NAEP Process Data & Accessibility.
Click-stream measurement for digital assessment.
An IES Innovation Grant investigating how accessibility features in NAEP digital mathematics assessments are actually used by students with disabilities — and how process-data signals predict outcomes that proficiency scores obscure.
Accessibility features aren't a checklist. They're data about what learning requires.
Assessment accessibility is usually treated as compliance: does the test have text-to-speech, magnification, contrast, breaks, extended time? Our project asks the sharper question. When those features are available, who actually uses them, and how does their engagement pattern predict the outcome?
Uptake varies sharply by feature, learner group, and item type. The process-data signal predicts outcomes that proficiency scores alone obscure — reshaping how we evaluate whether a test is fair.
Read the paper Research areaAccessibility features are not a checklist. They are data about what learning actually requires — and who the current test is failing to measure.— NAEP Process Data guiding principle
Assessment, rethought.
What shifts when accessibility moves from a compliance audit to a measurement data source.
Accessibility as a checklist
- Features listed in test documentation
- One-size-fits-all default settings
- Pass/fail audit for availability
- No insight into who uses what
- Score gaps treated as proficiency gaps
Accessibility as data
- Click-stream capture of feature usage
- Differentiated design by learner group
- Engagement patterns analyzed per item
- Process data predicts outcomes
- Test design changes based on findings
Four methodological fronts.
Each track uses NAEP process data differently — and feeds back into test-design recommendations.
Accessibility Feature Uptake Analysis
Which of the five NAEP accessibility features (text-to-speech, magnification, color contrast, breaks, extended time) are actually used by students with disabilities — and how usage varies by item type, learner group, and session duration.
Response Timing
Item-level response timing and its relationship to accessibility feature use — surfacing patterns proficiency scores obscure.
Outcome Prediction
Testing whether process-data signals (engagement patterns, feature use, timing) predict mathematics performance for students with disabilities better than pre-test accommodations alone.
Test Design Recommendations
Translating process-data findings into concrete recommendations for NAEP and next-generation digital assessments — differentiated design, default settings, and accessibility documentation.
Published from this project.
Building an assessment accessibility grant?
RISEI Lab serves as named evaluator on IES, NSF EDU, and Spencer Foundation grants involving digital assessment, accessibility research, and process-data methodology. Bring us in at the proposal stage.
Start a partnership →