forward, overcoming in the process conceptual and practical problems that seemed daunting only a decade ago. In view of these examples, and others like them, it hardly seems unduly optimistic to hope for similar gains in other substantive areas of political science in the decade to come.
Quantitative political methodology in the past decade has been a strong growth stock. We see no signs of a downturn in the intellectual vitality of the field or in the quantity or quality of research. On the contrary, there is every indication that further investment will pay even greater intellectual dividends. But what sort of investment is required at this stage? We see the same three glaring needs that Achen identified a decade ago. In each case there has been notable progress since he wrote, but in each case the underlying problem remains far from solved.
First, as Achen ( 1983a, 87) argued a decade ago, political methodologists and political theorists alike need to develop "formal theories with measurement models built into them." Where such theories have been developed, as in the fields of dimensional analysis, survey response, event counts, polytomous choice, and some areas of time-series analysis, significant methodological progress has followed. In the field as a whole, however, there is still far too much data analysis without formal theory -- and far too much formal theory without data analysis.
Second, "to provide for the future, much better mathematical training will have to be provided at both the graduate and undergraduate level" ( Achen 1983a, 89). Our traditional patterns of undergraduate training demand reform; as the research literature throughout the discipline becomes more sophisticated methodologically (and theoretically), it will be increasingly difficult to pretend that an undergraduate curriculum with little or no formal analytical training can provide more than a superficial caricature of the field. The problem is even clearer at the graduate level, with the added difficulty that we must also devote sufficient intellectual resources to train new and more sophisticated generations of methodological specialists. As long as otherwise reputable political science departments rely upon economists, psychologists, and sociologists to supply their graduate coursework in methodology, it will continue to be the case that "the pages of our journals frequently have the look of living rooms decorated at garage sales" ( Achen 1983a, 70).
Third, as Achen ( 1983a, 89) also argued a decade ago, "fundamental research must come to take priority, at least some of the time, over applied work. It should be possible to get funding, publish respectably, and make a career studying the principal agenda of political methodology." In this respect there has been greater progress in the past decade. Young methodological specialists who were graduate students or assistant professors when Achen wrote now fill senior positions in leading political science departments; their students are entering the field's junior ranks in increasing numbers; and their teachers continue to publish not only respectably but with great distinction.
Progress on these fronts has been, and will undoubtedly continue to be, stimulated by the institutional development of political methodology as a subfield. The Methodology Section of the American Political Science Association, a goal when Achen wrote, is now a functioning organization with hundreds of dues-paying members. Its annual publication, Political Analysis ( Stimson 1990, 1991b, 1992; Freeman n.d.), publishes a significant fraction of the best research in the field. And its newsletter, The Political Methodologist, is an increasingly ambitious forum for news, reviews, and intellectual debate. Summer methodology conferences at Michigan ( 1984), Berkeley ( 1985), Harvard ( 1986), Duke ( 1987), UCLA ( 1988), Minnesota ( 1989), Washington University ( 1990), Duke ( 1991), and Harvard ( 1992) have provided an important new forum for presenting and discussing methodological research. Many of the articles described in this review were first presented at these conferences. With funding from the host universities and the National Science Foundation, recent conferences have included dozens of graduate students in addition to faculty members from across the United States.
Nevertheless, any survey of the field must recognize that the number of political scientists working and thinking primarily as methodologists is still exceedingly small and that, in the long run, the number and energy of the field's part-time adherents cannot adequately compensate for a dire shortage of trained, committed specialists. As long as most political science departments have more Latin Americanists, urbanists, and voting behavior specialists than methodologists it cannot be surprising that we know more about the vicissitudes of politics in Nicaragua, New Haven, and the New Deal coalition than about the central methodological problems of our discipline. And as long as colleagues in other fields count anyone who does applied statistical work as a political methodologist, and vice versa, we cannot be surprised that our principal methodological agenda remains unconquered.
The authors thank Simon Jackman, Gary King, Bradley Palmquist, and Paul Sniderman for constructive criticism and advice,