Shiny app to measure the self-assessment of undergraduate students, using Likert scale questions, PRIDIT and t-SNE analysis.
Beyond the original PRIDIT/t-SNE pipeline, Gradquest adds:
- RIDIT – ref: group-referenced ridit comparison (e.g. by sex, academic performance, digital habits), with Mann-Whitney U significance testing per group.
- t-SNE × PCA robustness check: cross-validates t-SNE-derived clusters against clusters obtained directly on principal components, via Adjusted Rand Index.
- Response-style bias diagnostics: tests whether the composite score reflects genuine content differentiation or general response elevation, including ipsatization (person-centering) and double centering (person- and item-centering) of the response matrix.
Full changelog, interface documentation, and methodological references (with formulas) are in app_Shiny.md / app_Shiny.pdf.