Thinking About Think Tanks: Politics by Techno-Scientific Means

  • Stephen P. Turner
Part of the St Antony's Series book series (STANTS)


The creation of think tanks has in part been motivated by the desire for an apolitical politics, a politics of facts and standards rather than a politics of interests and public ignorance, and opposed to political machines. This chapter brings out some of the features of this kind of politics in the USA through the historical example of the Massachusetts Bureau of Labor Statistics, which illustrates the place of the construction of a factual world by think tanks as part of policy processes. The history of this bureau also shows how think tanks become surrogates for party intellectuals—a non-category in the USA. Think tanks then proliferated elsewhere as surrogates for politics by normal means.

Hermínio Martins was a thinker introduced to me by Irving Velody and Peter Lassman, who revered him. His chapter, ‘Technology, Modernity, Politics’ (1998), is an extraordinary and indispensable account of the great theme of the technical mastery of nature and the world, and its relation to ‘history.’ In this chapter he notes that A.A. Cournot, one of the ‘technological Prometheans’ at the core of his account, comments that ‘the post-historical phase presents itself in the typical form of gazettes and statistical bulletins, the end of history being also the “end” of historical narrative and narratability’ (1998: 155). Today, it might be said, paraphrasing Weber, that whereas the Prometheans wanted to live in a world of technical mastery, we are forced to. The thinkers Martins called Faustian, such as Heidegger, who rejected this fate, did so at the cost of being unable to comprehend their own societies. This fate is increasingly evident in the domain formerly known as politics, which is increasingly occluded by ‘policy’ understood in the sense of a form of technical mastery that displaces politics. To be in politics is now to be in the business of policy formation and in the thrall of experts.

One symptom of this new situation is the rise of think tanks as an organizational phenomenon but also as what I have elsewhere called a ‘knowledge formation’ (2017), that is, to say a more or less stable social structure or organization in which knowledge is produced. In this chapter I discuss this general phenomenon in relation to politics, and particularly to party structures, which think tanks partly replace and partly support. My central example will be a historical one, contemporary with Cournot: a think tank-like institution which produced statistical bulletins on labor.

The Policy Process and Knowledge Formations

Arguing about definitions of social phenomena is rarely profitable, but the problem of defining ‘think tanks’ is perhaps an exception. The phenomenon of think tanks is one in which the category itself was retrospectively constructed to take account of a novel kind of organization that had emerged out of the mass of organizations that were part of various other organized movements and specialized government bodies. As an organizational form, however vaguely defined, think tanks were copied and became ubiquitous worldwide phenomenon. In what follows, I will try to make some theoretical sense of this phenomenon by placing think tanks in the two more general categories alluded to above: the policy process and the organizational category of ‘knowledge formations.’ This will allow for some rough comparisons and help to identify more clearly what the definitional issues are. My broader concern, however, will be to see why think tanks proliferated, what they replaced, and why this proved to be such a successful organizational form.

Margaret Thatcher once observed that ‘the facts are Tory.’ There is a small truth in this but only a small one. There is an element of fact in policy recommendations of all sorts, but they are not alone. One can think of policies as the joint product of four elements: the ‘facts,’ Weltanschauungen or values, interests, and administrative policy decision practices, or legal expedience. All policies have these elements, but they enter in different ways. We can think of the process that think tanks are part of a process that starts with the construction of facts and ends in actual policy. Facts are of course never enough to generate policy or determine policy. There are many steps, each of which involves construction, then reconstruction, and more reconstruction, to produce a policy.

The language of ‘construction’ can be a distraction, so let me first defuse some objections. Many years ago Richard Rudner wrote a famous paper on the role of values in science (1953) in which he argued that in the determination of a scientific result there were always elements of judgment and choice. These acts of judgment could not be reduced to matters of ‘facts’ and therefore needed to be understood in terms of non-facts, that is, values. For my purposes, this sense of construction is sufficient. In each step of the process of forming and indeed implementing policy, there are transformations from a prior state: values themselves do not directly determine policy; they must be reconstructed, with an element of judgment, into something closer to policy principles or election platforms, for example. And even this is not enough: crafting an actual implementable policy requires decisions about legal form. There is a long similar process at the early stages of policy development with respect to the factual basis for the policy: the creation of a factual world, in the form of digestible boundary objects, such as reports or statistical analyses, all of which requires a good deal of reconstruction, then the addition of elements of interest, values, and considerations of realizability.

