Olcay Akman
Department of Mathematics and Statistics
Coastal Carolina University
Statistical Inference Based on Weighted Models
In many situations, experimenters do not work with a truly random sample
from the population in which they are interested, either by design, or
because experimental conditions make random selection from the target
population impossible. The usual statistical analysis assumes that a random
sample from the original distribution is obtained; however, since the
observations do not have equal chances of entering the sample, the resulting
sampling distribution does not follow the assumed original distribution.
Statistical models that incorporate these restrictions are called selection
models, or weighted models. In this talk some examples and relevant
estimation properties under weighted models will be discussed.
Back to Colloquium Series