It is an article of faith for social scientists that substantive considerations should guide the choice of research topics and dictate the selection of analytic tools, never the reverse. Most scholars share a certain disdain for colleagues who learn a new statistical technique and then apply it willy-nilly to every set of data within reach. Although we would not want, in general, to quarrel with the venerable maxim that technique should be the slave and not the master of substance, exceptions exist. In the Middle Ages a tremendous amount of scientific and creative energy was invested in the attempt to transform base metals into gold—surely an activity of great substantive interest but, ultimately, no payoff. On the other hand, vast new areas of scientific achievement were opened when someone built a better telescope and turned it on the stars or when cathode ray equipment, developed for other purposes, led to the discovery of X-rays. As Thomas Kuhn (1962) notes, increased precision in instrumentation is usually undertaken to test hypotheses under an existing paradigm. Yet these technical advances in measurement occasionally lead to surprising anomo-lies that open up entirely new substantive areas.