|
Abstract:
|
Often we can think of our data as an imperfectly observed version of a hypothetical complete data set in which some or all observations are known only imprecisely. When the degree of imprecision is so great that standard continuous-data models are no longer reasonable approximations, we say that the data are coarse. Some prominent examples of coarse data include censored data (e.g., survival times in clinical trials), grouped data (as incomes), and heaped data (data with multiple degrees of coarseness, such as reported ages or daily cigarette consumption). I will present a general analysis strategy for coarse data, discuss ignorability conditions, and illustrate the methodology with one or two live examples.
|
