Keyword

When the Bolshevik government came to power in October 1917, it based the legitimacy of its action on the scientific nature of its decisions, figures being one of the core elements thereof. Its leaders asserted the ambition of building a state in which science would form the basis for political decisions, for the well-being of everyone and fundamentally for the fulfilment of the communist plan to create a new society and new human being. From this perspective, statistics—an information and decision support tool—was also an instrument of power designed to prove the soundness of state action. Statistics would help symbolically construct the Soviet social and economic world. That notwithstanding, can one discern specific forms of production of figures in such a state, that is to say, as Alain Desrosières has conceptualized, a particular relationship between concepts, methods, technical instruments, statistical institutions and representations of the social and economic world, that could be construed as characterizing the nature of such a state (Desrosières, 1985)?

In particular, did certain tools take on a specific form in the USSR, in a context of construction of a socialist society underpinned by centralized and planned management of the economy? If that is the case, how were they adapted to new political representations of society, the economy and the role of science in such a state? This newly developed form of producing figures serving a new type of state and a new political plan included the creation of a new central government statistics department and an effort to formulate a new theory of statistics and design new methods and new tools. Various attempts to that effect gave rise to debates and tensions between leaders and statisticians, and between the statisticians themselves. After trying to characterize some of them, we will endeavour to underscore the specific nature of state statistics in the Soviet Union.

Inventing a New Form of Statistics for a New Model of Society

A New State Statistics Administration

The Bolshevik state’s new Central Statistics Administration, the TsSU (Tsentral’noe Statisticheskoe Upravlenie), was set up on 25 July 1918. One can consider that it was the product of two plans tending towards change, but plans in which the role of statistics was defined differently. For the Bolshevik leaders, the production of figures had to play a key information role for developing the plan and for managing the economy and society. But for the TsSU’s statisticians, most of whom were formerly employed by the statistical offices of the zemstva, local self-governed authorities of the Tsarist state provinces founded in 1864, the challenge consisted above all in creating the statistical institutions and tools of a modern state, in line with the recommendations of the international statistics congresses of the nineteenth century (Mespoulet, 2001; Mespoulet & Blum, 2003; on statistical internationalism in the late nineteenth century, see Brian, 1989).

So notwithstanding the political message conveyed by the Bolshevik government, the TsSU was organized on the model of the statistical agencies of late 19th-century European states, structured around a series of departments reflecting the main divisions of statistics of the time, and moreover constructed on the institutional and methodological bases of regional statistics as practised by the zemstva. The texts regulating its foundation and missions adhered to the spirit of the debates and resolutions of the international statistics congresses of the nineteenth century, as asserted by its director, Pavel I. Popov, speaking at the national congress of Russian statisticians held between 8 and 16 June 1918 (Popov, 1918). Encouraged by the experience of the statisticians in the regional offices of the zemstva, the Bolshevik state statistics administration was also resolutely organized in line with statistical internationalism, its scientific positions and organizational principles (on this point, see Brian, 1989).

This continuity with the pre-revolutionary period, relying also on continuity of the staff, was simply the visible part of other forms of inheritance in the representation of statistical work, methods and observational tools, and in the maintenance of certain administrative practices behind the apparent institutional changes (see Mespoulet, 2001, chapters 5 and 7). This strong continuity of individuals and practices coexisting with a disruptive political message was at the root of a great deal of tension between statisticians and political leaders from the early 1920s (see Blum & Mespoulet, 2003).

For the Bolsheviks, planned management, centralization and accounting had to be the bedrock of the organization of production in the future socialist state, and the statistics administration had to serve this political plan. While in line with the new plan model formulated by Lenin and the Bolsheviks prior to 1917, it should be noted that this conception of a centralized and planned economy was strengthened by the experiences of the First World War (see Holquist, 2002) and the civil war between 1918 and 1921 (Sapir, 1997), which provided a breeding ground for formulating and experimenting with a state-controlled economy (Stanziani, 1998), in particular to resolve supply chain issues.

In such a context, it is hardly surprising that the statisticians in charge of setting up the TsSU subscribed to this goal, in particular those who had played a part in the management of public and economic affairs during the war, in the Tsarist Ministry of Agriculture, or within the framework of the All Russian Union of towns and of the zemstva, or in the Provisional Government. However, such an endorsement of the Bolshevik plan did not preclude tensions and clashes between leaders and statisticians about the very definition of the role, methods and tools of state statistics. What form of quantification should be adopted to construct a socialist economic and social model tending towards communism? The first challenge consisted in how to define the connection between accounting and statistics.

A Complicated Demarcation Between Accounting and Statistics

The expression “socialist accounting” was used by Lenin as early as 1917 to describe the form of quantification aimed at providing figures to construct a socialist economy and society. In his speech to the Petrograd Soviet, Lenin said: “Socialism is accounting. If you want to record each piece of iron and fabric in the books of account, then that is socialism” (Lenin, 1955, vol. 26). Before the Revolution, he had already stressed the key role of accounting in his book entitled The State and Revolution: “Accounting and control, these are the chief things necessary for the organizing and correct functioning of the first phase of Communist society” (Lenin, 1955, vol. 21).

