Document ymw4dY7pKgjrJo1OLwj48447r

cc: W. F. Armstrong C. B. Hunt ETHYL CORPORATION G- So,"lon INTF.R-OFFICF JTO: Dr. J. G. Burdick FROM: B. A. Sobel SUBJECT: Mortality, Study - Houston Plant 10/1/60 to 8/1/76 DATE: March 14, 1977 I have studied the subject report and agree with its basic conclusion: that Houston plant workers and general Texas death rates do not differ significantly. Actually, inspection of the data (Table 5-2, for example) indicates that the plant death rate is lower than for populations as a whole except for cancer, where 19 deaths were observed with 17*1. expected. * p = ig/14490 .001311 A * .000131 p H 17.1/14490 - .001180 s V 39as8a . .00030, 14490 Evidently, there is no significant difference. I have calculated some of the Chi-square values and found them to be non-significant at any rational level. 1 would have preferred that the paper include the statistical tests. The statement that "none of the tests were significant at the .05 level" does not preclude the statement "ail the tests were significant at the 0.20 level". This iatter would be of tremendous significance in interpreting the data. The loss of a few points, as apparently happened, did not affect the conclusions. This is almost always the case with large amounts of data - that an absent or aberrant point has neglibible effect on the conclusions. BAS:mf1 B. A. Sobel FC 3003