This OSF provides the empirical data, the software implementation of the RDEX model, including the code for fitting the data and replicating the parameter-recovery studies for “A Hybrid Approach to Dynamic Cognitive Psychometrics” by Charlotte Tanis, Andrew Heathcote, Mark Zrubka and Dora Matzke. - **supplemental_figures.pdf** - Prior and posterior distribution of the population-level location and scale parameters inferred from fitting the hierarchical RDEX model to the empirical data. - Posterior predictive results for each cell of the experimental design. - Distributions of the posterior means from the parameter-recovery studies corresponding to the 18 parameter regions derived from fitting the RDEX model to the data of the individual participants. - **posteriors_individual.pdf:** Posterior distributions derived from nine sampling runs of fitting the RDEX model to the empirical data of each participant individually. Each page corresponds to a single participant. The nine different colors show the posteriors corresponding to the nine sampling runs. Block Type: blockwise vs. trialwise; Bias: Blue vs. Orange; Accumulator: BLUE vs. ORANGE; Difficulty: easy vs. hard; Accumulator Match: true vs. false, corresponding to the accumulator matching or mismatching the stimulus; gf=PGF; tf= PTF. - **posteriors_hierarchical.pdf:** Participant-level posterior distributions derived from nine sampling runs of hierarchically fitting the RDEX model to the empirical data. Each page corresponds to a single participant. The nine different colors show the posteriors corresponding to the nine sampling runs. Block Type: blockwise vs. trialwise; Bias: Blue vs. Orange; Accumulator: BLUE vs. ORANGE; Difficulty: easy vs. hard; Accumulator Match: true vs. false, corresponding to the accumulator matching or mismatching the stimulus; gf=PGF; tf= PTF. - **DataFitting:** This directory contains the empirical data, the precomputed posterior samples, and all R scripts necessary to reproduce the model fits, the posterior predictive simulations, and the parameter inference reported in the main text. - **dmc:** Dynamic Models of Choice package (DMC; Heathcote et al., 2019) used for model fitting. - **DataSamples.RData:** "StopData": empirical data; "hStop": precomputed posterior samples resulting from fitting the hierarchical RDEX model to the data. - **DynamicCognitivePsychometrics.R:** R script for specifying the RDEX model and the priors and for post-processing the posterior samples, i.e., checking convergence, running and plotting the results of the posterior predictive simulations, and posterior parameter inference. - **fit.R:** R script for fitting the RDEX model to the empirical data and for performing some preliminary post-processing steps, i.e., plotting the MCMC chains, generating and plotting posterior predictions, summarizing the posterior distributions, computing the Deviance Information Criterion (DIC), and checking convergence. - **ParameterRecovery:** This directory contains all scripts to reproduce the results of the parameter recovery study. Due to size limitations, the results themselves are not included. The data required to regenerate the figures is provided. - **0_essentials_ss-nopert.R:** Script to specify the design, true parameters for each participant, and priors. This script defines the settings of the parameter recovery study and is sourced by all other scripts. - **1_data-models.R:** Script to simulate data sets for each participant from their true parameter values defined in 0_essentials_ss-nopert.R. For each participant, 200 data sets are generated and the data is combined with the model in the data model. - **2_check-data.R:** Script to check if the simulated data looks as expected (i.e., normally distributed, no bulk at 0, reasonable response rate and go omissions, and probability of stopping around .5). - **3_start-values_smallsamples.R:** Script to produce start values for each dataset that will be used to sample from the posterior. We performed the sampling procedure on a high performance computing cluster, see /grid for the scripts. The start values are saved as .RData and are the input for these scripts. - **4_check-chains_smallsamples.R:** Script to check the samples from the posterior distribution (i.e., convergence, and posterior means and medians updated and unimodal). - **5_plot-post-means_caveats.R:** Script to produce Figure 7, an example of the distribution of posterior means that looks as desired (unimodal, centered around the true value), and one that shows undesired behavior (bi-modal). - **5_plot-post-means_smallsample.R:** Script to produce Figure 3-20 of the Supplementary Materials, the distribution of posterior means and the associated true values for each participant. - **6_plot-pairs_smallsample.R:** Script to plot the posterior means of each pair of parameters and their true values to check the relation between parameters. - **7_data-plot-recovery_smallsample.R:** Script to prepare the data for the recovery plots (i.e., scripts starting with “8_”). - **8_plot-recovery_smallsample.R:** Script to plot the parameter recovery results per parameter. The estimated average posterior means are plotted against the true parameter values for each participant. The plots also include the correlation between estimated and true values, and the average coverage over participants. - **8_plot-recovery_smallsample_collapsed.R:** Script to produce Figure 6. The difference between this script and 8_plot-recovery_smallsample.R is that in this case the parameters are grouped by parameter type. So all rate parameters and all threshold parameters are combined in one pane. - **/data** - **Singlet0_postMeans.RData:** Posterior means resulting from fitting the model to the data individually. These values are used as true values in the parameter recovery study to ensure realistic settings. - **/young** - **/grid-output:** Folder to store the samples objects. Sample objects are not added due to size limitations, but can be generated using the scripts provided. - **young_plot_param-recovery.rds:** Data produced by 7_data-plot-recovery_smallsample.R and can be used as input to 8_plot-recovery_smallsample_collapsed.R to recreate Figure 6. - **dmc.zip:** Dynamic Models of Choice package (DMC; Heathcote et al., 2019) used for the parameter recovery study. - **/figures** - **/young** - **young_plot_param-recovery_collapsed.pdf:** Figure 6. - **young_plot_post-means_caveats.pdf:** Figure 7. - **young_post-means_part[participant number]_n960.pdf:** Figure 3-20 of the Supplementary Materials. - **/grid** - **young_part[participant number]_n960.R:** Scripts for each participant to draw samples from the posterior distribution on the grid. This procedure is automatically performed for each of the 200 data sets. Note: each script takes about a day to run on the grid.