ABSTRACT
Logistic regression for matched case-control studies is largely used to analyse data retrospectively from case-control studies where a disease D is present for m individuals (patients) and absent for a further m individuals (controls). This chapter investigates the propensity for a patient with explanatory variables. The choices of pairing were based upon two stratification variables, namely age and gender. The regression function was modelled, using Splus online and during a subjective discussion with an opthamology expert, who at each opportunity recommended the next variable to include in the analysis. This provides an excellent example of Aitken’s inductive synthesis. The chapter concludes that the propensity for cataracts is related to a variety of factors interacting simultaneously, and not to any particular factor or pairwise interaction of factors.
