Wordtrade LogoWordtrade.com
Social Science

 

Review Essays of Academic, Professional & Technical Books in the Humanities & Sciences

 

Statistical Methods in the Study of Society 

Elementary Statistics in Social Research (11th Edition) by Jack A. Levin (Author), James Alan Fox (Author), David R. Forde (MySocKit Series: Allyn and Bacon) The Eleventh Edition of Elementary Statistics in Social Research provides an introduction to statistics for students in sociology and related fields, including political science, criminal justice, and social work. This book is not intended to be a comprehensive reference for statistical methods. On the contrary, our first and foremost objective has always been to provide an accessible introduction for a broad range of students, particularly those who may not have a strong background in mathematics.

Like its predecessors, the Eleventh Edition contains a number of pedagogical features. Most notably, step-by-step illustrations of statistical procedures continue to be located at important points throughout the text. We have again attempted to provide clear and logical explanations for the rationale and use of statistical methods in social research, and we have again included a large number of end-of-chapter questions and problems. Finally, we have again ended each part of the text with a section entitled "Looking at the Larger Picture," which carries the student through the entire research process.

For more than three decades, Elementary Statistics in Social Research has undergone extensive refinements and improvements in responding to instructor and student feedback. This edition also departs, in important ways, from earlier versions of the text. We have expanded our introduction to multiple regression, including a discussion of dummy variables, and to logistic regression. We have also added the inter-quartile range in Chapter 4, t tests with unequal variances in Chapter 7, and two-way analysis of variance in Chapter 8. Moreover, we have updated data and tightened or eliminated material, where needed. Finally, thanks to the addition of David R. Forde as a co-author, this edition is much more SPSS friendly. It contains a number of end-of-chapter problems to be worked out in SPSS using data sets available on-line and an appendix in which procedures for applying SPSS software are carefully explained. For those instructors who do not teach SPSS, however, these new aspects of the book can be easily excluded.

We have checked and verified all numerical exercises and illustrations to minimize the frustrations that students feel when they encounter computational errors. In addition, a solutions manual is available for instructors in which all problems are carried out in detail.

The organization of the book remains unchanged. Following a detailed overview in Chapter 1, the text is divided into five parts: Part I (Chapters 2 through 4) introduces the student to the most common methods for describing and comparing data. Part II (Chapters 5 and 6) serves a transitional purpose. Beginning with a discussion of the basic concepts of probability, it leads the student from the topic of the normal curve as an important descriptive device to the use of the normal curve as a basis for generalizing from samples to populations.

Continuing with this decision-making focus, Part III (Chapters 7 through 9) contains several well-known tests of significance for differences between groups. Part IV (Chapters 10 through 12) includes procedures for obtaining correlation coefficients, and an introduction to regression analysis. Finally, Part V (Chapter 13) consists of an important chapter in which students learn, through example, the conditions for applying statistical procedures to research problems.

The text provides students with background material for the study of statistics. A review of basic mathematics, statistical tables, a list of formulas, and a glossary of terms are located in appendixes at the end of the book. Finally, the companion website provides data sets for completing the SPSS exercises as well as ABCalc, an easy-to-use statistics calculator developed for this textbook.

The ancillary package for this edition includes:

For Instructors:

  • Instructors Manual and Test Bank.
  • Pearson MyTest Computerized Test Bank.
  • Solutions Manual. This supplement includes worked-out solutions to all end-of-chapter exercises.

For Students:

  • NEW! MySocKit (www.mysockit.com). MySocKit is an online supplement for Elementary Statistics in Social Research, Eleventh Edition, that includes material previously published in the Student Workbook and Companion Website, plus a number of new features. It contains a series of exercises and computer applications; SPSS data sets used in those exercises; a tutorial on basic SPSS commands; an Excel statistics calculator; flashcards for learning key terms and formulas; and worked-out solutions to the odd-numbered end-of-chapter exercises. MySocKit is available with this text at no charge when a MySocKit Student Access Code Card is packaged on request.
  • Student SPSS Software. The most current release of SPSS for Windows (on CD-ROM) can be packaged with the text at a significant discount from the retail version of Student SPSS. Ask your Pearson Arts and Sciences representative for details.

