Statistical considerations of the random selection process in a drug-test program.

Abstract

In a prospective drug-testing program, individuals whose job classifications have been defined as sensitive are placed in a selection pool. Periodically, individuals are chosen from this pool for drug testing. Random selection is a fair and impartial approach for this because each individual has an equal chance of being selected. Random selection also serves to reinforce the deterrent potential of a drug-testing program because individuals in the pool never know when or if they will be selected. This uncertainty discourages drug usage because of the risk involved in being selected and tested. A random selection process generates a Poisson distribution of probabilities that can be used to predict how many times an individual will be selected during a specific time interval. This information can be used to model the selection part of a drug-testing program to determine whether specific conditions of testing are met. For example, the probability of being selected a given number of times during the testing period can be minimized or maximized by varying the frequency of the sampling process. Consequently, the Poisson distribution and the mathematics governing it can be used to structure a drug-testing program to meet the needs of any given situation.