**10. Uncertainty in qualitative analysis**

Several measurements performed in forensic toxicology are qualitative in nature. These measurements typically take the form of a binary response (i.e., pass/fail, yes/no, over/under, present/absent, etc.). They are classification in nature where materials are assigned to discrete groups based on measurement results. Diagnostic tests are one important example of qualitative analyses. Their qualitative results are important indicators of whether some specified threshold has been exceeded or not and are important for the determination of further confirmatory analyses. In some cases the measurement system will respond simply with binary results (green light/red light). At other times the measurement system is quantitative on a continuous scale which can be dichotomized. For example, a prearrest breath test instrument employing a fuel cell might measure the breath alcohol on a continuous concentration scale but is interpreted as being greater than or equal to 0.080 g/210L or less than 0.080 g/210L. In either case, the response is considered binary and thus qualitative. The uncertainty associated with qualitative analyses has received much less attention than that of quantitative analysis. The uncertainty in qualitative analyses is basically probabilistic in nature - that is, we are interested in the probability of being correct in our decision. We are concerned primarily with the probability of false positive and false negative results. While there are a number of statistical methods for estimating the uncertainty associated with qualitative or diagnostic test results, there is no consensus as to which is to be preferred. (EURACHEM/CITAC, 2003, Pulido, et.al., 2003, Ellison, et.al., 1998) Some methods involve the simple determination of false-positive (FP) and falsenegative (FN) fractions which in turn assess the probability of making a wrong decision. (Pepe, 2003) Other qualitative and quantitative methods employ Baye's Theorem which is argued by many as a superior approach to estimating and interpreting measurement uncertainty. (Gleser, 1998, Weise and Woger, 1992, Kacker and Jones, 2003, Phillips, et.al., 1998). Space does not permit further discussion of these important and useful methods.
