SDT Assistant
With the publication of the third edition of Detection Theory: A User’s Guide, SDT Assistant has been updated to perform many of the analyses that are detailed in the book.
SDT Assistant has four main modules that are designed to undertake different types of analyses. Full details and examples are available in the manual, which is included in the download file.
|
SDT Assistant Manual - V. 2.0 - 2021.pdf Size : 951.729 Kb Type : pdf |
ROC Analysis for Rating Data
Rating data can be entered into a grid or via file. Models include: equal-variance (EV) and unequal-variance (UV) normal for yes-no ratings; EV normal for 2AFC ratings; EV normal for yes-no reminder ratings (both difference and independent-observation strategies); EV normal for 2AFCR ratings (both difference/independent-observation and absolute-difference strategies); EV normal for same-different ratings (both difference and independent-observation strategies). SDT Assistant provides best-fitting parameters, their variance and covariance, goodness-of-fit information, and various other statistics.
Estimates of d′ and Associated Dispersion Statistics from Binary Responses
Binary data can be analyzed with respect to all EV normal models identified above, and in addition, the EV normal model for the ABX task (both difference and independent-observation strategies). A direct calculation method is employed, and statistics discussed in Chapter 13 of Detection Theory: A User’s Guide are reported. These include the statistical bias, variance, and mean-square error of d′.
Generate Data to Plot Normal Density Functions and ROC Curves
SDT Assistant can produce curve data, in the form of a text file containing (x, y) coordinates, that can later be imported into graphing software. Parameters for the curves are easy to enter. ROC curves for which plot data can be generated include all models specified in the previous sections. In most cases, the only parameter required is d′. For the normal-normal SDT model the intercept and slope parameters are required.
Extend the Tables of Appendix 5 of Detection Theory: A User’s Guide
Tables of the form presented in Appendix 5 of Detection Theory: A User’s Guide present a selection of all possible values that may be encountered. Interpolation is the usual practice for values in between those tabulated. SDT Assistant allows all values to be calculated directly without interpolation.
Finally
While the use of the software is free of charge, it would be appreciated if you could mention SDT Assistant to other users, and suggest that your institution’s library purchase a copy of Detection Theory: A User’s Guide. You may find benefit from a copy of the book as well. We would like the methods of Detection Theory to be disseminated broadly, and are asking for your help to do this.
If you find the software useful, please send a brief endorsement to me (less than 100 words), along with details of your institution. Institution badges and/or user endorsements may be placed on the website (hautus.org) to show where SDT Assistant is being used.