explore operating characteristics factors

Language: Английский

On a generalizable approach for sample size determination in Bayesian t tests DOI Creative Commons
Tsz Keung Wong, Jorge N. Tendeiro

Behavior Research Methods, Journal Year: 2025, Volume and Issue: 57(5)

Published: March 31, 2025

Abstract The Bayes factor is often proposed as a superior replacement to p values in testing null hypotheses for various reasons, with the availability of many user-friendly and easily accessible statistical software tools facilitating use Bayesian tests. Meanwhile, design analysis (BFDA), counterpart power analysis, also ensure maximum efficiency informativeness study. Despite conducting BFDA being limited mostly relying on Monte Carlo methodology, methods based root-finding algorithms have been recently developed (e.g., Pawel Held, 2025), overcoming weaknesses simulation approaches. This paper builds these advancements by presenting method generalizing existing approach sample size determination t major advantage current that it does not assume normality effect estimate, allowing more flexibility specification priors. We showcase Shiny app BFDA, illustrated an empirical example. Furthermore, using our method, we explore operating characteristics factors

Language: Английский

Citations

0