Mortality history and the misleading mean.

he global transition to higher life-expectancy levels doubled the length of the average human life in less T than a century (Easterlin 1996, 69). Initially, life expectancy at birth values for most countries, Western and non-Western, were in the 20-to-40-year range; at present, they are in the 60-to-80-year range. For decades, social scientists have investigated this biological revolution without reaching a consensus about its causes. As Roger Schofield (1991,2) wrote in his introduction to The Decline of European Mortality, “It would only be a small exaggeration to say that our understanding of historical mortality patterns, and of their causes and implications, is still in its infancy.” If he is correct, we are dealing with a case of very prolonged infancy, because research on the decline of mortality was conducted throughout the twentieth century, and an enormous amount of solid factual information has been accumulated. The absence of an interpretative consensus after so much research reflects the covertly political nature of health and mortality history. Specific interpretations of the past have specific policy implications for the present that lead politically involved researchers to put a “spin” on the data (Kunitz 1987; Johansson 1994, 1996). But continuing controversy also reflects professional rivalry among social scientists who work in specialized fields of knowledge. More often than not, academics from different departments tend to treat each other as competitors instead of partners in the production of knowledge about the past; their rivalry encourages the habit of ignoring or dismissing evidence and arguments produced by the other side. This mentality discourages genuine interdisciplinary research in every aspect of demography and legitimizes the existence of highly artificial explanatory dichotomies (Lesthaeghe 1998). In mortality history, the artificial dichotomy currently generating more heat than light concerns the relative importance of better nutrition versus more public health in the world’s earliest mortality transitions, that is, those that began in Western Europe (see the essays in this issue by Haines and Kintner, Woods and Shelton, and Johansson). But both sides agree on the appropriateness of using national-level data to test competing hypotheses to the exclusion of doing more localized research. Before 1950, nationallevel research included some recognition of the importance of local differences to general explanations (Kreager 1997); but, subsequently, as the demography of currently developing countries came to be based on the analysis of nationallevel data and little else (Hill 1997), so did historical demography. Anthropological demographers objected to the increasing reliance on macro-level data and continued to do smallscale, micro-level population research.’ By now, they have exposed the empirical pitfalls of relying on national-level data as the basis for understanding fertility behavior (Kertzer and Fricke 1997). Most developing countries contain a wide variety of small-scale populations with distinctive demographic patterns; all their variation gets lost at a high level of aggregation. To the extent that local variation contains significant information about how real people perceived and responded to the disruptions and opportunities created by development in specific contexts, that information gets lost in national-level research, For decades, some historians have raised similar objections to national-level demographic history. As long as twenty years ago, Michael Anderson’s (1980, 30-32) Approaches to the History of the Western Family warned that the statistical preoccupation with mean household size in national populations was damaging family history. That history was much too complex to be reduced to such simple terms. Recently, Keith Wrightson (1998, 16) wrote that “there is no such thing as the history of the English family.” There are only specific groups of families coping with specific concerns in specific contexts in early modern England as elsewhere. Those who use national-level means to explain fertility history have also fared badly since Ansley Coale and Susan Watkins (1986) summed up the findings of the Princeton