The term ‘fact-finding’ to specify the first stage of the process of constructing facts for the purposes of directing policy or supporting policy thinking is useful because it implies the active role and selective work of the fact-finder. What is found may be fact, but it is not fact as such, but the facts that someone finds and presents to others because they are taken to be relevant to the specific purposes for which the project of fact-finding was created and negotiated. Moreover, fact-finding has an organizational aspect. The production of the project is done by some specific organization, which has a specific character as a knowledge formation, with a particular structure, membership, practices, and pre-existing knowledge base, as well as biases, known and unknown.

What I will be calling a knowledge formation is a form of organization that persists over time and must thus sustain itself. To sustain itself it must meet certain requirements, which must be reconciled with one another and are typically in tension in some respect. Indeed, most historical knowledge formations have involved a central tension, which has defined it. The list includes the following:

Knowledge sources: Knowledge has a history and source, and the sources constrain the way a knowledge formation is configured. There are multiple sources of ‘knowledge’ but the type of knowledge involved is important as a determinant of the way in which it is produced. Normally the means of knowledge production is shared by others, such as academic non-think tank producers, or journalists, and are open to criticism and appraisal by others.

Resources: Producing and reproducing knowledge requires people whose lives are to a significant extent dedicated to these tasks, and this means they must have sources of income that support the intellectual work that they do.

Means of communication: To the extent that new knowledge is generated or new interpretations are proposed, ‘publication’—in the literal sense of making known to some relevant persons—is essential.

Norms of conduct and conventions of discourse and exchange: Norms are part of the conditions for knowledge production and also of the reproduction of knowledge. These may vary significantly by field, and across time, but without them it is difficult for communication and exchange to result in something commonly recognized to be ‘knowledge.’ These norms, however, limit as well as facilitate communication, and because they vary from group to group and discipline to discipline, they are also the source of mutual incomprehension and disagreement.

Exclusion/inclusion and marks of recognition: A pervasive feature of intellectual communities is the existence of marks of membership, explicit or implicit. Certification in the form of degrees, membership in societies or academies, peer-review in a variety of contexts, implying a definite notion of ‘peer’ and the like are examples.

External legitimacy: Normally the community or group communicating knowledge has some sort of respect and recognition by non-members. This may be highly formal and come with a developed theory of the status of the particular kind of knowledge. The theory may be accepted by those who do not share the knowledge, or be part of the rationale for a particular institutional structure, such as an education system, bureaucratic order, or religious system.

Think tanks are knowledge formations, and we can regard think tanks as a particular solution to the general problems that all knowledge formations face.

The list allows us to deal with some of the definitional issues, or at least to identify them. The relevant comparisons, at a first approximation, are two: the individual expert, that is to say the expert not attached to a knowledge organization, and expertized social reform movements, that is to say organized movements promoting some reform cause which make or publicize expert claims but do not claim to originate them. Commissions depend for their external legitimacy to a great extent on the personal reputations of its members (Turner 2014a: 71–92, especially 74–75). In contrast, think tanks tend to efface the personal authority of the members in favor of the authority of the organization itself, not surprisingly, as the aim of think tanks is to last longer than the immediate knowledge task and policy problem it is addressing.

One can regard think tanks as beginning in the USA with such organizations as the Carnegie Peace Foundation and such Progressive Era institutions as the Russell Sage Foundation and the Municipal Research Bureaus. Projects of these organizations, such as the Pittsburgh Survey supported by the Russell Sage Foundation, have characteristics of these ongoing bodies. What is striking about them is that to some extent, and almost entirely in the case of the Pittsburgh Survey, the aim was to influence policy through educating the public at large. In the case of this survey and other similar surveys, this was done through a complex educational public exhibit attended by many thousands of citizens, with the aim of creating the conditions for the election of a reformist municipal government (Turner 2014b: 142–144).

