Citation

Maxwell JC, Pullum TW. Eval. Program Plann. 2001; 24(3): 257-265.

Copyright

(Copyright © 2001, Elsevier Publishing)

DOI

10.1016/S0149-7189(01)00017-9

PMID

unavailable

Abstract

A Poisson regression applied the capture-recapture model to a time series of data for admissions to treatment from 1987 to 1996 to estimate the number of heroin addicts in Texas who are 'at risk' for treatment. A method was developed to summarize and use all of the data. The entire data set produced estimates which were lower and more plausible than those produced by drawing samples. But these better estimates were still sensitive to the assumptions of capture-recapture analysis: unique marking, independence, and consistency over time and population. The analyst must understand the local 'drug scene' and changes in treatment programs when making estimates.