 

The Statistical Mind in Modern Society: The Netherlands 1850-1940, 2 volumes by Jacques van Maarseveen (Editor), Paul Klep (Editor), Ida Stamhuis (Aksant Academic Publishers) In the period 1850-1940 statistics developed as a new combination of theory and practice. A wide range of phenomena were looked at in a novel way and this statistical mindset had a pervasive influence in contemporary society. This development of statistics is closely interlinked with the process of modernisation of society at the time, and with the rapid progress in the sciences. The increasing influence of the statistical approach sometimes evoked strong opposition on the part of government, business and science, and gave rise to lively debates. This two-volume publication is a follow-up to The Statistical Mind in a Pre-statistical Era: The Netherlands 1750-1850 (Paul M.M. Klep and Ida H. Stamhuis, eds) (Aksant Academic Publishers), published in 2002.The inclination to view issues in statistical terms and its development in the pre-statistical era between 1750-1850 is validated in this landmark study. A quantifying spirit began to develop during the eighteenth century that manifested itself in numerous distinct areas, such as mathematics, astronomy, standards of living, mortality, water levels, population, taxation, finance, insurance, trade, and crime. This study is the first to analyze measurement, theoretical statistics, and statistical activity as one phenomenon in all aspects. It has profited from the expertise of a variety of qualified historians: economic and social historians, historians of demography, of natural science, of mathematics, and of law. The contributors pay attention to the extent to which a typically Dutch statistical mind developed, while at the same time providing insight into the nature of influences from abroad and their possible interactions.   

The Statistical Mind in Modern Society: The Netherlands 1850-1940, 2 volumes by Jacques van Maarseveen (Editor), Paul Klep (Editor), Ida Stamhuis (Aksant Academic Publishers) The contributions in this first volume, produced by experts from various disciplines, cover a great diversity of topics. In addition to the institutionalisation and internationalisation of official governmental statistics, attention is paid to statistics sup-porting policies for modernising society, in areas like agriculture, social legislation, education and justice. The application of statistics in trade and industry (such as banking and insurance, and the railways) is also discussed, as well as the growth of state power to combat social and economic problems such as child labour, the fight against alcoholism and economic crises.

The contributions in the second volume, produced by experts from various disciplines, cover a great diversity of topics. The application of statistics in the sciences (demography, geography, genetics, economic historiography, agricultural and medical sciences) is discussed in an international context. Special attention is given to the general emergence of thinking in terms of probabilities and the influence of mechanisation in statistics, as well as to the way Dutch scholars and scientists tried to solve statistical measurement problems (in meteorology, demographic forecasting, business-cycle research and unemployment).

In this review the following topics are discussed. First we examine the notion of the 'statistical mind'. Then we look the place of this publication within the historiography of statistics and brief summaries present the various contributions.

The statistical mind

In the period 1850-1940 statistics gradually became more and more important in society and by the end of the period it was deeply rooted. It influenced the way people looked at their social environment, how they formulated problems and how they tried to find solutions. Statistics could mean different things to different people and in different periods. Some statistical tools were only developed towards the end of the period under review. We can therefore speak of a variety of statistical mindsets, different mental attitudes involving different forms or different aspects of statistics. The term 'the statistical mind' refers, somewhat metaphorically, to this variety of mindsets. The history of the statistical mind can be viewed, we propose, as a branch of social history in which the succession and sometimes rivalry of contemporary statistical mindsets is traced.' The statistical mind embraced on the one hand a style of scientific thought and on the other hand a style of practical description and imagination about the social environment. It is always connected with concrete problems, decisions and actions.

The notion of 'statistical' in the terms 'statistical mind' and 'statistical mindset' refers, as was the case in The statistical mind in a pre-statistical era, to what was considered to be statistics and what was practised as statistics in the period under review. Just as in the proceeding period, the concept of statistics was multi-layered in the period 1850-1940. The different layers coexisted alongside each other, and their relative importance shifted over time. What layer predominated, depended on the sector of society or science in which it was present and the period under study.

In the period 1850-1940, statistics was still occasionally considered the general description of society, qualitatively as well as quantitatively. Statistics in this sense was also called Staatenkunde (state description); it had been dominant in the period 1750 to 1850. It can still be discerned at the beginning of the period 1850-1940. Statistics in the sense of Staatenkunde was therefore one layer.

