Colloquium

Department of Statistics

Dissertation Defense
October 3 (Fri) 11 a.m.
Alavi Commons Room, 6625 Everett Tower

Inference on Differences Between k Means for Data with Excess Zeros and Detection Limits

Haolai Jiang
Department of Statistics
Western Michigan University

Many data have excess zeros or values falling below detection limit. For example, data on hospitalization costs incurred by members of a health insurance plan will have zeros for those who did not get sick. Benzene exposure measurements on petroleum refinery workers might fall below detection limits. Traditional methods of analysis like one-way ANOVA are not appropriate since the point mass at zero violates typical assumptions.

For testing the equality of k distributions, we propose a likelihood ratio test that accounts for both excess zeros and observations below detection limits. We conduct a simulation study on its finite sample properties under both lognormal and gamma distributions. An imputation method is proposed as an alternative approach.

Past colloquiums