
Title
Estimating the number of classes using auxiliary information
Speaker
Lynne Stokes, Department of Management Science and Information Systems, University of Texas
Abstract
I discuss methods of estimating the number of classes (species) in a finite population using a sample of units from the population. Previous estimators using sample information alone have not performed well, so here I discuss a method of making use of auxiliary information about the observed classes to improve estimation. The estimator is similar to Alho's estimator of population size from capture-recapture data. It generalizes estimators of number of classes in an equal-class-size population by developing a regression model linking the size of observed classes and class characteristics. I discuss uses of the method for improving the performance of database software and for measuring disclosure risk.