Some psychologists argue strongly for bayesian stats
Only last week, we heard scathing (and well-argued) remarks against the use of Bayesian methods for scientific hypothesis testing. Yet, some very smart people argue in precisely the opposite direction…
Kruschke, J. K. (2011). Bayesian Assessment of Null Values Via Parameter Estimation and Model Comparison. Perspectives on Psychological Science, 6(3), 299–312. doi:10.1177/1745691611406925
Dienes, Z. (2011). Bayesian Versus Orthodox Statistics: Which Side Are You On? Perspectives on Psychological Science, 6(3), 274–290. doi:10.1177/1745691611406920
Kruschke, John K. (2013). Bayesian estimation supersedes the t test. Journal of Experimental Psychology: General, 142(2), 573–603. doi:10.1037/a0029146
Despite the appeal of Dienes’ preferred method of model choice, given it’s analogous structure to Neyman-Pearson Null Hypothesis Significance Testing (NHST), the results appear to vary quite readily given your choice of prior. Of course, Deines – and some other Bayesians – think this is a strength.
However, it just feels safer to use Kruschke’s preferred method of the construction of “credible intervals” – much more analogous to frequentist confidence intervals. Of course, maybe confidence intervals are superior to NHST in the frequentist world as well…