ABSTRACT
In a retrospective study of the possible effect of blood group on the incidence of peptic ulcers, B. Woolf obtained data from three cities. Those cities are London, Manchester and Newcastle. The example in this chapter illustrates a type of retrospective investigation widely used, in particular in epidemiology under the name case-control study. For fairly obvious reasons, it is convenient to collect data in an inverse fashion, taking a set of peptic-ulcer patients and a set chosen from non-peptic ulcer individuals and then observing blood group for each individual. The main assumption limiting the conclusions is that the control series can be treated as effectively randomly chosen from that part of the target population without peptic ulcer. Effective matching between ulcer and control groups might increase precision and any form of 'cluster' sampling could decrease precision that is increase variance.
