STP891

    Advantages of Using Regression Analysis to Calculate Results of Chronic Toxicity Tests

    Published: Jan 1985


      Format Pages Price  
    PDF Version (164K) 11 $25   ADD TO CART
    Complete Source PDF (6.2M) 11 $57   ADD TO CART


    Abstract

    Although results of most chronic toxicity tests have been calculated using hypothesis testing to detect statistically significant differences from the control, calculation of results using regression analysis offers several major advantages. By deriving an equation for the concentration-effect relationship, regression analysis allows estimation of the concentration corresponding to a specified amount of an adverse effect or estimation of the amount of effect corresponding to a specified concentration. In addition, whereas the number of replicates, agreement between replicates, and selection of α (the acceptable level of Type 1 error) have no impact on the estimate of the endpoint concentration calculated using regression analysis, they have a great impact on decisions concerning whether tested concentrations caused statistically significant differences. Further, use of regression analysis emphasizes the necessity of considering the biological importance of the adverse effects observed in toxicity tests. By using hypothesis testing, aquatic toxicologists have been able to avoid this difficult, important question by presenting results in terms of a variable “statistically significant difference.”

    Keywords:

    aquatic biology, regression analysis, statistical analysis, aquatic toxicology, chronic tests, hypothesis testing


    Author Information:

    Stephan, CE
    Environmental scientist, U.S. Environmental Protection Agency, Environmental Research Laboratory—Duluth, Duluth, MN

    Rogers, JW
    Statistician, Center for Lake Superior Environmental Studies, University of Wisconsin—Superior, Superior, WI


    Paper ID: STP33584S

    Committee/Subcommittee: E47.07

    DOI: 10.1520/STP33584S


    CrossRef ASTM International is a member of CrossRef.