Laserfiche WebLink
T Appendix <br /> ® Section <br /> Description of Statistical Analyses1 <br /> 1 <br /> The statistical analyses used to determine if a discharge from the landfill is impacting <br /> groundwater in the detection monitoring wells were selected in accordance with EPA <br /> guidance(EPA, 1989)and State of California Regulations (23 CCR 2550).The procedure is <br /> outlined in Figure D--1 and each step is discussed below. <br /> Identification of Outliers , <br /> -- The test for outliers is to determine if there is statistical evidence that an observation <br /> appears extreme and does not fit the distribution of the remainder of the data. Any <br /> observations identified as outliers were assigned a value of"4"in the Lab Flag field in the <br /> - database and were not used for statistical analysis. <br /> To identify outliers,the statistic T„was calculated as shown below,where X,,is the largest ' <br /> observation,X is the mean of the observations,and S is the standard deviation. <br /> T"; K'-X)/S <br /> The statistic T„is compared with a critical value for the sample size(EPA,1989). The sample <br /> was identified as an outlier if it met three conditions: (1) the statistic exceeds the critical <br /> value, (2) the observation is at least four times greater than the mean,and (3) the <br /> — observation was at least four times greater than the next highest values. <br /> Test for Normality <br /> The statistical procedures recommended in the EPA guidance and California regulations are 1 <br /> robust with respect to departures from many of the normal distribution assumptions(EPA, <br /> 1989). Extensive study of the distribution is not considered necessary unless a <br /> nonparametric method of analysis-is selected. <br /> The coefficient of variation was used to determine if the assumption of a normal <br /> distribution was appropriate.The coefficient of variation is the standard deviation,S, <br /> divided by the mean of the observations,X. If the coefficient of variation exceeds one, there 1 <br /> is evidence that the data are not normal and the assumption of a normal distribution is not <br /> -- appropriate for the data set. <br /> If the data set appears to not be normally distributed, the residuals for the data set will be 1 <br /> checked for normality.If these are normal,the distribution of the data set will be assumed <br /> normal and the pararneteric test will be selected.If the residuals are not normally <br /> distributed,the data set will be transformed and checked for normality.If the transformed <br /> data set is normal, the parametric procedure will be selected.If the transformed data set is <br /> not normal,a nonparametric procedure will be used for the data set. <br /> 1 <br /> RDD/APPD4o97.Doc 1 <br />