Laserfiche WebLink
APPENDIX D sEcnoN 1 DESCRIPTION OF STATISTICAL ANALYSES <br />The corrected mean and standard deviation are then used to determine the tolerance limit <br />as described previously. <br />This test statistic is not useful if the mean is slightly larger than the detection limit or if there <br />are multiple reported detection limits. When these conditions interfered with the <br />application of Cohen's method, the nondetect values were replaced with one-half the <br />detection limit and the tolerance interval was calculated. <br />Test of Proportions <br />If less than 50 percent but not fewer than 10 percent of the values are above the detection <br />limit, a test of proportions was used to compare the background data with the detection <br />well data. The test determines if the proportion of detects in the detection well differs <br />-significantly from the proportion in the background well. if there is a significant difference, <br />there is statistically significant evidence of contamination. This test requires at least five <br />detected values and five nondetects. <br />To perform the test, the proportion of detects in the background well Pb and in the detection <br />well Pd are calculated. The standard error of the difference in proportions is calculated as <br />shown below, where B and nb are the number of detects and the total number of samples in <br />the background monitoring well, and D and nd are the number of detects and the total <br />number of samples in the detection monitoring well. <br />S, = [[(B + D)/(nI, + nd)]*[1.-(B + D)/(nb + nd)]*[1/ nb + 1/ nd JjSz <br />The test statistic is then calculated as follows: <br />Z = (Pb - Pd/ S, <br />~ The absolute value of Z is compared with 1.96, the 97.516 percentile from the standard <br />normal distribution. If the value exceeds 1.96, there is statistically significant evidence at the <br />5 percent significance level that the proportion of detects in the detection monitoring well <br />exceeds the proportion in the background monitoring well. <br />References <br />United States Environmental Protection Agency, Office of Solid Waste. April 1989. <br />Statistical Analysis of Ground -Water Monitoring Data at RCRA Facilities:Interim Final <br />r - Guidance. PB89-151047. <br />ROD/APPD4097.DCC <br />