This paper discusses the screening of effluent toxicity by comparing survival proportions between effluent-treated and control groups of test organisms. If the effluent is nontoxic, survival proportions should be similar for both groups. However, because of differences in sensitivity of individual test organisms, these proportions will not necessarily be identical. Furthermore, the number of organisms surviving from test to test will vary, even though identical effluent samples and identical laboratory procedures are used for all tests. This paper describes a statistical test for determining when effluent and control group survival proportions are significantly different, thereby indicating a potentially toxic effluent. The paper emphasizes the statistical power of the test for detecting effluents of various toxicities. It also evaluates the probability that an effluent will be incorrectly judged to be toxic.

The four key statistical parameters that determine the usefulness of the test are described: statistical power (i.e., the probability of correctly recognizing a toxic effluent); Type 1 error rate (i.e., the probability of incorrectly judging an effluent to be toxic); effect size (i.e., the true but unknown difference between effluent and control long-run survival proportions); and sample size. (Graphs showing the relationships between these four parameters are included.) As the number of organisms used in the test is increased, the power of the test also increases. An experimental design procedure using this test and the statistical power-sample size tradeoff is also described.