More common was the view that statistics consisted of numerical representations of mass phenomena with a tenuous link to probability theory. This concept of statistics was present, mainly in official circles, in the first half of the nineteenth century, before the period covered by this publication, but it was not yet dominant at that time. It was stimulated in the second half of that century by the growing need for verifiable, 'objective' information, by the demand of official institutions amongst others. This 'descriptive' quantitative statistics was the second layer and a quantitative continuation of the previous Staatenkunde.

Thirdly statistics was viewed as the evaluation or analysis of quantitative statistical information with the help of probability theory. This third concept of statistics had also been present in the period from 1750 to 1850, but had not been very influential. In the second half of the period 1850-1940 this last idea of statistics would no longer be linked to 'mass' phenomena, but would appear to be more generally applicable to quantitative observations of phenomena in general, often of a scientific character. Old and new characterizations of sets of statistical numbers, like mean, variation, correlation, regression, index numbers and probability distribution, were used in such an evaluation. Internationally this third layer would develop strongly during the first half of the twentieth century. This publication will make it clear that it made only minor inroads in the Netherlands in this period.

A statistical mindset in the sense of a style of scientific thought is to be found in the emergence of statistics in the thought of political theoreticians, as was shown by Robert Schware in 1981 in his study of T Hobbes, K. Marx and J.S. Mill. Similar, even clearer, results are to be found in historical studies of statistical thinking that emphasise probability and statistics as tools for sum-marizing and analysing experimental and observational data. Examples are to be found in various publications, the most striking of which is the second volume of The probabilistic revolution, entitled Ideas in the sciences.

This approach offers the important insight that the 'statistical mind' employs one of the six 'styles of scientific thinking' distinguished by Alistar Crombie in 1993, that of statistical analysis, alongside postulation, experimental argument, hypothetical modelling, taxonomy and historical derivation.' The development of a 'statistical mindset' viewed as a style of scientific thought is encountered in a number of contributions in this publication, especially in the second volume. Statistics is here presented as such a style of thinking, interactive with science, and involving phenomena like debate, resistance, unevenness, and practical difficulties.

However, this is only part of the story. The 'statistical mind' also embraces a new style of practical imagination, description, and application, often con-structed and used by non-statisticians and non-scientists. This was apparent in a rapidly increasing number of themes in concrete political, social and commercial contexts in which numbers and numerical arguments were used. In this fast growing domain of society a form of statistics was often practised that was not directly connected with great scientists, but that did involve sometimes-advanced, social and technical forms of observation, comparison and application, not only in government departments but also in political and social organisations and business management. This growth of the 'statistical mind' is especially encountered in the contributions to the first volume.

In our view the history of the 'statistical mind' should be viewed as a history of the social construction of a statistical 'reality', in the sciences as well as in political, social and economic life. The historical course of this construction is partly determined by the questions, challenges and constraints of individuals and organisations in the social environment.

Pierre Bourdieu's theory of habitus provides a wonderful integration of sta-tistical thinking and practice. The development of the statistical mind can be viewed as a new combination of thinking and practice, a new habitus.° Bourdieu appreciates individual operations of institutions or persons who structure and restructure the field. Habitus is a 'socialized subjectivity' that accounts for the fact 'that, without being rational, social agents are reasonable."

This new habitus is not meant as a personal characteristic of individuals, but as a phenomenon that is deeply rooted in society. The enormous diversity of observations became measurable by a common standard, an 'objective' procedure. In order to become commensurable, information had to be processed and a wide area was opened to the study of mass phenomena and to the testing of hypotheses. The development of this new habitus had far-reaching consequences for existing discourses, ways of making decisions and for undertaking actions. Existing moral and ideological views and representations of phenomena and the conclusions drawn from them experienced heavy competition from often successful or at least challenging conclusions based on quantification. The new approach distanced itself consciously — and not without struggle — from styles of thought that based themselves on the unique case and deductive method. People and observations were considered in a different light and texts were read in different ways.

There have always been historical circumstances — for example times of emergency and crisis — when more counting and quantification was under-taken than at other times, and this was no different in the period 1850-1940. Crisis periods with markedly increased counting activities were the agricultural crisis in the 1880s, the First World War and the economic depression of the thirties. At the same time the 'statistical mind' was stirred and stimulated in novel ways and deepened as a result. After 1850 statistical reflection and activity experienced the stimulating impact of broad modernising changes in society, such as the growth of the nation state and democratisation, bureau-cratisation, the industrial revolution and urbanisation, new views of social justice and social progress, and the rise of experimental scientific research. A new Lebensgefiihl came into existence.'

