Gerhard Tintner’s memorial survey presents Oskar Anderson’s statistical work as a distinctive contribution to the methodology of the social sciences. Published after Anderson’s death in February 1960, the article combines a biographical introduction, nine thematic sections, and an extensive bibliography of Anderson’s publications. Its central concern is how Anderson connected probability theory with empirical inquiry while questioning the suitability of prevailing statistical assumptions for economic and social data. Tintner’s commemoration is also an argument for keeping this methodological alternative available to Anglo-American readers.
Anderson’s career links intellectual traditions as well as national institutions. Educated in mathematics, physics, economics, and law in Russia, he worked with A. A. Tschuprow, taught in Russia, Bulgaria, and Germany, and participated directly in agricultural sampling censuses. Tintner identifies this movement between theoretical instruction and practical investigation as part of his significance:
During his very distinguished career, he provided a link between the Russian school of statistics, from which he originated (Markoff, Tschuprow) and the Anglo-American school.
The article places Anderson within a “continental” tradition associated with Lexis and von Bortkiewich, potentially nearing its end as German textbooks increasingly adopted Fisher’s and Neyman’s approaches. Yet Tintner immediately qualifies this division into schools: their members communicated, and Anderson both appreciated Anglo-American achievements and questioned their applicability to social science. His distinctiveness therefore lies less in allegiance to an isolated tradition than in his scrutiny of the conditions under which statistical procedures become meaningful.
This concern is clearest in the section on probability. Anderson’s “social-statistical probability” refers the frequency of a characteristic in a given population to a higher-order population from which that population has been taken. Application requires specifying the relation between these populations; the higher-order population may be finite or infinite according to the circumstances. Probability is thus connected to a concrete account of how the observed population arises. Tintner then explains Anderson’s “Cournot bridge” between mathematical probability and practical inference. An empirical proposition—that very improbable events occur infrequently—is combined with mathematical results showing that sufficiently large samples make substantial deviations from expectations less probable. Together they justify expecting small deviations frequently when observations are numerous. The conceptual move is to distinguish formal theorems from the empirical warrant needed to apply them.
Survey sampling gives this argument an operational form. Tintner describes Anderson’s pioneering involvement in agricultural surveys in Russian Turkestan and Bulgaria, including the Bulgarian agricultural sample census of 1926 and annual surveys begun in 1936.
In his many theoretical contributions to the subject he stresses the point of view that the sample census must be based upon a probability model. The level of tolerance and the desired accuracy of results should be fixed in advance.
Sampling is consequently more than an economical substitute for complete enumeration. Its credibility depends on a specified model and advance decisions about acceptable uncertainty. Anderson’s practical experience supports, rather than merely illustrates, his methodological insistence on explicit foundations.
The longest technical discussion concerns the Variate Difference Method, Anderson’s contribution best known in America and a field in which Tintner himself worked.
The Variate Difference Method is an attempt to deal with time series while making a minimum of assumptions.
Tintner makes the qualification behind this aspiration clear. The method assumes a smooth systematic component, encompassing trend and longer cycles, with superimposed independent, non-autocorrelated random errors. Successive finite differences eliminate a polynomial systematic component or progressively reduce a suitably behaved one. Anderson developed formulas comparing the variances of consecutive difference series to help determine when sufficient differencing had isolated the random component. Tintner situates these results within subsequent improvements and his own investigations of their sampling distributions.
The account is appreciative without suppressing objections. Higher differences may be inaccurate, while autocorrelation and short periodic fluctuations can invalidate the procedure. Nevertheless, Tintner defends its continuing relevance to economic series containing trends. The widespread econometric use of first differences is presented as a practical adaptation: it removes a linear trend, or an exponential trend in logarithmic data, and can sometimes reduce autocorrelation. Anderson’s legacy here survives partly through a routine technique rather than broad adoption of the complete method. His criticism of the Harvard approach to economic time-series analysis likewise exemplifies methodological intervention that helped displace less efficient procedures.
The shorter sections establish the breadth of Anderson’s applied work. Tintner highlights statistical investigation of the quantity theory of money, divergent agricultural and industrial prices, studies of Bulgaria’s interwar economy, and the introduction of game theory to German-speaking readers. Work on production and cost-of-living indices combined practical construction with scepticism toward aggregation approaches and Wald’s ideas. In correlation analysis, Anderson’s objection to normal-population assumptions in social science led him toward distribution-free significance tests for correlation and autocorrelation coefficients. Across these topics, methodological criticism generates alternative procedures rather than simply rejecting statistical inference.
Tintner closes with Anderson’s textbooks, praising their clarity and restrained mathematical demands. His recommendation that the latest methodological textbook be translated gives the memorial a forward-looking purpose:
Because of the difference of some of Anderson's ideas about statistics from the prevailing Anglo-American school which is very well brought out in this text, a translation might stimulate discussion on these problems.
The concluding bibliography supplies the documentary basis for further study. Ultimately, the survey preserves Anderson’s work as a resource for reconsidering statistical practice: mathematical sophistication must remain answerable to population definitions, data characteristics, and the particular requirements of social-scientific explanation.
This work was divided into 6 sections when it entered the library's research corpus—an apparatus for search and citation, not necessarily the author's own table of contents. Each title opens its summary.
Put a question to this work; the Librarian answers from its 6 sections and cites the passage.
Ask the Librarian