STP1333: Comparison of LC50 Results from Commonly Used Computer Programs

    Sebaugh, JL
    Statistical consultant and owner, Sebaugh's Information Services, Lake Ozark, MO

    Pages: 15    Published: Jan 1998


    Abstract

    Using 50 data sets, this paper compares LC50s computed by several, commonly used computer programs. Programs evaluated include: Stephan, TOXSTAT®, ToxCalc™, U.S. Environmental Protection Agency (PROBIT and TSK), and this author's SAS® program. Numeric comparisons are made of both the LC50 estimate and its 95% confidence limits. An evaluation was made of the robustness of the programs to unusual data sets, the ease of use, and the clarity with which the programs provide results or error messages. The SAS LC50 program provided estimates for the greatest number of data sets. The probit LC50 point estimates provided by the EPA program and the SAS LC50 program were in the greatest agreement. TOXSTAT had the greatest disagreement with the other programs for the probit confidence limits. There was more agreement among the TSK estimates than among the probit estimates. Both EPA programs and the Stephan program require manual data entry with no ability to store the data. If a Windows® operating system environment is desired, either the SAS LC50 or the ToxCalc program is useful. Since SAS LC50 runs in the SAS environment and ToxCalc runs in the Excel™ environment, each is easier to use if one is already familiar with the respective environment.

    Keywords:

    LC50, EC50, software, probit, trimmed Spearman-Karber


    Paper ID: STP12177S

    Committee/Subcommittee: E47.06

    DOI: 10.1520/STP12177S


    CrossRef ASTM International is a member of CrossRef.