Spot the flaw. Tabershaw and Lamm (1977) compared the observed and expected numbers of different leukemia types in a study group of workers who had been exposed to benzene during their employment. They were testing a previous suggestion that exposure to benzene increases the probability of acute leukemia while not changing the occurrence of other leukemia types. Their expected numbers are based on the relative
frequencies of these diagnoses in the population as a whole. The numbers are presented in the following table.

The researchers applied a  goodness-of-fit test to the data and calculated a  value of 3.33, with df=3.  From the information given, what is the largest error made in this analysis?

Transcribed Image Text:

Type of leukemia Chronic lymphocytic Chronic lymphocytic Chronic myelogenous Chronic myelogenous Monocytic Monocytic Acute Acute Observed 00 11 22 44 Expected 22 11 11 33