roc_toolbox_v1.1.4

Update to correct minor bugs.

• Fixed bug in dpsd_gen_sim_data, msd_gen_sim_data, and uvsd_gen_sim_data that threw an error when checking for the 'ignoreCond' option.
• Added 'verbose' optional argument to roc_solver.
• Updated plot_roc_summary to avoid plotting 'NaN' values in the 'bootIter' option was not used (i.e., bootstrapped standard errors were not estimated).

roc_toolbox_v1.1.3

Minor Bug Fix:

• when fitting the MSD model, the parameter for dprime1_lure was incorrectly assigned in the output. The value was incorrectly set to be identical to the most conservative response bias parameter. This had no impact on the underlying model fit, and the correct parameter value (if fitting the MSD model with a version prior to v1.1.2) can be recovered by accessing the msd_model.opitmization_info.bf_pars part of the data structure output by roc_solver.m.

roc_toolbox_v1.1.2

Fixed a minor bug in the msd_gen_pars.m function that returned an error when trying to initialize the x0, LB, and UB variables.

roc_toolbox_v1.1.1

This release adds a function to check the version of the toolbox, adds some alpha code for estimating the posterior distribution of the parameter estimates, and updates the manual and tutorial 1.

Changes/Updates:

1. Updated Tutorial 1 to include an example of the non-parametric bootstrap technique for estimating the standard error of the parameter estimates.

2. Added the function roc_version.m to report the version of the toolbox that is currently downloaded.

3. Updated the manual document to be consistent with changes that have been implemented since v1.0.0.

4. Added a function to compute the likelihood value and scrip to estimate the posterior distribution of the parameter estimates to the folder in_alpha. The purpose of this is to begin developing a set of functions that can be used to flexibly define priors for parameters in models to estimate posterior distributions, as well as to potentially develop a routine for Bayesian estimation of the parameters using MCMC. Please contact us if you are interested in helping develop this.

roc_toolbox_v1.1.0

This release of the ROC Toolbox includes a non-parametric bootstrap method to estimate the standard errors (SE) of the parameter estimates from fitting a signal detection model to ROC data.

roc_toolbox_v1.0.2

UPDATE 7/1/16: The standard errors of the parameter estimates output by this release, which use the Hessian matrix, are incorrect. This is due to an issue with the Hessian matrix calculation in FMINCOM. An updated version of the toolbox (1.1.0) will be released in the next week which using a non-parametric bootstrap routine to estimate the SE of the parameters.

This release fixes some minor bugs in the code, and adds computation/presentation of the standard error of the parameter estimates.

Changes:

1. Updated roc_solver.m to include calculation of the SE of the parameter estimates using the Hessian matrix output by fmincon.m.

2. Updated plot_roc_summary.m to show the SE in the summary plot.

3. Fixed a bug in msd_calc_pred_data.m that caused roc_solver to crash when the MSD model was fit to the data.

roc_toolbox_v1.0.1

This is release 1.0.1 of the ROC toolbox. Only two minor changes were made:

1. A minor issue with the formulas for the mixture signal detection model that, in some circumstances, could render the model under identified was fixed.

2. A new input was added to the simulate data function to allow for seeding the random number generator, which allows randomly generated data sets to be reproducible.

roc_toolbox_v1.0.0

This is the initial release of the ROC Toolbox for Matlab.