Must psychologists change the way they analyze their data?

Area of Science:

• Psychology
• Statistics

Background:

• Frequentist statistical analyses are standard in psychology.
• Bayesian analyses offer an alternative approach to data interpretation.
• Wagenmakers et al. (2011) critiqued frequentist methods, advocating for Bayesian approaches using Bem's (2011) psi experiments.

Purpose of the Study:

• To re-evaluate Bem's (2011) psi experiments using Bayesian analysis.
• To demonstrate the impact of prior distribution selection in Bayesian analysis.
• To highlight the advantages and potential challenges of adopting Bayesian methods in psychology.

Main Methods:

• Bayesian reanalysis of Bem's (2011) psi experimental data.
• Comparison of results obtained with different prior distributions.
• Critique of the prior distribution selection in Wagenmakers et al. (2011).

Main Results:

• The original Bayesian analysis by Wagenmakers et al. (2011) used an unrealistic prior, leading to non-supportive conclusions for the psi hypothesis.
• A revised Bayesian analysis with a more appropriate prior distribution yielded strong evidence supporting the psi hypothesis.
• The study illustrates that Bayesian methods, while advantageous, require careful consideration of prior assumptions.

Conclusions:

• Bayesian analysis, when properly applied with suitable priors, can provide evidence supporting parapsychological phenomena.
• The choice of prior distribution significantly influences the outcome of Bayesian statistical inference.
• Psychologists should exercise caution and gain a deeper understanding of Bayesian "hidden traps" before widespread adoption.