Home

The following contains a list of links to various HTML vignettes with the mirtCAT package, generated using R markdown via the knitr package and hosted on Github. For more details on using the mirt package, visit the mirt wiki.

Workshop in Jena, 2017

The following workshop presentation files will likely be useful to understand some of the features presented in the mirtCAT JSS publication (Chalmers, R. P. (2016). Generating Adaptive and Non-Adaptive Test Interfaces for Multidimensional Item Response Theory Applications. Journal of Statistical Software, 71, 1-38) as well as some of the newer features that were not present in that article (e.g., optimal test designs, shadowCATs, customised item selection methods, test/person information components, DOGS, etc).

• Workshop slides
• Examples
• Exercise questions without and with answers

Generating GUI Examples

• Simple Survey. Define a survey with only 3 Likert-type items, without any scoring/CAT definitions
• Unidimensional CAT. A unidimensional CAT session using Maximum Information as the selection criteria with EAP estimation
• Customizing a small number of GUI elements. Customize elements such as the title page, authors list, last page, include graphical item stems, etc, and include a pre-CAT section to collect information prior to beginning the CAT
• MCAT example. Define two factor MCAT using Drule selection and terminating when SE(theta) < .3 OR 50 items have been administered

Public hosting

• Hosting applications on personal server. Instructions on how to setup a personal server with Ubuntu to host mirtCAT applications.
• Hosting a survey on shinyApps.io. One potential way to host public surveys is with the shinyApps.io project, and this example stores the results to a local Dropbox folder as .rdata files

Offline Examples (no GUI)

• Define population IRT parameters. Run a simple offline unidimensional CAT using predefined CAT parameters
• MCAT example. Run a multidimensional CAT using predefined CAT parameters with dichotomous data
• MCAT example with graded response model. Run with Kullback-Leiber information and ML estimation
• MCAT with multi-unidimensional test structure. Defines an MCAT structure where items are administered in unidimensional packets rather than allowing for freely selecting items from the complete item bank. Makes use of the customNextItem() input.
• Shadow CAT. Demonstrates how to construct a Shadow CAT design (a.k.a., an adaptive test with an optimal selection criteria, but with a potentially large number of item selection constraints)
• Collecting unscored item responses from existing CAT bank. Highlights how to use mirtCAT to collect new item response data while still administering items in an adaptive session. Run off-line to demonstrate the mechanics.

Simulation examples

• Unidimensional 3PL model. Run a Monte Carlo simulation for a unidimensional CAT
• Customized item selection for different age groups. Run a Monte Carlo simulation for a unidimensional CAT using the customNextItem() and person_properties inputs for different age groups

User contributions

If you would like to contribute your own to the collection then feel free to send along your source file (.Rmd) to rphilip.chalmers@gmail.com and I'll add it to the appropriate package locations on Github.