At the opposite pole from this, one finds the Congressional Research Office of the US Library of Congress. This is often treated as a pioneering think tank, but its research is not public—it is done for congressmen, and the results are only disclosed to them, though they may use it and disclose it. It is a government agency, though one that is governed by congress rather than the executive branch. But the research is precisely the kind that is carried out by more traditional think tanks. There are European analogues to this in the present, in Germany, for example, but one might also look back to earlier formations which look very similar. The US National Academy of Sciences was established with the explicit aim of providing scientific advice, as an officially recognized body, and in this it followed various European academies. Ernst Engel’s Prussian statistics group was part of a bureaucracy but functioned as a kind of policy-relevant research body as well as a university class training institution, which is the case also for many present organizations classified as ‘think tanks,’ such as CIDE (The Center for Research and Teaching of Economics) in Mexico.

Both kinds of think tanks are concerned with policy: they simply intervene in different points of the policy process, and this difference accounts for many of the apparent differences between think tanks. We can define think tanks as non-temporary knowledge formations whose purpose is to intervene at some stage in the policy formation process through the production of what in science studies are called boundary objects (Star and Griesemer 1989). These include things like exhibits, but also such things as reports, which contain results that are usable and understandable to public audiences but obscure the messy, uncertainty filled process of their production. This definition allows us to concentrate on a more meaningful problem: to explain how, out of all the rival forms of expertization and policy intervention present during and before the rise of think tanks, these specific kinds of organizations were able to find a niche, flourish, and expand to other niches and to the different niches provided by other countries and other contexts.

The Case of Labor Statistics: Think Tank or Bureaucracy?

In what follows I wish to explore the phenomenon of think tanks in terms of an early historical example: the Massachusetts Bureau of Labor Statistics, a pioneering institution of the late nineteenth century. What the example will reveal is how tenuous most of these definitional distinctions are. But it will also reveal something about the conditions for a successful policy-oriented knowledge formation and the means available to it for intervening in the policy process. The founding of the bureau is revealing about the relation between this kind of distinctive think tank-like organization and commissions in the narrow sense. The motivation for the creation of the bureau was, as with many think tanks, a perceived crisis. As a consequence of the economic dislocations during the decade after the Civil War, the problems of ‘labor’ became an important part of popular consciousness and, in the eastern states especially, an element in state politics. Massachusetts was the home of shoe and textile manufactures, which had benefited from the demand created by the war, and also provided soldiers from the working class who had worked in the mills.

Why was this bureau a solution, and what problem did it solve? The creation of the bureau was preceded by a series of acts of labor legislation over the previous four decades, much of which involved working hours. This was an international phenomenon. The short-hours movement was the central issue for labor during the nineteenth century and well into the twentieth. Socialist parties themselves were primarily supported for their commitment to this goal, and when it was achieved, the Left typically lost the unity it had. There were parties of this kind in Massachusetts, but they were relatively small, and that is an important part of the background to the founding of the Massachusetts Bureau: the elected officials had no desire to provide these parties with short hours as an organizing issue, and sought to at least appear to support the working classes.

The bureau itself was preceded by two commissions, one in 1866 and a second in 1867. The first commission was empowered to collect statistical information, and sent out a thousand circulars to towns and cities, asking for reports on the condition of the working classes and particularly about child labor. Its reports presented pro and con arguments for a short-hours law, and recommended against it, concluding that ‘the change desired can be better brought about by workingmen outside the state house, than by legislators inside’ (Pidgin 1876: 19). But the commission did make recommendations for legislation and enforcement in the case of child workers, and also asked that ‘provision be made for the annual collection of reliable statistics, in regard to the condition, prospects, and wants of the industrial class’ (Pidgin 1876: 19). The second commission went farther, and recommended a ten-hour day law (Pidgin 1876: 21).

The commission form solved some of the issues discussed above: its legitimacy was a product of the status of the members of the commission, respected worthies, including an academic, Francis Amasa Walker, whom we will shortly hear more of. Politically, however, it was inadequate for the purpose of providing a political or policy resolution of ‘the labor question,’ and the commissions themselves pointed to means of doing so, by suggesting that statistics be collected on an annual basis.

