Figure 1
Schematic presentation of an experimental trial. Participants first saw trial-specific information about the involved planet, categories, and feature. The contingency table then appeared. After 5000 ms, the generic sentence appeared beneath it. Then, after an additional 3000 ms, the justifiability question also appeared and participants responded on a six point scale: (–3) very unjustifiable, (–2) unjustifiable, (–1) rather unjustifiable, (1) rather justifiable, (2) justifiable, (3) very justifiable.
Figure 2
Histograms displaying predicted data patterns for the simulated empirical data and for each model. A) Normalized justifiability ratings provided by simulated participants, if participants rate justifiability using P(f|G)–P(f|¬G). B) Normalized justifiability ratings provided by a model where justifiability is predicted by absolute prevalence. C) Normalized justifiability ratings provided by a model where justifiability is predicted by a subtraction. D) Normalized justifiability ratings provided by a model where justifiability is predicted by a general fraction. E) Normalized justifiability ratings provided by a model where justifiability is predicted by a specific fraction. F) Normalized justifiability ratings provided by a model where justifiability is predicted by the fraction used in the model of Van Rooij and Schulz (2020).
Figure 3
Scatterplots comparing simulated data to simulated predictions. A) justifiability rated by simulated participants when using P(f|G)–P(f|¬G). B) Fit of simulated data and justifiability predicted by absolute prevalence. C) Fit of the simulated data and justifiability predicted by a subtraction. D) Fit of the simulated data and justifiability predicted by a general fraction. E) Fit of the simulated data and justifiability predicted by a specific fraction. F) Fit of the simulated data and justifiability predicted by the fraction used in the model of Van Rooij and Schulz (2020).
Figure 4
Histograms displaying the empirical data (N = 175) alongside the predicted data patterns for each model as was pre-registered (see Figure 2). A) Normalized empirical justifiability ratings provided by participants (error bars indicate 95% confidence intervals, corrected with the Cousineau-Morey method to take into account within-subject variability). B) Normalized justifiability ratings provided by a model where justifiability is predicted by absolute prevalence. C) Normalized justifiability ratings provided by a model where justifiability is predicted by a subtraction. D) Normalized justifiability ratings provided by a model where justifiability is predicted by a general fraction. E) Normalized justifiability ratings provided by a model where justifiability is predicted by a specific fraction. F) Normalized justifiability ratings provided by a model where justifiability is predicted by the fraction used in the model of Van Rooij and Schulz (2020).
Figure 5
Scatterplots comparing empirical data (N = 175) to simulated predictions as pre-registered (see Figure 3). A) empirical justifiability ratings. B) Fit of simulated data and justifiability predicted by absolute prevalence. C) Fit of the simulated data and justifiability predicted by a subtraction. D) Fit of the simulated data and justifiability predicted by a general fraction. E) Fit of the simulated data and justifiability predicted by a specific fraction. F) Fit of the simulated data and justifiability predicted by the fraction used in the model of Van Rooij and Schulz (2020).