This preview shows pages 1–6. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: POISSON REGRESSION Example applying the Poisson distribution A retrospective occupational followup study was conducted among workers of a telecommunications company in the State of New York. The authors wanted to test the null hypothesis that there was no association between exposure to magnetic fields and the occurrence of breast cancer among men. Male telephone workers, who were actively employed in the company during 19761980, were matched to the cancer registry of New York to identify breast cancer cases. Personyears and cases were counted only if the person was still working for the telephone company during the study period. The annual rate of breast cancer among all men in New York was 0.15/100,000 during the study period. Two cases were observed among 50,582 male telecommunications workers who had contributed 206,067 person years of followup (Matanonski GM, et al. Electromagnetic field exposure and male breast cancer, Lancet 1991; 337: 737). 2 Summary of data Observe 2 cases over 206,067 person years Expect (under H ): 206,067[(0.15)/100,000] = 0.31 cases over 206,067 person years Expected cases = (Total person time) (Rate) = L = expected events L = person time = underling rate 3 Poisson distribution Let Y be a random variable which can take on the values 0, 1, 2, 3, (e.g., Y = the number of cases) P(Y=y) = ! y e y  where y=0, 1, 2, 3, The Poisson distribution is defined in terms of a single parameter ( ) Once a value of is assumed, then the distribution is specified 4 Male breast cancer example (continued) Under H (no association between male breast cancer and magnetic field exposure): = 0.31 Let Y = the number of breast cancer cases, assume Y is distributed Poisson P(Y=0) = !...
View
Full
Document
This note was uploaded on 01/06/2011 for the course CHE 6270 taught by Professor Dr.dmitry during the Spring '10 term at University of Florida.
 Spring '10
 Dr.Dmitry

Click to edit the document details