In recent years there has been a rapid increase in pancreatic cancer. One method of searching for clues to the causes of cancer is the case-controlled epidemiologic survey. Investigators ask patients with a particular disease questions about their family and personal history, education, occupation, and medical history and ask the same questions of persons without the disease. Computers are programmed to store the data, search for differences in the patient and control group, and calculate the probability that the observed differences are due to chance. Such studies are successful only if the control group is chosen properly and the questions asked include some that ultimately prove relevant to the causes of the disease.
By such a method, Lin and Kessler (p 147) have identified several factors that may cause pancreatic cancer, and some of these seem to be different in men than in women. They created a model for estimating