Research

Education Leadership Data Analytics

At Teachers College's ELDA group, I study how schools and districts can use the data they already collect more effectively, through evidence-based improvement cycles, analytics, dashboards, and interactive data visualizations.

Feedback Timing in a Real Medical Learning Platform

I embedded a randomized experiment into a medical learning platform to test how feedback timing and recalling an initial answer shape later learning and confidence.

This project is ongoing and centers on how rigorous experiments can be run inside real educational systems.

When Should Feedback Arrive?

In a meta-analysis of 51 studies, I found no simple overall winner between immediate and delayed feedback. What mattered more was the learning task, the learner group, and the kind of feedback being used.

The project helps clarify a long-running question in digital learning design and points to the need for clearer reporting across studies.

The Right Difficulty for Learning

Using data from medical students, I found that learning works best when training is neither too easy nor too hard. The optimal level depends on the learner and on the specialty being studied.

This project shows why adaptive systems should aim for the right amount of challenge rather than a one-size-fits-all approach.

Modeling Learning in Real Time

I adapted a learner model to estimate student proficiency and question difficulty in real time on a large medical learning platform, making adaptive support possible while keeping strong predictive performance.

I also showed that historical data can improve the model's early predictions when a system is first launched.

What Happens After a Decision Is Made?

This project explored how new evidence arriving after an initial decision can still change both the decision itself and the confidence attached to it.

It showed that people do not simply stop processing information once they respond; later evidence can still reshape both judgment and confidence.

Why Vision Feels Stable Over Time

This study examined serial dependence, the tendency for what we just saw to shape what we see next, helping visual experience feel smooth and stable rather than noisy.

The findings suggest that this effect depends partly on the response stage and not only on perception itself.