VII. [Die Methode der standard population] (1905)

This published seminar report records Joseph A. Schumpeter’s lecture on the standard-population method; the reporter is unidentified. It explains procedures for adjusting mortality comparisons, then examines their causal assumptions and practical usefulness. Its central concern is whether standardization isolates the influences under investigation or removes conditions that partly constitute them.

Crude mortality coefficients register the combined effects of many causes. Although useful for practical purposes, they do not directly establish the general healthfulness of a population:

Im übrigen muß man sich, wie auf der Hand liegt und bis zur Banalität wiederholt wurde, vor Schlüssen aus ihnen, namentlich auf die Salubrität im allgemeinen, hüten.

English translation: For the rest, one must — as is obvious and has been repeated to the point of banality — beware of drawing conclusions from them, in particular as to salubrity in general.

Standardization responds to this limitation by holding selected population characteristics constant. Its analytical ambition is stated succinctly:

Der Grundgedanke unserer Methode ist nun, von einigen dieser Ursachen abzusehen, um andere zu untersuchen.

English translation: The fundamental idea of our method is now to abstract from some of these causes in order to investigate others.

The report first considers Körösi’s procedure for neutralizing differences in age composition. Since mortality varies markedly with age, populations can display different crude death rates without correspondingly different health conditions. A high proportion of infants, for example, can raise aggregate mortality. The procedure begins by establishing comparable categories:

Die Methode Kőrösis ist nun folgende: Man möge in allen Ländern die gleichen Altersklassen unterscheiden.

English translation: Kőrösi's method is now the following: let the same age classes be distinguished in all countries.

Each population’s observed age-specific mortality rates are then applied to a common distribution of age groups and combined into an index. Age composition is thus made uniform rather than literally eliminated. The report also explains Ogle’s correction factors, which provisionally assume identical age-specific mortality rates to estimate the effect of differing age structures. Following Westergaard, it limits the reliability of these factors to a specified period with a known age composition.

The principal theoretical question concerns the legitimacy of separating causes. Körösi distinguishes essential from accidental, or qualitative from quantitative, influences and defends adjustment for age through a contrast between natural and empirical laws. The report rejects this contrast: natural laws, too, are empirically established. The relevant distinction must instead concern the actual relationships among the causes being examined.

Where an age distribution is substantially accidental, as in certain occupational groups, adjustment can improve comparison. For a country or nation, however, age composition may itself belong to the complex of health conditions. Drawing on Bortkiewicz, the report emphasizes reciprocal effects: social circumstances influence age structure, while age structure can affect social circumstances and mortality. Standardization may therefore subtract part of the phenomenon it seeks to explain.

Extending adjustment to sex, marital status, religion, social structure, or climate intensifies this difficulty. Without a defensible criterion for distinguishing removable influences from constitutive conditions, there is no clear stopping point. Moreover, the same influence may operate differently across countries. A common standard does not by itself guarantee a valid causal comparison. The report nevertheless preserves a place for controlled isolation in investigations directed toward particular relationships.

Technical choices reinforce this qualified assessment. The division into age classes affects the index: finer categories improve precision but require more calculation, while broad classes can conceal important variation. Infant mortality demands particular care; simply excluding infants would be a crude substitute for adjustment. The selection of a standard population also matters, since different standards can change the proportions between mortality indices. Körösi’s comparison of European and Swedish standards produced only small changes and preserved the ordering, but this result does not settle the theoretical objection.

The conclusion distinguishes methodological possibility from practical necessity. Standardization is defensible where age composition is accidental and potentially valuable for specialized inquiry. Yet the examples discussed suggest that adjusted indices often differ little from ordinary mortality coefficients, which may suffice for many practical purposes. The report’s enduring methodological point is that numerical comparability requires causal justification: holding a characteristic constant does not establish that it is independent of the conditions being studied.