Document 7KRRqvwnZRop26XgMbQeG2rE

GRACE TO: R. C. Walsh FROM: A. L. Stockett cc: H. A. Eschenbach J. W. Wolter Construction Products Division DATE: July 30, 1984 On July 18 and 19 1 met with Ben Armstrong at McGill University in Montreal to discuss his preliminary report on the Libby exposure data analysis. Ben reviewed his methodology with me, and I am satisfied that he is using techniques which are appropriate* He pointed out that the preliminary analysis followed procedures which Corbett McDonald had used in previous analyses of similar data sets. However, he expects to modify somewhat the breakdown intervals for his groupings in future analyses to obtain a more uniform distribution of the data. In response to our question regarding the number of points to plot on the dose-response type graph, Ben explained that limitations are imposed by the use of a standardized mortality ratio. This statistic requires some group averaging to become stable, and he feels that no more than four groups (i.e. plotting points) can be effectively used with only 23 total data. In response to our question regarding the horizontal axis plotting locations used on the X-ray opacities graph, Ben conceded he had used what he felt were close enough approximations. As you suggested I called to ask him to not use an approximation for this graph. I will touch base again with Ben after he returns from his two weeks vacation. ALS/j s 10001*73 GRACE TO: R. C. Walsh FROM: A. L. Stockett cc: H. A. Eschenbach J. W. Wolter m r Construction Products Division DATE: July 30, 1984 On July 18 and 19 I met with Ben Armstrong at McGill University in Montreal to discuss his preliminary report on the Libby exposure data analysis. Ben reviewed his methodology with me, and I am satisfied that he is using techniques which are appropriat^. He pointed out that the preliminary analysis followed procedures which Corbett McDonald had used in previous analyses of similar data sets. However, he expects to modify somewhat the breakdown intervals for his groupings in future analyses to obtain a more uniform distribution of the data. In response to our question regarding the number of points to plot on the dose-response type graph, Ben explained that limitations are imposed by the use of a standardized mortality ratio. This statistic requires some group averaging to become stable, and he feels that no more than four groups (i.e. plotting points) can be effectively used with only 23 total data. In response to our question regarding the horizontal axis plotting locations used on the X-ray opacities graph, Ben conceded he had used what he felt were close enough approximations. As you suggested I called to ask him to not use an approximation for this graph. I will touch base again with Ben after he returns from his two weeks vacation. ALS/j s A. L. Stockett 10001^79 I