Psychologie statistique avec R

Abstract

This book provides a detailed presentation of allbasics of statistical inference for psychologists,both in a fisherian and a bayesian approach. Althoughmany authors have recently advocated for the use ofbayesian statistics in psychology (Wagenmaker et al.,2010, 2011; Kruschke, 2010; Rouder et al., 2009)statistical manuals for psychologists barely mentionthem. This manual provides a full bayesian toolboxfor commonly encountered problems in psychology andsocial sciences, for comparing proportions, variancesand means, and discusses the advantages. But allfoundations of the frequentist approach are alsoprovided, from data description to probability anddensity, through combinatorics and set algebra. Aspecial emphasis has been put on the analysis ofcategorical data and contingency tables. Binomial andmultinomial models with beta and Dirichlet priors arepresented, and their use for making (between rows orbetween cells) contrasts in contingency tables isdetailed on real data. An automatic search of thebest model for all problem types is implemented in theAtelieR package, available on CRAN. ANOVA is alsopresented in a Bayesian flavor (using BIC), andillustrated on real data with the help of the AtelieRand R2STATS packages (a GUI for GLM and GLMM in R).In addition to classical and Bayesian inference onmeans, direct and Bayesian inference on effect sizeand standardized effects are presented, in agreementwith recent APA recommendations.