De-identified Usability and Assessment Data for "STACK in an African Open University: Automated Mathematics Assessment, Usability, and the Limits of Training"

This dataset contains the de-identified data supporting the findings of the article "STACK in an African Open University: Automated Mathematics Assessment, Usability, and the Limits of Training." Data were collected through a user feedback survey administered to students and instructors using the STACK automated assessment system in Mathematics, Statistics, and Physics courses at the Open University of Kenya (n = 715 respondents). The dataset comprises anonymised survey responses covering overall experience, ease of use, usefulness of immediate feedback, usefulness of question randomization, frequency and purpose of use, training received, perceived improvement in understanding, and open-ended feedback. Likert-scale items are scored from 1 (Very Poor) to 5 (Excellent). All direct identifiers (respondent names and email addresses) and full response timestamps have been removed, and each respondent is assigned an anonymous identifier. Free-text fields were screened and contain no embedded personal identifiers. Summary tables providing the values behind the reported means, standard deviations, and figures are included. Raw, identifiable student-level data are not shared due to participant-privacy and institutional data-protection requirements. Files: stack survey deidentified.csv (full response dataset), table likert summary.csv (descriptive statistics per item), fig likert distributions.csv (per-score counts), fig categorical frequencies.csv (categorical frequencies), and README.txt (documentation).