Initially, and for reasons of efficiency, the production of figures was to be controlled by their immediate practical applicability, that is their usefulness in guiding actions and decisions. This conception tended to strengthen the comparison of statistics with accounting, regarding it as a set of tools rather than a science. This demarcation of the role of statistics clearly comes across in the letter Lenin sent to the TsSU’s director on 16 August 1921: “For practical work, we need to have figures, and the TsSU should have them before anyone else. But we will defer the verification of the accuracy of the figures, the estimated percentage errors, etc., to a later period” (Lenin, quoted in Iastremskii & Khotimskii, 1936, pp. 14–15).

Lenin does not provide a precise distinction between accounting and statistics. Rather, he differentiates between statistics as a “bourgeois science” inherited from the nineteenth century, and a newly emerging field of (socialist) statistical practice, accounting, which he sees as a practice of factual recording and counting, which does not involve any analysis or interpretation of reality, contrary to what, in his view, (bourgeois) statisticians did when producing their data. For Lenin, it was accounting which was needed for the construction of a communist economy and society. In his eyes, this construction project was an urgent undertaking and the “academic” efforts of (bourgeois) statisticians a waste of valuable time, standing in the way of proceeding with it as quickly as possible.

Economic and political urgency justified paying less attention to the use of statistical theory to verify the accuracy of figures, which led statisticians to restrain their scientific ambitions. The scope of statistics had to relate above all to the need to plan economic activity, which demanded a state statistics administration totally dedicated to this task:

The central administration of statistics must not be an “academic” and “independent” body, which it currently is, for 9/10ths following old bourgeois habits, but one for constructing socialism, for verifying and controlling the accounts of what the socialist state needs to know now, today. (Lenin, 1955, vol. 28)

These Lenin quotes are not provided merely for form’s sake. For it was in the name of his own vision of the role of statistical surveys in the management of Soviet economy and society that he regularly intervened in the TsSU’s affairs until his death in January 1924 (see Kotz & Seneta, 1990). Then, at the 13th Congress of the Party in May 1924, Stalin in turn linked statistics back to accounting, while remaining very vague about the nature of the connection between them:

No construction work, no work for the state and no planning is imaginable without correct accounting. But accounting is inconceivable without statistics. Accounting without statistics will not make a single step forward.1

The link between statistics and accounting was clearly reaffirmed in the early 1930s and symbolized by the TsSU being taken over by the State Planning Commission Gosplan in 1930. In the reference manual of statistics published that year under the direction of B. I. Iastremski and V. I. Khotimski (1931),2 the authors reiterated the tasks Lenin had assigned to the TsSU in the introduction: statistics had to be associated with accounting, even in the study of social phenomena. In such a conception, the attention paid to the tools was essential. To better understand the decisions made for their development and use, we need to recall the definition of the scientific nature of statistics given in the 1930s, a key period in the formation of the Soviet system under Stalin.

In fact, just like scientists or professional specialists in other fields, the statisticians had to justify the status of statistics as a science in the early 1920s (on the status of science in the USSR, see Graham, 1993; Krementsov, 1997). They strove to develop an analytical framework for social and economic phenomena adapted to the plan to construct a socialist economy and society, setting themselves apart from the writings of European statisticians, described as “bourgeois” by the Soviet political leaders.

In this task of revision and reinterpretation, the Soviet statisticians were faced with a profound contradiction. Though rooted in an intellectual tradition inherited from the nineteenth century, which regarded statistics as a science on a par with chemistry or mathematics, during various conflicts they were forced to justify the socialist nature of their work and its conformity with the political plan to construct a new state, and its economy and its society. But for these statisticians, who associated figures with objectivity and scientific truth, all science sought to establish was a truth that could not be imposed by political leaders (on the notion of objectivity, see Porter, 1995). The TsSU’s statisticians were subjected to various forms of political pressure that they nonetheless had to come to terms with, at times under duress, and that drove them to work out a theory in conformity with the Bolshevik political message about the construction of a socialist economy and society. How much of their theoretical constructs were based on compromise? In the initial stages, they embarked on a task of theoretical deconstruction.

A Task of Theoretical Deconstruction

In reality, it was rather a hollow definition of Soviet state statistics that was produced, as opposed to the statistics qualified as “bourgeois” to describe the theories developed in the nineteenth century. The manual published by B. S. Iastremskii and V. I. Khotimskii in the early 1930s begins with a denunciation of the latter:

Like any science, statistics is a science of class. The whole system of statistics in capitalist countries is constructed to serve the interests of the ruling classes. Bourgeois statistics does not just serve to significantly reduce income tax of different capitalist groups, it also hides the actual amounts of military spending. Furthermore, bourgeois statistics gives a false picture of the situation of the capitalist economy by embellishing it and attesting to the absence of antagonisms. (Iastremskii & Khotimskii, 1931, p. 6)

The production of statistics in capitalist countries is presented as an undertaking of falsification serving a non-egalitarian state. Accordingly, its fundamentals cannot be used in a state that plans to achieve equality between all human beings. On that basis, the authors of the manual are led to refute certain theoretical contributions of the pre-revolutionary period by associating them with the fundamentals of statistics as used by the bourgeoisie of capitalist countries:

Bourgeois statistics as practised is based on the bourgeois theory of statistics, which is organically connected to the whole system of bourgeois political economy and philosophy. The authors and theorists of the bourgeois science of statistics (Süssmilch, Quetelet, Lexis, Bortkiewicz, Pearson, Mittchell, Bowley, Moore, Chuprov, the fascists Pareto and Gini, and others) give arguments, with the aid of statistical constructs, extolling the “unshakeable” and “eternal” nature of the capitalist system. (Iastremskii & Khotimskii, 1931, p. 6)

