Supplemental Material for Cognitive and Cultural Models in Psychological Science: A Tutorial on Modeling Free-List Data as a Dependent Variable in Bayesian Regression
Psychological Methods · 2023
In Supplementary Section A, we implement an alternative likelihood to the zero-one inflated beta, namely the ordered beta distribution.For technical details and background, see Kubinec ( 2022) and for practical implementation, see the R package ordbetareg (https: //github.com/saudiwin/ordbetareg)and source code for the present supplementary document.In Supplementary Section B, we illustrate and implement alternative likelihood models that do not require transforming free-list data to salience scores, as discussed briefly in the main manuscript.We integrate key code chunks in the text but refer to the source code for full reproducible code.For simplicity, we restrict the examples to the Model 1 data and exclude field-site specific 'random effects' in all models.In Supplementary Section C, we first present plots for Models 3-8 (Figures 10 and11).Then, we present results for the reduced models, where only non-zero free-list responses are retained .Finally, we present results from leave-one-group-out crossvalidation (Table 3), an alternative model comparison to the approximate leave-one-out cross-validation as presented in the main manuscript.