The specific event that precipitated the founding was the application for a state charter by a labor organization (a shoemakers union). The grant of a charter was at first refused, but the legislature did not wish the refusal to appear antagonistic to the working man, and so, as a gesture, a state Bureau of labor statistics was created to study the issues around the ‘labor question.’ The initial concern of the bureau in 1869 involved an effort to sort out the claims about the number of unemployed and to determine if returning Civil War veterans were taking the jobs of the already employed or were themselves unemployed. The idea behind this was a product of the ‘lump of labor’ fallacy: the idea that there was a finite demand for labor. But the remit of the bureau was much larger, and also very vague. Indeed, as became apparent, there was little known about how to ‘fact-find’ about labor: the methods needed to be devised by the bureau itself. Nor was it clear what the limits of factual construction were: simply tables of statistics, or something more? Who was the audience, and what was the message for this audience? These were questions that were answered through practice and trial and error.

The first director of this bureau, Henry Oliver, was faced with the vague statutorily defined task of ‘the gathering of statistics of labor […] together with what might be ascertained of their influence upon the health, education, manners of life, and industrial habits of the workings class’ (Bureau of Statistics of Labor [Mass.] 1870: 7). The interpretation placed on this wording of the law by the officers of the bureau involved creating a broad theme within which the bureau was to define specific research topics. The purpose was not simply, or even primarily, to aid legislators in making policy or executives in administering the laws effectively but to ‘modify, improve, and reform public sentiment’ (Bureau of Statistics of Labor [Mass.] 1872: 11). This is a succinct formulation of at least one aim of some think tanks: to provide the conditions for public acquiescence for policies that they might not otherwise have accepted. The politics behind this particular case are not clear, but the bureau did survive attempts in its early years to shut it down. One can infer some motives for support: legislators did not want to legislate without public backing. Having the bureau transform public opinion was a convenient and risk-free alternative to doing so themselves: if the public was not persuaded, the recommendations could be ignored.

The means by which the statistics were to be gathered were not specified by the legislature when it created the bureau, and Oliver soon discovered that he did not possess any particular legal authority to compel testimony, as would a census or legislative commission. He at first attempted to get the authority to call ‘witnesses’ in the fashion of a court or a legislative inquiry, but for legal reasons this proved difficult. He was then made a justice of the peace, which enabled him to legally compel testimony, but the bureau had no funds to pay the required witness fees so he was forced to resort to the collection of voluntary responses. This was a pivotal moment in the development of the social survey, for it created the modern relation between the surveyor and the surveyed.

Oliver had what he called ‘schedules’ or ‘blanks’ printed containing questions for employers and employees. He sent these by mail to large numbers of subjects and asked them to respond by mail. Assessors provided lists of employers, but in the case of employees, many devices were used, including newspaper advertising and rolls of unions. The inadequacy of these samples was understood in an informal sense, and the ideal of a complete enumeration was always used as a standard of comparison in assessing the adequacy of the results with respect to representativeness. But there were no good means of overcoming the difficulties within the legal limitations imposed on the bureau. The questions on the ‘blanks’ were, by modern standards, badly worded, and the lists of questions were very long. Many of the questions called for quite simple numerical answers about the number of persons employed and the like, but many were requests for opinions—and not simple questions of approval or disapproval of some statement or position, but complex questions about the respondent’s beliefs about the causes of various undesirable features of the contemporary situation, such as the high price of provisions.

Few workers or employers proved to be willing to answer the complete list of questions. Most simply failed to respond. The answers of those who did were generally ‘curt and unsatisfactory’ (Bureau of Statistics of Labor [Mass.] 1870: 23). The bureau did follow-up, with correspondence, on responses that it found ‘strange or unmeaning’ (Bureau of Statistics of Labor [Mass.] 1870: 12), but apparently not to improve the response rate. In one of these early surveys, the ‘circular’ or questionnaire addressed to workers contained 137 questions, and was sent to 268 workers, of whom 114 replied. When possible, unskilled laborers who were not literate enough to complete a blank were interviewed orally. The equally demanding circular for employers, also sent with a prepaid envelope, received only 217 replies out of 1248.

The bureau was well aware of the inadequacy of the formulations of questions they were asking: in the first report they quote Rousseau to the effect that ‘the art of asking questions […] is the art of masters rather than of scholars, and one must have learnt many things to know how to ask about a thing one does not understand’ (Bureau of Statistics of Labor [Mass.] 1870: 16). This was used to justify the effort the bureau spent on gaining background information, especially historical information, and producing reports on existing laws in other countries relative to the topics listed above and generating statistics on them.