The authors discredit the statistical reasoning developed in the nineteenth century (on statistics in the nineteenth century, see Porter, 1986), accusing it of encouraging the status quo and of curbing economic and social progress by basing natural and unchanging laws on the “regular stability of statistical figures”:

In the first half of the 18th century, pastor Süssmilch spoke of “the divine order” that manifested itself in the regular stability of statistical figures. In the 19th century, Quetelet, who admittedly said so without referring to God, spoke of a natural order that found expression through statistical figures, of an “average man” having a specific number of crimes, good deeds, etc. (Iastremskii & Khotimskii, 1931, p. 6)

This accusation (nineteenth-century statistics being a curb on progress) led to the plan to construct a system of “socialist accounting” serving a political programme of figures being used by the people:

Lenin on many occasions said that accounting would become the business of the masses only after the overthrow of capitalism. But such accounting is impossible without statistics, which is also essential for the good of the masses. “In capitalist society, statistics was the exclusive reserve of ‘people of the state’ or narrow-minded specialists; we have to bring it to the masses, popularize it so that workers can gradually learn to understand themselves and see how and how much they have to work, how and how much they can rest…” (Lenin, 1955, vol. 21). The essential nature of disseminating socialist accounting among the people requires maximum simplification of its technique. (Iastremskii & Khotimskii, 1931, p. 13)

The production of figures in a socialist state thus becomes a tool for the people, one that the people must appropriate, which requires the contents of statistics to be reduced primarily to technical mechanisms, assumed to be accessible to the greatest number. However, the link between accounting and statistics is not broken, even though it is not clarified. Such a conception gave rise to numerous discussions between political leaders and statisticians and among the statisticians themselves.

Debates and Tensions Surrounding Statistical Theory

Scientific debates were mingled with internal political rivalries (see Maksimova, 1996). These discussions were all the more heated when statisticians trained in Marxism in the new Soviet higher education institutions, such as the Plekhanov Institute for the National Economy3 or the Institute of Red Professors,4 started being recruited by the TsSU. The rifts between statisticians keen to maintain theoretical mathematical statistics and those who preferred more descriptive statistics based on surveys were reinterpreted during their controversies. The interpretation of the mean and of the law of large numbers, as well as the use of probabilities in statistics, were in particular the subject of heated debate.

Tension Surrounding the Mean and the Law of Large Numbers

As already mentioned earlier, the interpretation of the mean by Quetelet was accused of leading to moral fatalism, which negated the freedom of individualwill, and facilitated a form of social fatalism whereby society depended on superior laws that one could not combat (on Quetelet and the mean, see Desrosières, 1988; Porter, 1986, ch. 2 and 4; Thévenot, 1994). The fact is that these conceptions were in contradiction with the Bolsheviks’ Promethean transformation plan.

The law of large numbers, which was one of the essential factors legitimizing statistics as a science serving the measurement of social facts, was condemned for the same reasons from the early 1920s. The Marxist theorists emphasized the fact that observed trends were not inevitable, and that such a concept led to putting the state’s action into perspective. In their view, this statistical law contradicted the principle of political action, as it portrayed social processes as fatalistic and the effects of political action on society as illusory. Moreover, it helped to justify the idea of stability of capitalism and thus its unchanging dimension.

On the other hand, for the TsSU statisticians, statistics could not exist without the law of large numbers. Consequently, they strove to present a use of the said law that did not challenge its theoretical foundation while at the same time taking into account the attacks on it. In 1936, Vladimir N. Starovskii, head of the Central Statistics Administration,5 proposed a wording full of stylistic acrobatics in a work published that year (Starovskii, 1936). True to Marxist thinking, he did not question the existence of laws that transcend individuals when having to deal with large populations, but he challenged the ineluctable nature of phenomena that appeared to him to be suggested behind the statistical regularities and the lack of references to history. He presented as essential the reference to the historicity of these laws, which he presented as specific to a given social and political system. That being the case, the state could transform the said laws and replace them with others. Human and political action was no longer at risk of being powerless. But a socialist use of the law of large numbers and the mean still had to be defined.

In fact, as they failed to establish the theoretical foundations of a form of statistics that could be described as socialist, the statisticians in charge of this task of reinterpretation endeavoured first of all to clarify what statistics should not be. Their reasoning consisted in calling into question the purportedly erroneous interpretations inherited from the pre-revolutionary past rather than in elaborating a new set of concepts adapted to a socialist economy and society. On the substance, it was therefore more a political debate on the use of the tools than disagreements on their theoretical foundations. Beyond the views and reasoning of convenience, the statisticians were forced to think about the interpretation of the tools they used and the creation or use of other tools better suited to the context of the moment. The tension around the concept of randomness and the use of probabilities is one such example.

Tensions Around the Shift to the Random Model

In 1917, Russia had already acquired extensive experience in the field of sample surveys. The first tests of surveys on a “part of the whole” probably took place in the 1870s. The period from 1885 to 1917 was marked by numerous efforts to perfect sampling methods, and by serious thinking by the Russian statisticians in this respect (Mespoulet, 2002). From 1895, they had the same discussions between themselves as their colleagues in other European countries about the shift from an exhaustive count to a partial survey, then about the shift from a sample constructed in a reasoned manner, based on the use of type, to a random sample based on calculated probabilities (on the history of sampling methods and sampling surveys, see Desrosières, 2002; Gigerenzer et al., 1989; Stigler, 1986). At the European level, the two questions of representativeness and confidence raised by a survey of “a part of the whole” were at the heart of a debate that started with Kiær’s first communication at the Congress of the International Institute of Statistics (IIS) in 1895 and ended in 1925 at the IIS Congress in Rome, where both sampling methods—reasoned selection, also called purposive selection, and random selection—were accepted after heated discussions about their respective merits (on the debates during the period 1895–1925, see Desrosières, 1998 [1993], ch. 7; Kruskal & Mosteller, 1980).

