Statistical Evaluation of Truncated Breath-Alcohol Test Measurements

    Volume 33, Issue 2 (March 1988)

    ISSN: 0022-1198

    CODEN: JFSOAD

    Published Online: 1 March 1988

    Page Count: 4


    Gullberg, RG
    Sergeant—breath test section supervisor, Washington State Patrol Crime Laboratory, Seattle, WA

    (Received 7 April 1987; accepted 6 July 1987)

    Abstract

    The reporting of breath-alcohol measurements truncated to two decimal places is a form of computational error. The magnitude of the error can range from 0.000 to 0.009 g/210 L. The truncation error will follow a uniform distribution. A total of 500 breath-alcohol test measurements were evaluated to determine the distribution of the third digit. There are 10 possible discrete values for the third digit. The frequency of each third digit was found to range from 44 for the lowest to 57 for the highest. The data closely approximated the uniform distribution. To conform exactly with the uniform distribution, there would have to be 50 of each decimal value. Given that the third digit approximates the uniform distribution, one cannot attach a greater probability to a particular third-digit value as opposed to another.


    Paper ID: JFS11964J

    DOI: 10.1520/JFS11964J

    ASTM International
    is a member of CrossRef.

    Author
    Title Statistical Evaluation of Truncated Breath-Alcohol Test Measurements
    Symposium , 0000-00-00
    Committee E30