The ‘assumptions’ behind the construction of the questions and the analysis of results differ from what was to become standard survey practice in America; and the differences need to be made clear, for they involve some basic methodological premises. The workers’ and employers’ responses are treated in a manner not unlike the model of legislative testimony, in that people are assumed to have coherent reasons or a particular experiential base for their ‘opinions,’ and the task of the interpreter was, in large part, to exhibit and explicate these reasons. The workers and employers were treated as responsible adults who were capable of making their own judgments. Questions often ask for judgments—for example, of the adequacy of ventilation in a factory—and their responses were taken at more or less face value. This was typical of ‘commissions’ in the narrow sense: because of their short life spans they were likely to take testimony or evidence in the form of reports which others had generated for them to consider.

So what did the reports do? Where there were systematic differences of opinion between workers and employers, the character of the differences was made apparent, but the descriptive categories and ‘values’ were those of the subjects. The role of the analyst was to classify and count the different kinds of reasons and opinions, not to supplant the subjects’ terms of reference. Not surprisingly, the reports were highly readable human documents in which the voices of the subjects come through clearly. The ‘testimonies’ in which workers’ lives are recorded were exemplary in this respect, since the educative aim of the surveys was to make the problems of real people evident. There were attempts to balance and present systematically the ‘facts’ about such things as child labor or divorce that might otherwise be known only through unrepresentative and misrepresented reformist horror stories. Thus, while the initial aim of the surveys was to get the facts about unemployment, these procedures were generalized to other ‘problems,’ especially those in which there was potential legislative action.

The reports thus went beyond ‘tabular statistics’ to become a kind of simulacrum of democratic deliberation, in which different voices were heard. In retrospect, of course, the bureau’s construction of this discourse, and more generally the attitudes of its authors, expressed in terms of filth and degradation, reflected the attitudes of literate middle-class Boston, which was indeed the public whose opinions the surveys were designed to improve (Watson 2002: 15–20). The credibility of the judgments was enhanced by the use of physicians as authorities in the reports: ‘12 In one section, no less than eighteen physicians are cited about the destructiveness of long hours in the factory’ (MBSL I 1870: 127–128). Similarly, the second MBSL Report sent surveys to ‘medical men in four principal factory towns’ (MBSL II 1871: 504) asking them to describe the physical effects of factory life under no fewer than 17 headings (504). This represented a tension: between sheer reporting of attitudes, facilitating democratic deliberation, reliance on expert opinion, and promoting a cause to an audience.

The topics that were examined included many of the themes that were basic to early academic research of the reformist variety, such as the value to the worker of cooperative savings banks. These were sometimes politically treacherous. Oliver’s report on this topic suggested that the workers did not use the banks as heavily as believed and that the major depositors were employers seeking to avoid taxation. The state senate passed a ‘resolve’ rejecting the report as based on insufficient returns and complained that the reports made Massachusetts look bad. As Oliver said in his next report, the bureau was placed on trial.

The Later History

The primary concern of the Massachusetts survey, and other state surveys as well, was the economic condition of the laboring classes: working conditions, child labor, the incidence of effects of employment on women, domestic labor, the cost of living, the numbers of strikes, the sanitary conditions of workers in the workplace and home, and so on. Given the criticisms of its methods and results, the Bureau had no alternative other than to refine their methods within its legal limitations. Oliver’s successor, Carroll Wright, made it one of his first actions in office to test the method of mailing questionnaires by sending it to clergymen, who, he supposed, should be responsible and literate enough to reply: of 1530 mailed, there were only 544 replies, several of which were blank, others ‘sneeringly expressed the intimation that what we were asking was none of our business’ (Bureau of Statistics Labor [Mass.] 1874: 24).

He learned from this. Wright abandoned the mail method, except for certain infrequent uses, and relied on personal interviews, especially visits to employers to examine their records, which produced much better cooperation. Much of the work of the bureau was secondary analysis, especially of data on wages. But a great deal of data was collected by ‘special agents’ of the bureau conducting interviews. These interviews were governed by the interviewer’s understanding of the objects of the inquiry rather than by adherence to any elaborate interviewing procedure, and this was understood to be one of the reasons for the superiority of the personal interview: Wright later explained that the questions on mailed blanks, no matter how precisely formulated, tended to be interpreted differently by different respondents, and were therefore inferior to the results obtained by personal interviews.

