EPA interim final report on WTC background study, April 2003

NYC 9/11 Public Portal Document

Table 8-3. An evaluation was conducted to determine how the estimated UCLs were effected by assigning 1/2 of the detection limit for samples that were below the detection limit (DL). In general, the variation between the different methods (i.e., ND=0, ND=1.2, and ND=DL) are directly related to the frequency of detection. As illustrated below, the lower the frequency of detection, the greater the variability, with values ranging from 100% less to 100% greater for data sets that contained all values below the detection limit. The use of 1/2 of the detection limit results in a mid­ point estimate. Freq. Of ND=0 ND=l/2 ND=DL Percent Difference Percent Difference Analyte Surface Detection (%) UCL UCL UCL ND=0/ND=l/2 ND=DL/ND=l/2 Lead BulkPast 100% 186 186 ? 0% Total Dust Hard Surface 100% 14.4 14.4 0% Lead Hard Surface 49% 1.69 1.78 1.88 -5% 5% Asbestos PCM-Air 42% 0.0022 0.0023 0.0025 -6% 7% Alpha-quartz Hard Surface 38% 68.3 79.6 91.8 -14% 15% Asbestos Hard Surface 13% 5,462 6,192 6,965 -12% 12% MMVF Hard Surface 10% 29 52 76 -44% 46% MMVF An- 8% 0.034 0.060 0.088 -44% 45% Lead Porous Surface 8% 1.00 1.98 2.95 -49% 49% Asbestos PCMe 5% 0. 2527 0.00024 0.00044 -72% 80% Asbestos TEMAHERA-Au- 4% 0. 2527 0.00024 0.00043 -72% 80% Calcite Hard Sur&ce 3% 60.8 132.3 232.3 -54% 76% Lead Air____________ 2% 0.0068 0.0304 0.0549 -78% 81% Cristobalite Hard Sur&ce 1% 12.2 103.7 200.5 -88% 93% Calcite Air 0% 0 93 18.6 -100% 100% Cristobalite Air____________ 0% 0 93 18.6 -100% 100% Gypsum Air 0% 0 93 18.6 -100% 100% Gypsum Hard Sur&ce 0% 0 49.9 99.8 -100% 100% Portlandite An- 0% 0 93 18.6 -100% 100% Portlandite Bulk Dust 0% 0 4.69 937 -100% 100% Portlandite Hard Surface 0% 0 99.8 199.7 -100% 100% Tridymite An____________ 0% 0 93 18.6 -100% 100% Tridymite Bulk Dust_______ 0% 0 4.69 937 -100% 100% Tridymite Hard Surface 0% 0 99.8 199.7 -100% 100%

rhese data sets were not included in the above table because the frequency of detection or the variability in the three values that were calculated were altered by additional factors, thus they are reported separately:

These values are also influenced by a change in distribution fiom either non-parametric to lognormal or lognormal to non-parametric due to altering the values between ND=0, ND=l/2, and ND=DL. This required different statistical methods to be used which resulted in the comparison of the three values not being similar to comparing values that were calculated using the same statistic. fThese values are also influenced by the way that the frequency of detection is calculated, as there are 17 different dioxin congeners. This results in the comparison with the other data being dissimilar as they are based on the fiequency of detection of a single compound. For example, one sample may have detected one congener, while a second sample detected 15 congeners. Each sample would be considered to t)e above the detection limit, however the values used to estimate the UCL would be altered for 16 congeners in one sample but only for 2 congeners in the second sample, thus creating greater vaiibility from the other data sets.

♦ These values are also influenced by a change in variability within the data set due to altering the values between ND=0, ND=l/2, and ND=DL. This changed the standard deviation of the data set which required a different confidence interval to be used to estimate the UCL. The comparison of values that use different confidence intervals resulted in the comparison not being similar to the other data sets. Freq. Of ND=0 ND=l/2 ND=DL Percent Difference Percent Difference Analyte Surface Detection (%) UCL UCL UCL ND=0/ND=l/2 ND=DI7ND=l/2 Alpha-quartz Bulk Dust* 89% 3.88 3.66 4.97 6% 36% Dioxin Hard Surfitcef ___77%__ OW 0^ 1.330 -87% 92% Alpha-quartz Air;:_______ 25% 45.5 61.9 48.0 -26% -22% Calcite BulkDust^ 11% 0.61 3.41 7.05 -82% 107% Gypsum Bulk Dus? 11% 1.42 2.33 4.71 -39% 102% Cristobalite Bulk Dus? 0% 0 4.69 10.70 -100% 128%

NYC-WTC_000145955