Preparing test data
Bringing item data, metadata, and checks together so that scaling and reporting stay traceable.
Educational Assessment / R / Methods
Educational Assessment, R, and Data Quality.
I work on tools and materials that make educational data easier to handle: test item data, metadata, IRT scaling, and reproducible workflows in R.
On GitHub, the visible thread is a calm methodological one: R packages, Shiny applications, TBA data preparation, and materials from the methods team.
Areas of work
Between data, models, and good documentation.
This page does not show everything, but it shows the throughline: robust preparation, auditable analyses, and materials that help others work.
Making models computable
Tools and examples for Rasch, 2PL, and partial-credit models in educational assessment.
Sharing methods
Materials, websites, and methodological notes that make R workflows more accessible across a team.
GitHub selection
Code and methods material
A small selection from visible repositories and fields of contribution.
eatPrepTBA
Wrappers and routines for TBA studies: accessing IQB Studio and IQB Testcenter APIs, preparing data from multiple sources, and bringing it together.
BT-ShinyApp
Interactive results display for the Bildungstrend studies: an R/Shiny application with a map of Germany, used as the basis for the IQB results browser.
eatPrep
Prepare test item data for IRT analyses: preparing data, checking structures, and making scaling workflows more reliable.
eatModel
Educational Assessment Tools for IRT Modeling: functions for Rasch, 2PL, partial-credit, and generalized partial-credit models.
eatRep
Replication methods for large-scale assessments: means, percentiles, GLMs, plausible values, jackknife, bootstrap, and multiple-imputation pooling.
eatAutoCode
Tools for automated coding in educational assessment: preparing text responses, using models, and making coding workflows more reproducible.
eatGADS
Metadata, SPSS imports, and data management for educational assessment: the part of the work that makes analyses dependable in the first place.
IQB-Methods
Materials from the methods team: R-SIG, Quarto pages, methodological notes, and reusable resources.
karolines.de
And the old archive stays.
The start page now foregrounds the professional work; the old images, record notes, and found pieces remain as a small, idiosyncratic side wing.
Art and old files
A deliberately small selection of older images and project traces, from portraits to clay figures.
Record notes
Three short, very time-bound music notes from the earlier karolines.de.