Table 2.

Fit indices of the choice models computed by multilevel Bayesian generalized linear modeling (ordered by fit).

Choice model	${\widehat{elpd}}_{diff}$	$se({\widehat{elpd}}_{diff})$	${\widehat{elpd}}_{loo}$	$se({\widehat{elpd}}_{loo})$
Model 2: PHS*Healthiness +RE	0.0	0.0	−20476.4	174.2
Model 1: Session*PHS*Healthiness+RE	−3.8	2.5	−20480.2	174.3
Model 3: Healthiness+RE	−79.9	5.8	−20556.2	174.3
Model 4: 1+RE	−4754.3	100.2	−25230.6	208.0

The endorsed model is indicated in bold. All models are compared to the best model (i.e., Model 2). More complex models are considered to be better if they show more than one standard error enhancement in ${\widehat{elpd}}_{diff}$ (Vehtari et al., 2016; Magnusson et al., 2020). RE, Random effects. For a detailed specification of the models, see Equations (1)–(4). ${\widehat{elpd}}_{loo}$, Expected log pointwise predictive density for a new dataset using the Pareto smoothed importance sampling (PSIS) leave-one-out cross-validation (loo) criterion (Vehtari et al., 2016; Magnusson et al., 2020). The closer to zero, the better the model is. ${\widehat{elpd}}_{diff}$, The difference between ${\widehat{elpd}}_{loo}$of two compared models. se, Standard error of the targeted variable.