The exuberant production of statistics in the USSR in the early 1920s created particularly favourable conditions for the rapid spread of sample surveys in the country (Mespoulet, 2002). Intensive use of such surveys from 1919 onwards stimulated thinking about sampling methods. After 1925, such thinking indeed developed much more as a logical extension to the questions raised by the already long-standing practice of Russian statisticians in this field than in relation to the questions discussed at the Congress of Rome, even though a summary on this subject was drawn up by the TsSU.

However, the shift to the random model took particular forms from the 1930s, in a country where the political leaders propounded a representation of society in which the collective prevailed over the individual, and in which an average collective behaviour was preferred to individual—and therefore diverse and dispersed—forms of expression. Moreover, the Bolsheviks’ purposeful plan to construct a new state and their Promethean representation of human action on the environment gave a new dimension to the debates about the adoption of random sampling: what place could chance have in a planned world, in which there was precisely no place for uncertainty, and in which the unpredictable nature of individual behaviour and the variability of individual cases could not be factored in?

Planning concerns influenced the thinking about statistical representation and uncertainty, and consequently the forms of recourse to the calculation of probabilities. The treatment of statistical dispersion and of the mean, on the one hand, and the construction of sampling methods for sample surveys, on the other hand, depended on the status of the individual in society. The way of considering the question of confidence in the method for constructing samples influenced the forms of the shift to random sampling in the USSR.

Random selection, the product of the law of randomness, could only be considered within the bounds of a prior breakdown of reality controlled by human reason, in accordance with the Marxist concept of the primacy of human action over the economy and society. Chance could only operate if confined within a framework demarcated by human reasoning and compliant with political choices and directives. From that stemmed the basic Soviet statistical sampling method, developed in the 1930s, subsequently modified in the early 1950s. Whether under the name “random stratified sample” (raionirovannyi sluchainyi otbor) or “typical stratification” (tipicheskoe raionirovanie), the method for creating a sample based on a combination between the demarcation of “typical groups”, firstly, and random selection, secondly, addressed the need to perform a prior breakdown of social reality into classified categories deemed relevant for analysing class structure or different types of agricultural or industrial production structures. This way of combining use of type and a specific interpretation of randomness characterized Soviet statistics from the 1930s to the early 1990s.

Mechanical selection was preferred to probabilistic random selection in the strict sense. The former was considered to provide more accurate results than random sampling, in the strictly random sense of the term, inasmuch as it guaranteed regular distribution of the sample’s units over the entire set. In actual fact what was called mechanical selection was a systematic selection (1/5 or 1/10 for example) without replacement. This preference for mechanical selection was also a constant of Soviet statistics until 1991.

This form of constructing samples combines the two symbolic characteristics of the preference for typical stratified sampling: attachment to territory as the basis for division into typical groups, which dates back to the association between territory and type of the nineteenth century, and the principle of an a priori breakdown of observed reality into classifications. Random selection could only occur subsequently. From the 1930s, the classifications developed served to form “typical groups” based on an economic breakdown of reality into branches of production. In the process, they resulted in excluding a part of the population of the samples used in budget surveys for instance, casting out to an unknown world all those who were not counted in the productive sphere, pensioners among others. They thus departed from the principle of representativeness in relation to the population as a whole. This way of focusing statistics on groups or entities that mattered economically, socially and politically in the eyes of the leaders can be explained by the priority given to targets. This appears no less in contradiction with the desire to know everything, which is often ascribed to the Soviet leaders. In this case, the search for information appears to have been selective, satisfying political choices rather than statistical laws.

At the end of the 1960s, A. Boiarskii clearly addressed the problems raised by this sampling method:

A sample formed on the basis of production is obviously a sample of individuals who work, and in that case the rest of the population is only included in the field of view inasmuch as it has a connection with those who work. However much we then adjust and correct the results, this sample will never replace a sample of another kind of population. With this sampling process, observation is still based on the habitual recourse to administrations and companies. But from data concerning all companies one cannot obtain an appropriate reflection of the life of all the population, all socialist society in its entirety. (Boiarskii, 1968, p. 16)

This tension between a realistic conception of statistics, which produces figures “reflecting” reality and was defended by the statisticians, and a representation of the figures, treated as one of the instruments for constructing a socialist reality and preferred by the political leaders, is regarded as one of the main components of Soviet statistics. In such a situation, the statisticians had to make efforts to adapt to the requirements of the moment; this took various forms according to the period and the circumstances. A. Boiarskii’s remark implies moreover that the adjustments and corrections of results may in certain cases have resulted from this effort aimed at better matching the data with reality, despite the choices of methods or techniques that resulted in ignoring or masking certain aspects or dimensions thereof.

What Form of Statistics for Constructing a New Order?