It is interesting that after about 1890 the 'statistical mind' seems to have moved closer to the centre of power in society. Something had changed. Important social topics in vital, medical, social and economic statistics had moved into the limelight, including death rates and causes of death, diseases, illiteracy, hospital statistics, labour, wages and prices. They had been made visible through quantification. A similar process took place in the scientific world. People had begun to believe in 'objective' statistical representations of reality and in quantifiable abstractions as a result of which the influence of traditional moral discourses diminished. They sometimes developed strong emotions and displayed purposeful behaviour when they were confronted with certain numbers, percentages and statistically constructed connections. Their public and scientific discourse was gradually cast in the language of statistics, which rationalized decisions and action. It helped to shape political goals, increased public control over citizens, introduced effective management methods and created answerable research questions. As a result, around 1940 quantification can be considered one of the dominant ways in which Dutch society structured and negotiated daily, political and scientific life.

How are we to account for the appeal and power of quantitative methods? Why were they seen as desirable? We should not see this as an element of scientific drive only. This was not only the product of a particular scientific style or culture. Political and social practices are extremely important in order to understand this, in addition to the pursuit of objectivity. This collection offers detailed insights into the accelerating drive in the nineteenth century towards improving social efficiency. Why would dull statistics help the quality of democratic discussion, promote social progress, produce a fairer distribution of scarce products, and better management? The need to communicate in an efficient way increased the attractiveness of 'objective' statistics. The emerging idea of 'objectivity' is closely connected with the 'statistical mind'.

However, the statistical approach is not just an intelligent method of rep-resentation and communication. By quantifying on the basis of concepts and by making information commensurable, the statistical approach is a deter-mining factor in the way problems are formulated, measuring concepts are shaped, and observations are collected, processed, compared and analysed. Although persons and institutions used statistics from the unreflected assump-tion that they were being 'objective' and that their material was indisputable evidence, this was actually socialized subjectivity. Quantification and statistics are produced in a process of social construction, but they are not 'innocent', value-free. Nevertheless, statistics transformed explanation, reasoning and action in science and society. The objectivity and success of statistics pro-vided legitimations and conferred power to those who used it: the state, the managers and the scientists.

This exploratory collection concerning one country, the Netherlands, in the period 1850-1940, the first of its kind, has the society-wide interpreta-tion of statistics as its central theme. This historising approach proceeds interactively. It relates problems in society to statistical thought, but it also looks at the effect of statistics on society. The modernisation of statistics is also thematised.

How is this publication related to the historiography of statistics and the re-lated concepts of quantification, chance and probability? In the first section, dealing with the statistical mind, the cultural approach of Crombie — who distinguishes different styles of scientific thought — was characterized as too narrow to encompass the subject of this publication: statistics in its different meanings as well as its embeddedness in practice. Bourdieu's theory of practice and his idea of habitus seemed a more appropriate way to describe what exactly the 'statistical mind' is about, and to highlight the fact that it involves both mindsets and actions, and that it is both a product of social construction and an innovative element in modern society.

We are not the first scholars interested in the broad field of the history of quantitative, statistical and probabilistic thinking and practising. This publica-tion has benefited from the attention statistics and probability have received in recent decades. A range of books and articles dedicated to these topics inspired us." Nonetheless both this and the preceding publication have their own characteristics. We will indicate several aspects to make this clear.

In the first place in our study the emphasis is not on the how and the why of the rise of probability or chance. Chance and probability are often used as synonyms, but a distinction can be made between probability as 'subjective' and chance as 'objective'. The word probability is then used when referring to the belief of people that some event will happen or not, and chance when referring to the possibility of the occurrence of the event.

Generally speaking earlier than statistics, probability and chance drew scholarly attention. In 1975 Ian Hacking gave impetus to the field through his book The emergence of probability." A group of publications appeared at the end of the 1980's: the two edited volumes of The probabilistic revolution, published in 1987, Lorraine Daston's Classical probability in the Enlightenment in 1988, The empire of chance in 1989 by Gerd Gigerenzer et al. and The tam-ing of chance in 1990, again by Hacking. Central in these publications were the development of the concepts of probability and chance and the influence these developments exerted in science and society.

