Table 5. Correlations between parameters used to generate simulated data and recovered parameters for 8 fitting methods and three sets of numbers of observations for the numerosity discrimination design.
N=1000, 1000 | N=100, 100 | N=40, 40 | |||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Method | a | Ter | v1 | v2 | a | Ter | v1 | v2 | a | Ter | v1 | v2 | |
0% contaminants | HDDM | 0.991 | 0.972 | 0.952 | 0.960 | 0.782 | 0.877 | 0.356 | 0.571 | 0.426 | 0.791 | -0.129 | 0.154 |
Fast-dm | 0.971 | 0.971 | 0.932 | 0.926 | 0.893 | 0.910 | 0.843 | 0.774 | 0.878 | 0.851 | 0.573 | 0.605 | |
DMATout | 0.913 | 0.888 | 0.810 | 0.921 | 0.755 | 0.808 | 0.265 | 0.591 | 0.493 | 0.719 | -0.041 | 0.366 | |
DMATno | 0.980 | 0.957 | 0.902 | 0.916 | 0.798 | 0.854 | 0.586 | 0.651 | 0.379 | 0.650 | 0.091 | 0.337 | |
MLH | 0.987 | 0.977 | 0.937 | 0.894 | 0.954 | 0.923 | 0.835 | 0.647 | 0.923 | 0.890 | 0.743 | 0.665 | |
Chi9q | 0.982 | 0.961 | 0.908 | 0.934 | 0.943 | 0.942 | 0.829 | 0.743 | 0.899 | 0.852 | 0.668 | 0.643 | |
Chi5q | 0.979 | 0.961 | 0.921 | 0.951 | 0.948 | 0.917 | 0.822 | 0.746 | 0.899 | 0.871 | 0.650 | 0.636 | |
EZ | 0.976 | 0.928 | 0.926 | 0.929 | 0.896 | 0.804 | 0.866 | 0.798 | 0.821 | 0.745 | 0.741 | 0.670 | |
4% contaminants | HDDM | 0.667 | 0.939 | 0.602 | 0.752 | -0.126 | 0.880 | 0.181 | 0.360 | 0.478 | 0.790 | 0.089 | 0.332 |
Fast-dm | 0.946 | 0.968 | 0.786 | 0.838 | 0.876 | 0.859 | 0.783 | 0.684 | 0.873 | 0.847 | 0.629 | 0.546 | |
DMATout | 0.930 | 0.930 | 0.674 | 0.838 | 0.525 | 0.771 | 0.221 | 0.494 | 0.352 | 0.666 | -0.026 | 0.471 | |
DMATno | 0.862 | 0.941 | 0.649 | 0.745 | 0.619 | 0.765 | 0.597 | 0.626 | 0.333 | 0.626 | 0.115 | 0.405 | |
MLH | 0.975 | 0.975 | 0.844 | 0.830 | 0.894 | 0.872 | 0.755 | 0.636 | 0.763 | 0.862 | 0.599 | 0.613 | |
Chi9q | 0.987 | 0.963 | 0.910 | 0.932 | 0.941 | 0.935 | 0.848 | 0.765 | 0.867 | 0.866 | 0.629 | 0.516 | |
Chi5q | 0.967 | 0.938 | 0.854 | 0.903 | 0.930 | 0.907 | 0.807 | 0.786 | 0.835 | 0.858 | 0.691 | 0.594 | |
EZ | 0.804 | 0.760 | 0.766 | 0.849 | 0.675 | 0.640 | 0.814 | 0.778 | 0.643 | 0.575 | 0.611 | 0.623 |
Note: DMATout was the DMAT method with contaminant correction, DMATno was the DMAT method with no contaminant correction, MLH was the maximum likelihood method, chi9q and chi5q were the chi-square methods with 9 and 5 quantiles respecitvely. a is boundary separation, Ter is nondecision component of response time, v1 and v2 are the drift rates for the easy and difficult conditions respectively.