Figures & data
Fig. 1 Boxplots of the original creative writing scores by treatment group. The dot represents the mean of each group.
![Fig. 1 Boxplots of the original creative writing scores by treatment group. The dot represents the mean of each group.](/cms/asset/2f060241-aa6a-4297-85d8-91c1788401d5/ujse_a_1920866_f0001_c.jpg)
Fig. 2 A lineup consisting of 19 null plots generated via permutation resampling and the original data plot for the creative writing study. The data plot was randomly placed in Panel #10. Can you discern the difference?
![Fig. 2 A lineup consisting of 19 null plots generated via permutation resampling and the original data plot for the creative writing study. The data plot was randomly placed in Panel #10. Can you discern the difference?](/cms/asset/1f900e08-62cf-479a-b148-62dc28f4a623/ujse_a_1920866_f0002_c.jpg)
Fig. 3 A lineup of residual plots. The null plots are generated via a parametric bootstrap from the fitted model. The observed data are shown in Panel #9. Can you discern the difference?
![Fig. 3 A lineup of residual plots. The null plots are generated via a parametric bootstrap from the fitted model. The observed data are shown in Panel #9. Can you discern the difference?](/cms/asset/dd06994e-6729-429b-afa0-77c6c2d3ea54/ujse_a_1920866_f0003_b.jpg)
Fig. 4 A lineup for a deficient logistic regression model. The data plot is simulated from a model with a quadratic effect, while the null plots are simulated from a model with only a linear effect. The observed residuals are shown in Panel #8 and are indiscernible from the field of null plots, showing the problematic nature of residual plots for logistic regression.
![Fig. 4 A lineup for a deficient logistic regression model. The data plot is simulated from a model with a quadratic effect, while the null plots are simulated from a model with only a linear effect. The observed residuals are shown in Panel #8 and are indiscernible from the field of null plots, showing the problematic nature of residual plots for logistic regression.](/cms/asset/9de23912-7032-405c-93c4-5dc3326a01b7/ujse_a_1920866_f0004_b.jpg)
Fig. 5 A binned residual plot from a binary logistic regression model. The average deviance residual is plotted on the -axis for each of 31 bins on the
-axis.
![Fig. 5 A binned residual plot from a binary logistic regression model. The average deviance residual is plotted on the y-axis for each of 31 bins on the x-axis.](/cms/asset/668d4eb0-dcd1-450d-bcf6-5339af1710db/ujse_a_1920866_f0005_b.jpg)
Fig. 6 A lineup of binned residual plots from a binary logistic regression model. The observed residuals are shown in Panel #3 and clearly stand out from the field of null plots, indicating a problem with linearity.
![Fig. 6 A lineup of binned residual plots from a binary logistic regression model. The observed residuals are shown in Panel #3 and clearly stand out from the field of null plots, indicating a problem with linearity.](/cms/asset/30adcb37-091a-4ef7-ac12-4f478e53abdc/ujse_a_1920866_f0006_b.jpg)