Chance on the one hand and statistics on the other hand are connected. The chance of an event can be estimated by its frequency in the past. Statistical data show up frequencies and chances can be estimated by determining frequencies. The consequence of this connection between chance on the one hand and statistics on the other is that, when chance and probability are being studied, statistics must usually be taken into account. And this is what is seen in the publications on chance and probability: statistics received attention as well, because its results served the increasing insight of probabilistic thinking.

Secondly, once it is concluded that in the study of probabilistic thinking the compiling of statistical data must also be taken into account, it is under-standable that in the 1980s, when chance and probability were more closely examined, greater understanding emerged about the processes of quantification and the compilation of statistical data and the contexts in which they took place. Scholars became conscious of the complexities of these processes. This is one of the possible explanations of why, over time, the focus of scholarly attention has shifted, although not absolutely. In a move away from the focus on chances and probabilities, the increasing use of numbers and statistics in science and society was studied. These studies naturally also drew attention to the sciences and the societies of which these numbers and statistics were an expression. Therefore the economic, social and scientific contexts were increasingly taken into account.

An interesting early example is Donald MacKenzie's book of 1981, Statistics in Britain, 1865-1930 (Edinburgh University Press), in which chance still plays an important role, but which is one of the first efforts to try to explicitly explain the interdependency of statistics and the social and political context in which it developed This was especially interesting because the author focused on what is now recognized as the emergence of mathematical statistics in England around 1900, while mathematics is usually considered the science for which interdependence with its context is the most difficult to demonstrate.

Although a greater interest developed in statistics and the environment in which these statistics were embedded, nevertheless chance and probability continued to play an important role. In 1986 Theodore Porter published The rise of statistical thinking, in which he discussed statistical thinking as it developed among social scientists, biologists and physicists.41 In the statistical thinking he discusses chances and probabilities were essential. This was also the case in Alain Desrosieres's textbook on the History of statistical reasoning of 1993, which appeared in English in 1998, and which was of great help in understanding the international circumstances in which Dutch developments took place.

Thirdly, many contributions in this publication deal with the massive expan-sion of statistics in the sense of quantitative expressions of mass phenomena in the period 1850-1940. This is connected with a historiography that examines the increasing use of, and belief in, numbers in society. After all, some of the publications on statistics in this tradition focus more on the use of numbers and quantification while probability and chance were less dominant. In 1995 Porter published Trust in numbers. Recently an issue of Centaurus paid attention to the importance of institutionalisation of statistics and statisticians in the nineteenth century.

Fourthly, this publication combines and relates official statistics to the socio-economic development of the country and the use of statistics in the sciences. In doing this we have taken a somewhat different course than is often done in the historiography of official statistics, which has been influenced strongly by the idea of the rise of the nation state. Recently several books have appeared that are dedicated to statistics in a national context. 44 Numbers and nationhood, published in 1996, discusses statistics in nineteenth-century Italy. Statistics and the German state of 2001 focuses on German statistics between 1900 and 1945. The politics of population of the same year is about the population censuses in Canada between 1840 and 1875. Disciplining statistics, published in 2006, compares demography in France and vital statistics in England between 1830 and 1885 and Een monument voor het land (A monument for the country) of 2008 discusses official statistics in Belgium between 1795 and 1870. In all these books official statistics plays a prominent role. They all argue, more or less explicitly, that official statistics was essential in the nation-building of these countries.

That the Netherlands is no exception can be concluded from several publications: two books dedicated to the Dutch Central Bureau of Statistics (CBS), from the chapters about official statistics in The statistical mind in a pre-statistical era, as well as from the chapters on official statistics in this work.45 However, in the present publication the emphasis is not on the role of statistics in nineteenth-century nation building, and the scope of this publication is expressly not limited to official statistics. It is rather the interaction between official statistics and the public debate that we focus on. In the design and the public effects of official statistics, social and economic questions (and the link with the idea of social progress) play an important role.

The comparison of our work on The statistical mind with The probabilistic revolution illustrates these points. A first impression may suggest that the similarities are predominant. Both edited publications look similar; they both consist of two volumes and several parts. The two publications are divided into two volumes in corresponding ways: volume one discusses the role of the central concept, the probabilistic revolution and the statistical mind respectively, in society, and the second volume discusses the role of this central concept in the sciences.