The Soviet state was founded on a system of representations of management of the country predominated by the image of a stable universe, in which a planned economy was based on an organizational method designed to lead to an increase in general living standards and to an egalitarian society. Figures occupied a central position, being a tool for information and action, a tool for evaluating results and an instrument for proving the rightfulness of the measures taken. Central to such a system of administration of the economy and society, quantification was a basis for legitimizing the state and power. Accounting seemed to be the new language that the Soviet Union needed to construct a system of quantification serving the construction of socialism in the economy and society. But that also needed new categories of classification to be constructed to characterize the composition of society and to redefine the criteria on which confidence in the data produced by state statistics was based.

The Relationship Between Statistics and Accounting

In 1932 Valerian V. Osinskii, then head of the Central Directorate of Accounting for the National Economy (TsUNKhU),6 defined the role of statistics in relation to accounting:

During the transition towards socialism, accounting must encompass all spheres of economic and social life, and must penetrate all its links, even the smallest.

[…] As the remains of capitalism were being swept from the economy and the consciousness of people, statistics was increasingly superseded by national accounting itself. The former TsSU did not become the TsUNKhU by accident. This change in name did not happen by accident, it characterizes the transformed orientation of activity of the system as a whole. (Osinskii, 1932)

V. V. Osinskii regarded statistics as a relic of capitalism. Henceforth serving the plan, statistics had to be transformed into accounting to draw up the national accounts. This unified form of accounting, from initial recognition to centralized processing, had to ensure uniform auditing and processing both of demographic and social phenomena and of economic activity, and guarantee continuity between companies’ accounts and the national accounts. This idea of a unified accounting system was not exclusive to the USSR. It is found with variants, after the Second World War, in European countries that set up their own system of national accounting (see Studenski, 1958; Vanoli, 2002). In the USSR, it was intensified by bureaucratic management at different levels, from the basic department of a company to the central planning bureau of the Gosplan (Kornai, 1996, pp. 141–164).

Was accounting destined to replace statistics? The answer to this question remained ambivalent until the end of the 1930s, reflecting the difficulty in inventing a new quantification language centred primarily on accounting devices without resorting to statistical theory as a resource. So Stalin’s formulation, at the 16th Party Congress in 1930, of the idea that “there can be no accounting without statistics” then gave rise to much prevarication and many changes in attitude in the definition of the connection between statistics and planning through accounting.

One such example is provided by the fate of the theory of the decline of statistics formulated by Osinskii in 1932 and subsequently denounced at his trial in 1937. The idea that statistics had to be transformed into accounting to serve the preparation of the national accounts was put to the test. This prompts one to put into perspective the formulations made during the 1930s about the conception of the scientific role of statistics, as the use of ideas for political ends was standard practice at the time. However, even if the role of statistics was once again officially recognized at the end of the 1930s, it was not at all well-defined. In reality, it was only in 1948 that statistics was fully restored as a field in its own right, when the TsSU regained institutional independence from the Gosplan, the figures of which it was tasked to audit. Its rehabilitation was completed in 1960 when Starovskii, then director of the TsSU, gave himself up to a form of self-criticism on the occasion of the publication of his work on the history of statistics in the USSR:

In economic literature, all sorts of leftist ‘theories’ were formulated, for instance the ‘theory’ of the decline of money under socialism or the ‘theory’ of the decline of statistics under socialism (the transformation of statistics into accounting, which appeared for the first time in the articles of academic V.V. Osinskii) [...]. The author of this article, in his works, in the period 1932-1935, also asserted this ‘theory’. Life has clearly demonstrated the pointless and erroneous nature of these ‘theories’, which distracted the attention of scientific managers from current questions surrounding statistical theory. (Starovskii, 1960, p. 16)

Nevertheless, the asserted integration of statistics into accounting had far-reaching consequences on the forms of the use of statistics and the tools used. Firstly, the period 1930–1955 was marked by the virtual disappearance of social statistics. The scientific and universalist ambition of nineteenth century statisticians, defended by the early heads of the TsSU, was supplanted by an immediate use of figures for accounting purposes to provide the Gosplan with the data needed to work out and evaluate five-year plans. The social figures produced between 1930 and 1955 were mainly demographic in nature. Statisticians became a kind of “right hand” providing the economistic planners with statistics needed for elaborating the plan, for instance for building blocks of flats, schools or leisure facilities.

On the other hand, a whole new area of statistics developed, based on economic forecasting and the construction of indexes. The statisticians elaborated new approaches directed at constructing a national accounting system, the reflections of which are already evident in the manuals published in the 1930s (see in particular Iastremskii & Khotimskii, 1936). The chapters dealing with “relative magnitudes” and indexes occupied a more important place in them, helping steer practices towards a purely accounting use of statistical techniques.

When the Methodological Council of the Central Directorate of Accounting examined the training programme on corporate accounting techniques and statistical theory in 1934, it stressed the need to limit statistical theory to what it described as the purpose of statistics: categorization, the calculations of averages, indexes and the sampling method.7 The construction of indexes, for instance, was based on ratios aimed to compare two quantities, for example industrial production to agricultural production. The indexes were built for planning purposes and for the evaluation of the plan objectives. However, not by any means unbiased, categorization brought into play considerations that were much more than merely technical, namely social and political issues, in particular when it was a matter of constructing other tools, such as classifications of population censuses.

