Next to last class. I spent it on the material in the last chapter: false positives. The only numerical example I worked was the artificial one in the text, with false negative rate 1%, false positive rate 0.1% and an actual incidence rate of 1 in 10,000. Then the probability that a positive test means you have the disease is just about 9%.

Then I talked about two way contingency tables qualititatively in several contexts, each time looking at the costs associated with both false positives and false negatives: spam detection, terrorist screening at airports, earthquake or hurricane prediction, responding to indications that a terrorist attack is being planned.

With the ten minutes left at the end of the class I discussed the connection between odds and probabilities (with some simple numerical examples) and how odds are determined at the race track by the amounts bet on the various horses (pari-mutuel) . It went by so quickly I suspect most won’t remember it.

blog home page