Wright was highly concerned with what he called the ‘reliability’ of results, which he viewed as a question of ‘representativeness’ and conceived as a problem of random selection (Bureau of Statistics Labor [Mass.] 1874: 251). Yet these concerns did not lead him to random sampling procedures but to finding a ‘fair ratio’ of types in the population. For he also often used the concept of representativeness in an inference-relative rather than a population-relative sense; his aim was often to find a community or case in which some problem could be clearly observed.

The connection between the activities of the bureaus and legislation was obvious: states gradually produced legislation governing child labor and uniform hours, as well as regulations on many other topics. The bureaus did not long outlive this wave of legislation, and at the end, the Massachusetts bureau was reduced to the performance of such tasks as counting summer residents in hotels for the purpose of justifying the issuance of additional liquor licenses in resort towns. But in the 50 years in which the movement flourished, from 1869 to the 1920s, when child labor remained an issue, research began virtually on all of the problems that were later taken over by academic sociology under the heading of ‘social pathology.’

Why treat this as a think tank? The bureau published reports on the consequences of divorce, the effects of prohibition, the rise of pauperism, the differential effects of certain forms of employment on women’s health, the labor conditions of teachers and clergymen, the situation in tenement houses, the cost of living and workingmen’s earnings, and the Canadian French in New England. They even published what would now be called an ethnographic study of profit sharing under Wright’s name conducted by the future sociologist Franklin Giddings, who was then a Springfield newspaperman. In later years, there were studies on such topics as the social and industrial condition of the Negro in Massachusetts; the level of compensation for female college graduates; the relation of the liquor traffic to pauperism, crime, and insanity; the state of home ownership; the accumulation of wealth through insurance; and the state of workingmen’s savings. The bureau also issued a large number of reports (also called surveys) on legislation and conditions in other countries with respect to such diverse topics as the existence of municipal pawnshops and the licensing of barbers. These are activities typical of think tanks.

The Massachusetts Bureau of Labor Statistics was, however, anomalous among the state labor statistics bureaus. Wright, a highly competent and politically adept leader, had used his discretion to expand the budget and the domain of the activity beyond its original construction. The imitators of the MBLS were not so successful, and many of them restricted their activities to purely industrial topics for the most part, as in fact Wright himself was compelled to do in the position he assumed in 1887 as director of the newly established Federal Bureau of Labor Statistics (which was the direct administrative ancestor of the present cabinet level Department of Labor). Wright did perform some research beyond the narrower boundaries of labor statistics in his time in Washington—the report on the Pullman Strike and the study of slums, including the support of the Hull House survey, and the study by W. E. B. DuBois of the Philadelphia Negro being the most prominent examples. These studies used the same kinds of ‘methods’ as previous ‘labor statistics,’ and were closely monitored by Wright. But these were not part of the administrative routine of the Bureau and were usually responses to special Congressional demands in which the implicit political task of defusing a controversy was paramount.

In the 50 years of the bureau’s activity, the means of asking questions improved, though the same kinds of ‘opinion’ questions were asked into the twentieth century, requiring a great deal of inference on the part of the respondents and treating them as, in effect, local experts with special insight into such topics as the causes of high prices of provisions. This succession raises its own questions. The academicization of reform topics, primarily in the new field of sociology, moved the issues from the symbiotic relation between reform organizations and statistics bureaus to foundations and universities. The problems also changed: the problems that could be solved by labor legislation were replaced by the more intractable problems of race, crime, immigration, and urbanization. These required a different and more comprehensive kind of knowledge that was ultimately at odds with reformism itself.