Categorization of the Population and Censuses

The work of constructing occupational classifications used for censuses combined the desire of the statisticians to reflect reality as closely as possible and the classification models officially accepted by the Party. The tension between a “realistic” conception of statistics, which produces figures ‘reflecting’ reality defended by the statisticians, and an understanding that treated statistics as one of the instruments for constructing a socialist reality, which was preferred by the political leaders, can be regarded as a key feature of Soviet statistics.

Each census involved a confrontation between the desired construction of a socialist reality and forms of resistance to transformation of society or the economy, and it was important for the results to provide a snapshot of the volume and structure of the population. The census, which had to confirm the successful construction of a socialist society, was the central link in the statistical tools for constructing reality. From the mid-1920s, its production was subjected to increasingly frequent attempts at intrusion by the Party at various different data collection, processing and publication stages.

While as in any country the elaboration of classification categories for the individuals and phenomena under study was the subject of negotiation between various administrations or institutions, it raised particular issues in the USSR, where the administration of society was based on a classification of the population into categories defined by the Party. The TsSU’s statisticians had to find the adaptations needed to match the bill of materials with, among other things, the structure of the social classes adopted by the Party or the official list of nationalities (on the classification of nationalities in the censuses, see Cadiot, 2007; Hirsch, 1997). This work of adjustment, which varied according to period, the political views and priorities of the moment, was a source of tensions between statisticians and political leaders and calls to order from the latter to the former. The example of the definition of occupational categories in the 1920s and 1930s clarifies certain forms of adjustment. In the 1926 census, the category “occupation” replaced “profession”, which had been used by choice in the 1920 census, but which created difficulties for constituting the main social groups defined in the theoretical Marxist social analysis model(Mespoulet, 2008). Subsequently, behind a simplified classification into major social groups, called “classes” at certain periods, there remained a very detailed classification of “occupations” that reflected social diversity.

The purges of statisticians following the 1937 census halted the TsSU statisticians’ efforts to reflect such diversity through censuses (on the Great Purges in the Soviet state statistical administration, see Blum & Mespoulet, 2003). From the 1939 census onwards, a stable image of social structure dominated through a classification of social groups that remained broadly unchanged until 1989. This virtual stability of the classification, which can be an advantage for comparisons over time, nonetheless resulted in obscuring some of the diversity of the social situations of individuals from the 1960s onwards. Generally speaking, the model of stable social aggregates, based on major groups, overlooked changes affecting small groups of individuals, regarding them as residual or marginal phenomena. In the process, it left blind spots with regard to the actual state of Soviet society and failed to spotlight its underlying transformations that broke out after the perestroika.

What room was left for the social in this model where the collective prevailed over the individual? In reality, social phenomena were studied mainly when they could be reduced to economic management. Furthermore, reflecting on how social matters were taken into consideration raises another question, namely the confidence one has in the data on society collected by the statisticians and in what the respondents say.

Confidence in the Data and the Status of the Statistician

As already mentioned, the question of how much confidence we can have in statistics explains the reservations one can have about using a random selection in the sample surveys method practiced in the USSR. This question is also expressed in a number of ways in connection with the collected data and the processing of individuals’ responses to the surveys. How much confidence can one have in what respondents say, and thus in the data based on their responses? Beyond that, this point also raised the question of confidence in the work carried out by the statistician.

From the end of the nineteenth century, surveys were considered a social situation in their own right, in which the effects of interaction between interviewer and respondent played their part (Mespoulet, 2008). However, whereas this thinking had helped to consolidate the role of the professional statistician in conducting field surveys up to the early 1920, in particular in questioning individuals, it went in quite another direction from the early 1930s. The effects induced by the leeway respondents had in responding made them suspect in the eyes of the authorities, and the Party started intervening in the organization of data collection operations, more particularly by controlling recruitment and the work of local census takers.

In areas other than state statistics, in the early 1960s, the local Soviets and trade unions appointed “civic controllers and engineers” to audit certain stages of statistics work at a local level (see Anisimov, 1968; Sbeglov, 1967). In the late 1970s, an article in Vestnik statistiki, the TsSU’s journal, pointed out: “The state’s statistics bodies, with the active help of the Party, Soviet organizations and trade unions, do their work by recruiting civic activists on a large scale to audit the accounts, administrative registers and the authenticity of the data in the registers” (Vestnik statistiki, 1978, p. 67). This auditing of statistics by the citizens, in reality by social activists, presupposes a careful choice of those hired to do the work, who were also known as “civic inspectors of state statistics”: “They were selected and confirmed nationwide by branch, according to worker qualification, and also according to the possibility of combining an inspector’s obshchestvennaia rabota with his main job” (Vestnik statistiki, 1978, p. 67). Who were these inspectors, also called obshchestvenniki, civic activists?

In principle, the most experienced specialists in the various branches of the national economy were recommended as civic inspectors of state statistics, and the nature of each district is also taken into account. These inspectors are accountants, economists, engineers in factories, administrations, organizations and kolkhozes (collective farms), workers in the plan and financial bodies. Applications are put forward with the agreement of the managers of the companies and organizations where the applicants work. They are examined with the leaders of the Party and the trade unions in the workplace, and only after that are they confirmed by the local executive committee of the district or town, the soviets of the people’s deputies. (Vestnik statistiki, 1978, p. 67)

What precise tasks were expected of these inspectors? First and foremost, their function entitled them to demand that the managers of factories, work sites, government departments, organizations and kolkhozes disclose their accounting records and work registers for inspection purposes. This consisted in auditing accounts and figures in factories and other workplaces, by examining the original documents. On that basis and with the aid of the managers of these establishments, the inspectors could take measures in situ aimed at eliminating the defects brought to light, and also ask for the collectives of workers subject to this inspection process to be convened in order to discuss the results thereof.

