ISSN:
1573-7845
Source:
Springer Online Journal Archives 1860-2000
Topics:
Sociology
Notes:
Summary This discussion has tried to illustrate that analytic induction as a technique used by qualitative researchers may be extended in several ways. This is not to suggest that numerical analogs should be used to supplant the purpose of analytic induction. The present analysis has merely tried to show, following Robinson's suggestions, the broad similarities between analytic induction and what Robinson calls enumerative induction. The transition from one to the other mode of analysis requires a trade-off, however; that is, numerical estimates of the analytic inductive model can be made only if the logical structure is modified. Thus, placing even a minimal value in cell 3 automatically precludes the possibility of stating the logically necessary condition for the phenomenon—which is the focal point of analytic induction. This sacrifice of logical necessity may be ameliorated somewhat by showing that if certain numerical values are assumed, then, from a statistical point of view, the strength of the relationship may suggest conditions under which both necessary and sufficient conditions may be fulfilled. Thus, even sophisticated statistical methods such as path analysis (Kerlinger, 1973) assume a logical relationship between the variables (usually the sufficient condition) without being overly concerned that the correlations between the paths are usually moderately large at best. Indeed, another question would be whether logical considerations in empirical research should even be of central interest. Of course, in a very minimal way, they are important in the sense of saying that variable A logically antecedes variable B. However, unless propositions of this kind are left at a very general level of analysis or, alternatively, are recognized to be tautologies (e.g., “if societies are stratified, then they will have occupational hierarchies”), the utility of strictly logical considerations is not central. Thus, enumerative induction stresses quantitative “proofs” over logical ones, while analytic induction stresses logical considerations over quantitative ones. The point made by Robinson, however, is that they are not contradictory approaches, at least at some levels of analysis. What relates the two approaches more directly, even though this is not made explicit by Robinson, is the assumption that the qualitative researcher must know the characteristics of his group and/or setting intimately. Only then will he be able to suggest the reasonableness of assuming other (quantitatively distributed) characteristics of his sample.
Type of Medium:
Electronic Resource
URL:
http://dx.doi.org/10.1007/BF02350812
Permalink