The Bureau as a Knowledge Formation

If we consider the basic elements of knowledge formations, we can see how the bureau constituted a solution to the problem of sustaining itself. The first issue, which proved to be a difficult one, was to identify a knowledge source. The conflict was this. The politicians said that the task of the bureau was ‘to collect the facts and leave them speak for themselves,’ as Governor Washburn said in 1874. But ‘the facts’ were not a well-bounded domain, and the explicit aim was to affect public opinion. Oliver insisted from the first that ‘Tabular array of figures alone,’ as the preamble to the 1869–70 Report says, ‘are inadequate to help the cause of the laborer […] be put into such light and view that its real status can be comprehended, its disease be determined, the right medicine be administered, its health be secured, and its true interests promoted’ (MBSL I 1870: 15). This reflected both values and expedience: the presentation was motivated by a desire to help, but to help in a particular way, by removing pathologies. In the terms given earlier, these are reconstructions of the facts with values and interests added in, in order to bring the reconstructed facts closer to and more relevant to policy.

This role was the accepted understanding of the bureau, but the problem of partiality was never far removed from discussions of the bureau’s work. As one of the early commentaries on the bureau’s reports put it,

The report itself is a model for clearness, system and practical arrangement, and especially for a thorough recognition of the nature of the facts required by a body of legislators, though it is much to be feared that they will be the last to be benefitted thereby. It does not need a very close scrutiny to recognize that General Oliver, the chief of the bureau, is in warm sympathy with the workmen, and, from conviction, inclined to their views. But, after a careful study of the volume, we acquit the authors of every suspicion of partiality in what was, after all, the main work, the collection of the facts themselves; and these are of the utmost importance.

That the reading of such reports as General Oliver’s will convince many persons of the gravity of the evil, and reconcile them to the freest and most exhaustive discussion of the remedies, is the chief, as it will be the most immediate, result of the establishment of the bureau itself. (Pidgin 1876: 66–67)

This lays out quite clearly the situation of the bureau. It was intervening in a contentious topic. It was certain to face suspicions of partiality. It nevertheless was expected to reveal ‘the evil’ to responsible persons, and ‘reconcile them to the freest and most exhaustive discussion of the remedies.’

The use of statistics, as the prominent Professor Francis Amasa Walker, a member of one of the commissions, put it, was a powerful means of doing this: the country is hungry for information; everything of a statistical character, or even of a statistical appearance, is taken up with an eagerness that is almost pathetic: the community have not yet learned to be half skeptical and critical enough in respect to such statements (quoted in Pidgin 1876: 59).

Statistics were accepted uncritically, and in a sense, this was the goal of expertization itself: uncritical acceptance. But this, as it turned out, was not enough, as Wright came to understand.

The initial resource base of the ‘labor statistics’ movement was explicitly political: it was a concession to the interests of the working classes, although a concession which, at the time, was perceived to be insufficient. The ‘objectivity’ of the results was not especially problematic since, like testimony, its credibility lay in its sources and in the recognizability of the facts in the testimony. The political problems of satisfying diverse supporters in the fractious labor movement and its reformist well-wishers of the time, however, made this resource base unstable. Where it went astray was when it came up with results, as in the savings bank survey, that were unflattering and served few interests.

Wright transformed the bureau into something different and politically more stable—a kind of research organization that would collect facts on public issues and produce ‘balanced’ reports on them which nevertheless shaped public opinion by agenda setting and defining problems in a vivid way that the intended audience would accept. Why did it work? In large part, it was because of the contrast with the theatrical character of the reform movements themselves, which claimed expert knowledge, sought to influence the public, and in some cases, the major one being prohibition, did so, but used methods of publicity that employed emotional manipulation and personal appeals, an inheritance of abolitionism, the source of reformism. The MBLS was a model of sober analysis compared to these performances. And this did a great deal to serve the purpose of external legitimacy.

Politics by Other Means

In a strong parliamentary system with strong parties, these educational tasks, as well as the agenda setting done by the commissions and the bureau, would be done by parties, or party intellectuals, or intellectuals speaking out in reform or religious contexts. In Britain, for example, in the twentieth century, the thinking was done by people like the Fabian Society, and later Richard Tawney, who addressed Christian socialist conferences and theorized the British welfare state, or John Maynard Keynes, who personally addressed the cabinet as an advisor. Yet here there was also a role for think tanks. The Tories had Chatham House, which served as a venue for their party intellectuals. The term has little application in the American context, simply because the parties were heterogeneous coalitions of voting groups which had to be bound together by platforms that reflected interests and to a lesser extent, a vague sense of shared values.