However, when we look more precisely the differences are obvious. In the first study, probability as a new style of scientific thought within various scientific cultures is central, whereas in the second work quantification and the production of statistics in practical contexts are dominant. In the Probabilistic revolution the different forms of quantification and the production and application of statistics play a minor role, while in the Statistical mind probability and chance are present, especially in Volume II, but only as a part of the story.

Another obvious difference is the focus on the historical embedding of statistics in the history of a particular country, the Netherlands, in the publication on the statistical mind, whereas in the publication on the probabilistic revolution the probability questions in different western countries are reviewed in a comparative manner. Statistics seen as part of a newly evolving habitus in social and scientific practice is apparently more connected to a specific national context than probability and chance.

The differences between The probabilistic revolution and The statistical mind reflect the shift in focus and interest this field of history has experienced: from probability and chance to quantification and statistics, from a scientific-conceptual approach to a historical-sociological one, and from general to national.

These studies contribution to the history of statistics

Volume I, Part 1. Official statistics

From the period 1795-1850 the Netherlands inherited a legal system of official administrative statistics the purpose of which was reporting by, and checking of, local and provincial authorities. By legal decree and with the help of prescribed model tables, lower authorities had to inform ministries in The Hague. The Dutch Republic of the Seven United Provinces (1581- 1795) had no tradition of strong centralisation: the cities and provinces were highly autonomous. In such a context a structure of official statistics hardly developed. Only the Batavian Revolution (1795) established a strong central authority that needed objective, standardised information about the state bureaucratic machinery and about social conditions. Administrative counting played an important role in this process. As a result the nation state of the Kingdom of the Netherlands, created in 1813, acquired a statistical 'face', an 'objective' basis for accountability and debate in a gradually increasingly democratic context. As Randeraad, Bracke, Patriarca, Curtis and others have shown, in many countries official statistics were an important factor in the process of nation building.°

Counting and compiling tables was usually done by the individual mu-nicipalities, by public services or by subsidised institutions like poor relief funds. The collecting and processing of part of these official statistics into national overviews, as well as the rather limited publication of the data was, until 1850, carried out in the Netherlands by individual government departments." After 1820 this decentralised form of organisation led repeatedly to the question of whether the processing of the incoming information should be centralised. Centralisation of the control and processing of administrative statistics would make uniformisation and professionalisation possible so as to bring to an end the disjointed and incomplete nature and the poor quality of official statistics.

However, for a long time nothing was done. In this respect the Netherlands lagged behind the surrounding countries in the period 1840-1892. In these years central statistical commissions (in Belgium and Austria) or central statistical bureaus (in Denmark, Norway and Sweden) were established, while in other countries (including Germany, France, and Great Britain) statistical bureaus were created, which advised on various departmental statistics, which were also called official or national statistics.

The term official statistics, also called national statistics, implies a distinction from the provincial and municipal statistics. Large amounts of statistical material were incorporated in the provincial annual reports and — particularly after the introduction of the new constitution of 1848 — in the municipal reports." Provincial bureaus for statistics were decreed by law in 1850 and officially instituted in 1858. However, they never fully developed. In addition to collecting information for government statistical reports and the annual Provincial Reports, they were charged with all kinds of administrative tasks that had nothing to do with statistics. When the collection and processing of data was transferred to the newly formed CBS in 1899 they lost their raison d'être and in 1905 they were abolished. Municipal statistical bureaus were only created in the large cities. In the first half of the twentieth century the statistical bureaus of the cities of Amsterdam and The Hague in particular were of great importance, as they compiled price index figures, which compensated to a certain extent for a lacuna in the official statistical data in this area.

Two aspects of Dutch official statistics in the period 1850-1940 are explored in these contributions: the institutionalisation and reorganisation of official statistics, and the development of the programme of the statistics. Concerning the institutionalisation special attention is paid to the rise of international statistics and its influence on developments in the Netherlands, and to the contributions made by the Dutch in the international arena. Concerning the programme of the government statistics, the development in a general and in a more specific sense is examined.