In 1978, there were nearly 30,000 inspectors covering most of the districts of the Soviet republics, half of whom were in the Federal Republic of Russia (Vestnik statistiki, 1978, p. 68). Most of them had a secondary school or higher level of educational attainment, and many of them were members of the Party or candidates for Party membership.

Whether it be a control of the data provided by factories or administration registers, by censuses or budget surveys, this inclusion of data collection operations in the sphere of civic activities demonstrates a special way of treating interaction between interviewers and respondents in an authoritarian state such as the USSR. The question of the degree of confidence one can have in the information provided by the respondents, a subject of discussion for Russian statisticians from the 1880s onwards, remained an important factor in Soviet statistics, but the way it was resolved took on a different form. Before 1917, Russian statisticians shared the idea that it was up to the interviewer himself to behave in such a way as to gain the trust of the respondents and thereby elicit more accurate and comprehensive responses from them. After 1917, the Party had a quite different view of the way of handling trust in relations with respondents.

In the censuses of 1919 and 1920, against a backdrop of grain requisition campaigns, the Bolsheviks suspected the peasants of holding back information, especially about their harvests and stocks of grain production. So they were considered guilty in principle. This suspicion of the leaders vis-à-vis the peasants was a constant in the 1920s and 1930s. So, an assessment of the confidence one would have in the information provided by the respondents was not left solely to the statisticians, suspected in principle of possibly colluding with the respondents. The degree of confidence one could have in the collected information therefore had to be handled differently, both by the Party and its social organizations. The control of the statistics would be done by new forms of social and political control. The census preparation and collection operations in the various districts were controlled by Party bodies and social organizations until 1989.

The intrusion of the Party and the social organizations in the conduct of field survey operations, both for censuses and for surveys of family budgets, socialized this moment of statistical work while at the same time divesting part of the statistician’s professional competence.

In that light, Lenin’s statements regarding the need to “popularize” statistics, in the sense of making it accessible to the people, take on a particular relevance. The ambition of the Soviet regime was to create a “new human being”. At the centre of this project, all aspects of social life were deemed susceptible to control by the mass organizations representing the people. Who better than the collectors of the people’s data to control the statements of citizens subject to survey? The verification of the reliability of the gathered information compared with reality took the form of social control. Such inclusion of certain statistical operations in the sphere of civic activity restricted the statistician even more to a primarily technical role, cutting him off from the very source of the information he had to process.

What Statistical Tools for a New Order?

Up to the 1950s, the management tool aspect of statistics was regarded as the main one in the apparatus of state statistics and was closely linked to planning. However, surveys of family budgets, which were precisely at the point where statistics and planning intersected, offer a good example of a technical mechanism that in certain cases could also be used as a tool for acquiring knowledge about society, which Boiarskii, then Director of the TsSU Research Institute, himself acknowledged in 1968 when, following new directives from the 1967 Party Congress aimed at developing the social sciences, the TsSU had to “intensify the statistical study of social phenomena” (Boiarskii, 1968). For Boiarskii, surveys of family budgets were the preferred tool for assessing what he included in the subjects of study of social statistics: problems at work, wages, consumption, services, housing, everyday life. Apart from that, health, the composition and movements of the population and the family structure were in his view in the realm of demography, which also encompassed social statistics. In fact, the demarcation of the latter was dictated by an imperative: all these questions had to be studied as “a reflection of the economy’s effects on all other phenomena of social life” (Boiarskii, 1968, p. 14).

In reality, from the 1950s, and even more so from the 1960s, many of the tools and mechanisms used in Soviet social statistics resembled those used in capitalist countries (Elisseeva, 2003). The difference was not so much in the type of tools used but rather in the way they were used, which Boiarskii himself pointed out, for instance in connection with the methods of sample surveys conducted by Soviet sociologists in the late 1960s:

We cannot compare our surveys with those of bourgeois sociologists without due consideration, on the basis of specific criteria. The main difference lies in the fact that their programme (both of observation and of processing) is constructed in such a way that by expressing itself in a manner full of imagery one can’t see the wood for the trees. Their specific nature does not stem from the fact that one proceeds by conducting sample surveys or by conducting a survey, but rather from the fact that the whole construction of these surveys is not designed to bring to light what matters most: the role of relations of production that, all things considered, in reality determine all these processes. (Boiarskii, 1968, p. 17)

Much more so than the tools as such, and the statistical theory underpinning them, the difference resided in the approach and the model for interpreting the reality in the framework of which they were used:

Our sample surveys are constructed on quite different bases and principles – the principle of the historical materialism of sociology, in our Soviet, Marxist conception of the word. It is precisely in that, and not in the technical question, that resides the dividing line between the surveys of sociologists in socialist countries and those of bourgeois sociologists. The very forms of the work may be similar. (Boiarskii, 1968, p. 17)

National accounts, demographic censuses, budget surveys, all these tools were used for a similar purpose in different European countries after the Second World War, whether to measure production levels or living standards or to analyse trade circuits for the purpose of macroeconomic regulation and increasing material well-being in a country. In the case of the USSR, victory against the United States to claim the title of superpower also depended on success in the economic competition. The Soviet socialist system thus had to demonstrate its superiority in the same economic and social fields as the capitalist countries, which meant that they tended to adopt the same tools, while re-interpreting their uses.

