Citation

Carlson WL. Accid. Anal. Prev. 1976; 8(3): 177-186.

Copyright

(Copyright © 1976, Elsevier Publishing)

DOI

unavailable

PMID

unavailable

Abstract

The problem of obtaining unbiased estimates using samples from sequential populations is examined using Monte Carlo simulation. These sequential populations are composed of superposed renewal processes. One example of a sequential population is the population of highway crashes within a geographic area over a specific time period. The simulation results indicate that samples at fixed or random count intervals provide unbiased estimates of the proportion of elements from each subpopulation, whereas samples at specific times which are defined at fixed or random time intervals result in biased estimates. A multiple regression analysis showed this bias to be related to the actual subpopulation proportions and the difference in the relvariances (squared coefficient of variation) of the subpopulation interarrival times. The results are shown to apply to an actual population of highway crashes obtained from a midwstern county over a one year time period.