First the institutionalisation. There is an enormous contrast between the state of government statistics at the eve of the Second World War and that around the middle of the nineteenth century. Around 1940 Dutch official statistics thrived. Official statistics were, with the exception of agricultural statistics, designed by the Central Commission for Statistics (CCS, 1892) and compiled by the Central Bureau of Statistics (CBS, 1899).

The creation of these organisations had not been an easy process. Why did this take such a long time, Ida Stamhuis asks in The long road to an enduring national organisation of statistics. The socialist member of parliament Domela Nieuwenhuis had asked the government three times, in vain, to create a national statistical organisation, as he was convinced that such an organisation was necessary and inevitable. But was such a development really inevitable, the author wonders. What were the arguments adduced in favour of and against institutionalisation, and to what extent were party-political positions the cause of this slow development?

The creation of the CCS (1892) and the CBS (1899) cleared the way for the revision and extension of government statistics. What did this institutionalisation amount to and what form did it take in the course of time? The two organisations were given the status of scientific independence. What exactly did this imply? In what sense did these institutions, which were managed

by statistical experts, guide the programmatic development and expansion of government statistics? In For practice and science. The institutionalisation and expansion of Dutch official statistics (1892-1940), Jacques van Maarseveen discusses these developments. He explores what political and economic factors played a role in the development of these institutions and to what extent the centralisation of statistics gave rise to further professionalisation. An important question is also to what extent major international events such as the First World War and the depression of the 1930s influenced the statistical programme.

In Europe there was great deal of activity in the second half of the nineteenth century in the area of official statistics and the improvement of the comparability of statistics. There were regular international statistical meetings. As a result of the rise of liberalism there was an atmosphere of ever increasing political openness between the European countries. The aim of these statistical meetings was, Nele Bracke explains in In search of comparability. The internationalisation of official statistics (1853-1945), to enhance the international comparability of statistics. She discusses the attempts to do this in a number of major areas of statistics (censuses, mortality statistics, population data). She confines herself to `internalist history', focusing on the international community.

What influence did these developments have on the Netherlands, a country that traditionally had strongly interactive relations with other countries? This 'externalise history, which centres on the impact of the international meetings and agreements on national statistics, is discussed in Jacques van Maarseveen's contribution Co-operative but ambivalent: Dutch international statistical relations (1850-1940). Although the Dutch were always actively involved in interna-tional statistical cooperation, the influence this had on official statistics and on the organisation of statistics was initially limited, even though the Netherlands organised an international statistical congress in 1869. A change in this situa-tion occurred only with the creation of centralised statistical institutions at the end of the nineteenth century. From then onwards the Netherlands partici-pated actively in the international harmonisation. In addition the Permanent Office of the International Statistical Institute (ISI) was established in The Hague in 1913. To what extent was this fact of influence on Dutch statistics and what contributions were made from the Dutch side in this cooperation? Or did the international influence remain limited even then, as is suggested by the fact that in the mid 1930s complaints were still being voiced in Parliament about the absence of Dutch data in international publications?

In addition to the institutional element in the history of Dutch official statistics, there is also the question of how these developed in the period 1850-1940. This development was closely related to that of government administration and policy. In 1940 the government produced many more statistics than a hundred years before. The Central Bureau of Statistics alone compiled more than a hundred statistics that covered a wide area of demo-graphic, social and economic themes. Methorst published an initial survey dealing with the nineteenth century in his Geschiedenis van de statistiek in het Koninkrijk der Nederlanden (1902), while for the twentieth century a number of studies have been published recently. 52 Four contributions in this publica-tion are devoted to the evolution of four classical administrative statistics that had been produced since the Batavian-French period (1795-1813): education, agriculture, justice, and alcohol consumption. Topics that gave rise to debates about these statistics were the aim and usefulness of statistics, the method of data collection and the presentation of the results, the quality (precision) of the data, the comparability of data over time, and sometimes also the cen-tralisation of statistics.

From 1860 a fierce political battle erupted in the Netherlands about educa-tion. The education statistics, which had been compiled since the French-Batavian period (1795-1813), experienced the repercussions of this conflict. In Education policy and the growth of a statistical mindset', Hans Knippenberg and Kees Mandemakers describe this episode, and they investigate with what political ends these statistics became interlinked, and how the collection of data was gradually expanded and professionalised. The education statistics played an important role in the public debate, especially during the school dispute (schoolstrijd) between liberal and more religiously inspired political groups and in discussions about child labour and school attendance.

