Laserfiche WebLink
SECTION 1 93 <br /> In all analyses there is a spread of results around lth e�generally(TV). The <br /> ThestaMD <br /> eliminates the false positive by having the reporting g Y <br /> deviations above the IDL,thus any signal seen is real. However for analyte spikes at the <br /> MDL itself,50%of the obtained results are going to be less than the MDL(the shaded 31 <br /> area in the graph),and not reporting them constitutes a Type II error, a falsvehncelgaauvset at <br /> To avoid the problem many labs use Practical Quantitation Levels (PQL), <br /> a variety of amounts,the most common being 12 times the standard deviation resulting <br /> from the MDL procedure. Three different types of result now can be a detect but less than the1PQL iven or a non - <br /> laboratory: a numerical value above the PQL, <br /> detect. See Figures 1-23 and 1-24. <br /> Frequency <br /> of Result <br /> -3 SD TV +3 SD <br /> I <br /> Figure 1-23. Distribution of results around a true value. j <br /> I <br /> tip <br /> it <br /> FFrequency <br /> i 1 <br /> IDL MDL <br /> -3 SD +3 SD 4- <br /> Figure 1-24. False negative determinations at the MDL. <br /> MDLs are always linked to a standard sample size such as 200.0 mL of sample <br /> titrated for alkalinity(310.1), 1.00 L of sample extracted and concentrated to 1.00 mL <br /> for sermvolatile organics analysis(8270),or 5.00 mL sparged for GC-MS analysis <br /> (8260). Attempts to lower MDLs through increasing the processed sample size are <br /> generally not productive due to the simultaneous increase in background interferences. <br /> I <br /> 131 L.Keith,Environmental Laboratory,June/July 1992.pp.58-61. <br /> Genium Publishing Corporation <br /> `v; <br />