Conclusion

Quantification can be analysed as a way of legitimizing a fragile and young power, if one considers the USSR of the 1920s and 1930s, but it can be said that this was still true in the 1950s and 1960s, in the period of confrontation with the capitalist states. Adam Tooze has shown how statistics have been used to boost the legitimacy of the German state under the Weimar Republic (Tooze, 2001). In his view, statistics offered the Weimar Republic a new, attractive and credible language for government. Theodore Porter for his part has shown how the use of quantitative language also goes hand in hand with a transformation in the bases of authority of the expert, in this instance the statistician (Porter, 1995). To what extent can one speak of socialist statistics in the case of the USSR?

At first sight, what is striking is the difficulty that political leaders and managers of the TsSU had in defining a new conceptual framework specific to state statistics, both in the 1920s and 1930s (the period of formation of the Soviet system) and subsequently. Marking a break with the pre-revolutionary economic management systems, planning had to have a quantitative language and new tools, radically different from those of statistics nourished by 19th century scientific representations and the bourgeoisie.

Accounting seemed to provide this new language that Soviet Russia needed to construct a quantification system serving the construction of communism. In reality, statistics was never abandoned, even if its existence was hotly contested in the period of formation of the Soviet system. The expression “socialist statistics” was used to refer to a different way of producing figures from the way that capitalist countries produced them. However, although we can speak of socialist statistics, it is not so much to refer to a specific theory, concepts and tools, but rather to a set of uses of them, which, depending on the period, were reinterpreted and adjusted to the objectives set by the political plan of the leaders of the Soviet state.

Against this backdrop, the Soviet statistician was caught between two poles: heroes of science, or engineers of figures in the service of the authorities. A way of resolving this dichotomy consisted in the state restricting the role of the statistician to a primarily technical role in a Directorate of Statistics or a computational centre, without any expectation of a scientific role, which was entrusted to various research institutes (on the Soviet organizational system of science after the Second World War, see Graham, 1993; Krementsov, 1997). This transformation of the status of the statistician is inseparable from a conception of quantification adapted to a planning model, but also a form of state where power was exercised in an authoritarian manner and figures were considered a propaganda tool.

The authority of the expert statistician was no longer based on his independence vis-à-vis the political leaders. On the contrary, the components of his professionalism were defined by the latter. The expert was kept at arm’s length from the authorities, but at their disposal. The part of his work that was closest to the population, such as data collection for censuses or household budget surveys, was until the 1970s placed under the political control of the Party, being included in the sphere of civic activities.

Notes

  1. 1.

    Source: XIII s’’ezd VKP (b), Stenograficheskii ochet (XIIIth Congress of the Communist Bolshevik Party of Russia (b), Shorthand report), p. 130.

  2. 2.

    A new copy of this handbook was edited each year from 1930 to 1936. One of its editors, Boris S. Iastremskii (1877–1962) was a mathematician specialized in the field of probability calculation. From 1918 to 1933 he headed the Department of Statistical Methodology of the TsSU. He became a member of the Party in 1931. The other editor, Valentin I. Khotimskii (1892–1939), was a former student of Aleksandr A. Chuprov and was specialized in the field of the probabilities. Although he was close to the Party he never became a member. After teaching mathematics from 1924 to 1927 at the Plekhanov National Institute for the National Economy, he became a researcher in the mathematical section of the Communist Academy. He directed it until 1932 and then was appointed at the head of the controlling sector of the Gosplan for Russia. In 1935 he was appointed at the direction of the Statistics Department for Population and Health of the Directorate Accounting for the National Economy (TsUNKhU), that was in charge of the organization of the 1937 and 1939 demographic censuses. The TsUNKhU had replaced the TsSU in 1930. V. I. Khotimskii was arrested in 1937 during the Great Purges. Most statisticians taking part in the organization of the 1937 census were arrested and put in prison. V. I. Khotimskii died in 1939.

  3. 3.

    The Plekhanov Institute for the National Economy was re-formed in 1924 in Moscow. It had a Department of Statistics.

  4. 4.

    The Institute of Red Professors was created in 1921 in Moscow. It was placed under the authority of the Central Committee of the Party.

  5. 5.

    Vladimir N. Starovskii (1905–1975), was recruited as a statistician to the TsSU in 1925. He became a member of the Party in 1939 and was appointed to the direction of the 1939 demographic census after Stalin had rendered void the results of the 1937 census. After being Deputy Director of the TsUNKhU in 1939 and 1940, he was appointed to the direction of the new TsSU when it was re-formed in October 1940. He remained in this function until he died in 1975. From March 1941 he was also vice president of Gosplan.

  6. 6.

    The TsUNkhU replaced the TsSU in 1930. Valerian V. Osinskii (1887–1938) was a member of the Bolshevik Party since 1907. Immediately after October 1917 he was director of the State Central Bank and president of The Supreme Council for the National Economy. After 1918 he fulfilled various leading functions. In 1926 he was appointed head of the TsSU to give a new political direction to the Central Statistics Administration. He was replaced in this function in 1928. A few years later he was appointed head of the new TsUNkhU. He was arrested in 1937 during the Great Purges and executed in 1938.

  7. 7.

    See RGAE, f. 1562, op. 1, d. 749, ll. 138–154.