VOC sampling locations table for Ground Zero area, Jan 2002

NYC 9/11 Public Portal Document

airway responsiveness studies, univariate linear regression variance ratios were greater than 10-fold then all of the was used in all subsequent tests. The models used in responses were ranked from smallest to largest across all these regression studies were analysis of covariance TRT and DAY combinations. Then ranks replaced the (COV) with tests for parallelism for each TRT and DAY original responses for the univariate ANOVA. Sometimes combination. In experiment A, natural logarithms were the variances of the ranks for TRT and DAY combinations used for both Meh concentration (C) and PenH AUC still indicated heterogeneity. Then additional judgment responses. The linear regression Log (PenH AUC) = al + was used to help insure that this heterogeneity of variance b(Log C) reduces to a power function of the form PenH = did not affect the overall conclusions. In experiment A, a aj C*’. No logarithms were used in Experiments B and C replicate was called DAY. When replication was shown due to several negative values resulting after baseline to have no significant contribution in the ANOVA results, adjustments. Techniques similar to ordinary stepwise DAY was not included in subsequent ANOVAs. For regression were used in COV analyses. Overall test of responses with many “zero” values, the residuals from the parallelism of regression lines was done first. Subgroups ANOVA were plotted and analyzed by univariate ofthe TRT and DAY combinations were determined to get techniques to determine if the residuals generally met the subgroups exhibiting a common slope. Within a subgroup assumptions required for ANOVA. When interactions with a single slope, subsequent tests were done to between TRT and DAY occurred, these were pointed out determine if the means were different (using individual and in some cases further ANOVA were done for each contrast tests). Body weight was determined DAY. When ranks were used for the response, the ranks repeatedly for animals. Due to animals being removed and were regenerated for each day separately. When TRT was killed, numbers of mice in each group were different on significant, follow-up comparisons of means were done different days after exposure. For days when few animals using Tukey’s multiple comparison tests. weights were determined, univariate analysis of variance The statistical tests examined only whether groups (ANOVA) was used to test for TRT effects. For those were significantly different from each other. In the days where most of the weight data occurred, MANOVA reporting of the results, for the sake of brevity, groups are techniques were used for statistical tests. sometimes referred to as having significantly greater For each TRT and DAY combinations with a values than other groups. These statements should be read univariate response, a determination was made if the as groups are significantly different from each other, and variances could be considered homogeneous. If the the mean of one group is greater than the mean of another.

13

NYC-WTC_000149156