Subsequently, he received an Open Venning Exhibition to study physics, specializing in quantum mechanics, at Worcester College, Oxford University.
He also taught popular courses in applied quantitative methods to more than 3,000 graduate students from departments throughout Harvard University and the Massachusetts Institute of Technology.
In 1990, under the auspices of the Harvard Seminar on Assessment—along with his colleagues Judith D. Singer and Richard J. Light—Willett authored the book By Design: Planning Research on Higher Education.
The book was written to facilitate the conduct of superior research in higher education and was dedicated specifically to the proposition that "you can't fix by analysis, what you bungled by design."
In 1991, Willett and his collaborators Richard J. Murnane, Judith D. Singer, James J. Kemple and Randall J. Olsen published a comprehensive portrait of the careers of more than 50,000 teachers who were serving in America's public schools, based on extensive discrete-time survival analyses of their longitudinal teaching records.
In 2003, Willett and his close collaboratorJudith D. Singer, authored their seminal volume entitled Applied Longitudinal Data Analysis: Modeling Change and Event Occurrence.
Its main thesis was that—to document the importance and impact of education effectively—one needed to analyze systematically collected longitudinal data on the participants in the process, whether they be students, parents, teachers or administrators.
To rectify this, in the book, the authors provide a detailed presentation of the methods of individual growth modeling and survival analysis, respectively, using research questions and data-examples drawn from the field.
Most recently, Willett and his colleague, Richard J. Murnane, published a book that presents and describes improved methods for making causal inferences from empirical data in social and educational research.
The book is organized around important substantive research questions in education and uses detailed accounts of exemplary research from a wide variety of fields to describe the optimal design of true experiments, to introduce the concept of natural experiments and regression-discontinuity strategies, to describe the rationale and implementation of instrumental-variables estimation and lay out stratification and propensity-score methods for making causal inferences.