STP1007

    Head-to-Head Comparison of SAS and ASTM-Proposed Probit Computer Programs

    Published: Jan 1988


      Format Pages Price  
    PDF Version (88K) 5 $25   ADD TO CART
    Complete Source PDF (9.5M) 5 $81   ADD TO CART


    Abstract

    Estimates of the median lethal dose (LD50) [or the median lethal concentration (LC50)] are used in toxicological and hazard assessment studies to indicate the dosage of a substance that produces a 50% response (death) rate in a population. The calculation of LD50 estimates is frequently accomplished through probit analysis. Probit analysis is based on certain underlying assumptions about the tolerance of the population in question to the substance being tested. The response rate is assumed to follow a cumulative normal distribution with respect to the dosage or the logarithms of the dosage.

    Recently we reviewed ASTM's proposed standard practice for using probit analysis (Draft No. 2). A BASIC computer program for performing probit analysis was included with the proposed standard. We evaluated this program using the PROBIT procedure in the wellknown SAS Institute package as a standard of comparison. The SAS procedure offers a default option that assumes the response rate of the population follows the cumulative normal distribution. In addition, two logarithmic options are also offered (natural and base 10), whereby the response rate is assumed cumulative normal relative to the logarithm of the dose. The BASIC and SAS programs give very similar results when the appropriate logarithmic option is used with the SAS procedure. Thus, the ASTM program would be a subset of the SAS procedure in that it is comparable to one of the SAS options, employing one set of possible assumptions. We believe that the assumptions one would make about the distributional response of a population are not necessarily standard, and that careful examination of the data relative to analytical assumptions prior to analysis is recommended.

    Keywords:

    probit computer programs, aquatic toxicology


    Author Information:

    Bromaghin, JF
    Statisticians, Animal Damage Control Program, Denver Wildlife Research Center, Denver, CO

    Engeman, RM
    Statisticians, Animal Damage Control Program, Denver Wildlife Research Center, Denver, CO


    Paper ID: STP10297S

    Committee/Subcommittee: E47.01

    DOI: 10.1520/STP10297S


    CrossRef ASTM International is a member of CrossRef.