chapter  20
24 Pages

Inverse probability weighted methods

ByAndrea Rotnitzky

In most epidemiological and clinical longitudinal studies the intended (full) data — the data that the study investigators wish to collect — are incompletely observed. In modern studies, the intended data are typically high-dimensional, usually comprised of many baseline and time-varying variables. Scientific interest, however, often focuses on a low-dimensional parameter of the distribution of the full data.