Toward More Cumulative Inquiry

My initial training in the systematics of social inquiry-cum scientific methodoccurred at a time when the pendulum had swung far in favor of deduction and pure theory. Mere inductive learning was seen as pedestrian at best, if indeed it stopped short of sheer numerology. The most common modifier of data-gathering was mindless. While the era was already empirical, no information was to be collected until a full-blown theory had been confected, with all of its corollaries and ramifying hypotheses smoked out by pure reason. Each hypothesis, so generated, was to be embodied in the marginal labels of a prefabricated tabulation sheet, well in advance of any instrumentation or data collection. Nothing was to be measured in the actual study which lacked some clear place in the web of predicted interrelationships. When the results were in, the data were less to be analyzed in a modern sense, than merely to be registered on the tabulation sheets, and as blindly as possible, against the chance that the roving eye might be quicker than the scientific conscience. What's more, upon completion of the last tabulation, the raw data were to be destroyed, for there was no telling what unsavory minds might otherwise fall upon them at a later date and be moved to wonder about their patterning in a less disciplined way. Anything, or so it seemed, to minimize the uncertainties of discovery. The inherited model was clear: a consuming reverence for Newton and Einstein (albeit both somewhat misconstrued), with a postscript or two prompted by early vulgarizations of Fisher, Neyman, and Wald. I thought this a curiously childish caricature of what most of science had ever been about. I was appreciative, even at the time, of the happy fact that Newton had been preceded by a mindless Brahe and a half-witted Kepler, and I have spent much of the intervening period vindictively collecting episodes as to what the other vital 90% of science has truly been about. Among other things, I wondered where theories and hypotheses could come from at all, save from observation of the real world; and if this be given, then I wondered what good it served to outlaw from the process of theory formation uniquely those observations of the real world which were most carefully collected and systematized. I also wondered, back then, how some few branches of the macrocosmic social sciences like economics and population demography seemed to display more consensual focus, more natural continuity of research, and more manifest cumulativity than the rest. There were ready answers to this curiosity, of course. It was obvious, for example, that both disciplines had rudimentary but highly serviceable units of analysis, like money and body counts. But it was also true, and in some degree a corollary, that both economics and population demography had been more handsomely munified than most neighboring social sciences with large series of data spanning many geographic jurisdictions and lengthy periods of time, less because of academic clairvoyance than because of administrative need to keep track of economic transactions and vital statistics. I was well aware that both disciplines had made noteworthy theoretical progress before these readymade data bases began to be much mined. But I wondered if the current cumulativity and the kinds of anchoring and ready reality-testing offered by these gift data bases were utterly coincidental. I rather doubted it. All by way of introducing the General Social Survey (GSS), the periodic measurements on a national-sample base begun by James A. Davis and his collaborators at NORC in 1972 with seed money from the Russell Sage Foundation, and currently being maintained through the good funding offices of the National Science Foundation. To my mind, the GSS is an important and happy development for the future of those areas of sociological researchers little serviced by the Census or other large-scale government data collections. It has its faults, but most of these spring more from external constraints than from basic conception, and in a broader scheme its limitations pale by comparison with its potentialities. In spite of my lengthy preface, my enthusiasm for the conception of the GSS does not arise simply because it is essentially an inductive vehicle. The fact of being inductive, once all context is set aside, is not to