The ‘American’ version of the relationship between think tanks and political parties is represented by the MBLS. The think tanks are non-partisan but serve to set agendas. In the important case of Worker’s Compensation legislation, the one relatively unambiguous success of Progressivism, the knowledge base, had been prepared by decades of research by bureaus of labor statistics and a commission in New York using these statistics. The actual passage of the landmark legislation was the result of a New York State commission charged with writing a law. The person appointed for the task, however, Crystal Eastman, was a Progressive activist who had done the volume on workplace accidents for the Pittsburgh Survey and was a trained lawyer. This was a top-down reform based on a general consensus of workers and employers that had been built in part by the Labor statisticians. Party politics was irrelevant. A second pattern was observed in Chile during the Pinochet repression of democracy. There, think tanks flourished because the ‘technocratic option’ was allowed while the possibility of political party challenges was not.1

A third pattern involves internal party politics: the creation of highly partisan ‘think tanks’ which aim at influencing parties themselves. The present Fabian Society, which has explicitly evolved into a party think tank, and the Center for American Progress are examples. These play an important role because they express as expert certainties for a limited partisan audience claims that think tanks which claim to be non-ideological, such as the Brooking Institute, could not, without jeopardizing their reputations. And there is at least one more pattern: counter-institutions, which contest the claims of think tanks like Brookings to be non-ideological, such as the Heritage Foundation, and supply alternative ‘expert’ claims.

The sheer variety of think tanks and the porousness of the boundaries between them and other kinds of expertized organizations are telling. Expertization is ubiquitous. The balance between traditional political parties and expertized organizations, whether official or private, is shifting. Parties no longer set agendas: they react to the agenda setting of others. This is an epochal change.



  1. Martins, H.G. 1998. Technology, Modernity, Politics. In The Politics of Postmodernity, ed. J. Good and I. Velody, 150–181. Cambridge: Cambridge University Press.CrossRefGoogle Scholar
  2. Massachusetts Bureau of Labor Statistics. 1870. Report of the Bureau of Statistics of Labor I. Boston: Wright & Potter, State Printers. Scholar
  3. ———. 1871. Report of the Bureau of Statistics of Labor II. Boston: Wright & Potter, State Printers. Scholar
  4. ———. 1872. Report of the Bureau of Statistics of Labor. Boston: Wright & Potter, State Printers.Google Scholar
  5. ———. 1874. Report of the Bureau of Statistics of Labor. Boston: Wright & Potter, State Printers. Scholar
  6. Pidgin, C.F. 1876. History of the Bureau of Statistics of Labor of Massachusetts and of Labor Legislation in that State from 1833 to 1876. Boston: Wright and Potter, State Printers.Google Scholar
  7. Rudner, R. 1953. The Scientist Qua Scientist Makes Value Judgments. Philosophy of Science 20 (1): 1–6.CrossRefGoogle Scholar
  8. Star, S.L., and J.R. Griesemer. 1989. Institutional Ecology, ‘Translations’ and Boundary Objects: Amateurs and Professionals in Berkeley’s Museum of Vertebrate Zoology, 1907–39. Social Studies of Science 19 (3): 387–420.CrossRefGoogle Scholar
  9. Turner, S. 2014a. Expertise and Political Responsibility: The Columbia Shuttle Catastrophe. In The Politics of Expertise, 71–92. New York: Routledge.Google Scholar
  10. ———. 2014b. The Pittsburgh Survey and the Survey Movement: An Episode in the History of Expertise. In The Politics of Expertise, 138–151. New York: Routledge.Google Scholar
  11. ———. 2017. Knowledge Formations: An Analytic Framework. In The Oxford Handbook of Interdisciplinarity, ed. R. Frodeman, J. Thompson Klein, and R.C.S. Pacheco, 2nd ed., 9–20. Oxford: Oxford University Press.Google Scholar
  12. Watson, W. 2002. The Facts Which Go to Form This Fiction: Elizabeth Stuart Phelps’s ‘The Silent Partner’ and ‘The Massachusetts Bureau of Labor Statistics Reports. College Literature 29 (4): 6–25.Google Scholar

Copyright information

© The Author(s) 2018

Authors and Affiliations

  • Stephen P. Turner
    • 1
  1. 1.University of South FloridaTampaUSA

Personalised recommendations