The workbench is a point-and-click app for people who would rather not write R code. It runs the same functions you would call yourself, and it writes the R code for everything you do, so you can rerun the analysis later or learn R along the way.
library(mirtverse)
launch_workbench()
# or start with your own data already loaded
launch_workbench(my_data)The seven steps
| Tab | What you do |
|---|---|
| 1. Data | Pick an example dataset or upload CSV, Excel, SPSS, Stata or RDS. Choose the item columns, enter an answer key to score multiple-choice items, and declare missing-value codes. |
| 2. Explore | Classical item statistics, Cronbach’s alpha and the total-score distribution. |
| 3. Model | Let mirtverse compare models and recommend one, or choose the item type and number of dimensions yourself. |
| 4. Results | Plain-language interpretation, item table with flags, parameters, all graphics, and item health cards. |
| 5. Scores | Person scores (EAP, MAP, ML or WLE) with standard errors, a CSV download, and the sum-score conversion table. |
| 6. Response times | Upload response times to fit the joint model and flag rapid guessing with the residual rule or the NT10, NT15 or NT20 normative thresholds. |
| 7. Report & code | Download the HTML report and the R script that reproduces the analysis. |
Sharing the app
The workbench is an ordinary Shiny app. To host it for colleagues or
students, put this in an app.R file and deploy it to
shinyapps.io or Posit Connect:
The workbench is a tool for doing an analysis. To show finished results to people who do not use R, build a dashboard instead: see Sharing results with a dashboard.