ABSTRACT
The statistical consultant spends much of his time in the borderland between statistics and the other aspects, philosophical and substantive, of the scientific search for explanation. This chapter focuses on several problems dealing with the interplay of statistics with the more general problems of scientific inference. Searching for causal factors among survey data is an old, useful sport; and the attempts to separate true explanatory variables from extraneous and 'spurious' correlations have taxed scientists since antiquity and will undoubtedly continue to do so. The classic statistical tests give clear answers only to some simple decision problems; often these bear but faint resemblance to the complex problems faced by the scientist. The differences among experiments, surveys and investigations are not the consequences of statistical techniques; they result from different methods for introducing the variables and for selecting the population elements.
