ABSTRACT
This chapter offers a classification of data-dredging procedures and some comments on their use. In survey analysis, which is commonly exploratory, it is rare for precise hypotheses to be formulable independently of the data. Snooping procedures, or at least such as have so far been developed, are of little practical use to the survey practitioner as opposed to the experimentalist. Whereas snooping predesignates all tests to be made, and fishing uses a predesignated set of candidate variables, HUNTING involves no predesignation: the data are simply dredged for information in the area of interest. A more elaborate form of fishing appears in many computer programs for regression, in which the independent variables are screened by the program to see if they account for a worthwhile proportion of the variation in the dependent variable; those that fail to meet this test do not appear in the final regression equation.