The relation of agricultural statistics with public debates and agricultural policy is investigated by Albert Niphuis in his contribution on Agricultural statistics. A matter for agriculturists. In public discussions of these statistics the way the data were collected and the quality of the statistical data — as col-lected by the Department of Interior Affairs — were criticized. The role of the agriculturalist was emphasised over that of the statistician. The late transfer of agricultural statistics to the CBS in 1941 is curious. Why did this take so long, the author wonders, what motives played a role? He also considers what results the debates about the qualitative aspects had in the end.

Judicial statistics had also been published since the French-Batavian period. After 1850 they began to play a greater role in professional circles of jurists and lawyers. Attention is devoted to the reception of these statistics by Sjoerd Faber and Harry Eggen in their contribution 'Love without passion. How judicial statistics were received, in particular in the Weekblad van het Recht (1850-1935). In this way the development of judicial statistics is indirectly sketched as well. The authors wonder why the reactions to these statistics in the periodical they discuss were so variable, sometimes enthusiastic, sometimes very lukewarm and reserved. They also pay attention to the discussion about the way the data were analysed and presented by the CBS and the cutbacks that had to be made in the publication of the figures. Cutbacks are, as a matter of fact, a recurrent theme in the history of official statistics, a theme that is especially highlighted in the contributions about the institutional history referred to above. A point of investigation is to what extent financial resources determined the ups and downs of official statistics, and to what extent war and economic depression had an influence.

How old administrative statistics can suddenly acquire a novel significance and how the quality of the figures can, in that case, suddenly become a hot topic is described by Ronald van der Bie in Old gin in new bottles. The official debate on the reliability and production of statistics on the consumption of spirits in the Netherlands (1917-1929) This contribution also shows how official statistics can be used for contemporary political aims, in this case the fight against alcohol-ism. How did the CBS, the author asks, try to satisfy the political demands for statistical information, while in previous years there had been a great deal of criticism of the reliability of these statistics. He investigates in particular the method of presentation, as the presentation was meant to enhance the reception of the statistical information in the public debate. The influence of the presentation on the reception of information by the users is also discussed in the contributions on agriculture and education already mentioned.

Volume II, Part 2. Statistics, social progress and modern enterprise

In the period 1850-1880 official statistics were mostly compiled on the basis of records made by local authorities and central services. These so-called administrative statistics were strongly aimed at the control and possibly reorganisation of the state machinery, and not at monitoring society. This changed considerably as a result of the modernisation of Dutch society. Novel, large-scale social and economic challenges arose as well as problems caused by industrialisation, rapid urbanisation, wartime conditions and intensified competition from abroad.

The pressure on the government to tackle social problems was already mounting in the 1870s, but the government and the majority of conservative liberals and confessional conservatives took the position that it was not the duty of the state to solve socio-economic problems, and therefore they refused as much as possible to undertake statistical research into social problems. They recognized the merits of existing administrative statistics as supporting 'objective' political publicity about the state and a rational debate on state matters, but as far as possible they avoided individual studies by the state into social problems.

How was it possible that, notwithstanding tenacious resistance on the part of government and parliament, large-scale studies were made of child labour (1863), working conditions (1886-1889) and the agrarian crisis (1886)? How did a Dutch governmentality — a new science of governing a population — arise which produced entirely new statistical knowledge, which changed the political discourse about social problems, and in addition increased the power of the state over its citizens? The influence of the state had grown considerably, partly as a result of the new statistics. Such an increase in the power of the state had earlier only occurred in the Napoleonic period under conditions of occupation and war, and had subsequently, after 1813, led to a major reaction. Why was the increased power of the state then accepted after 1890? In Governmentality, statistics and state power Dutch labour and agricultural inquiries (1840-1914) Paul Klep discusses the fundamental significance of the ever-stronger idea of social progress. He examines how new statistical knowledge changed the discourse on social problems. This new knowledge made Dutch society transparent as an 'objective' system, ready for change. In addition the new statistical information created new technologies of governance and new institutions of state power in the various ministries.

In the name of social progress and — during World War I — of the national interest, state control supported by statistics expanded in the period 1890-1918 not only to the remote corners of bureaucracy, but also increasingly to citizens themselves. How could one shirk providing detailed statistical information to the government? Loes van der Valk shows this in her contribution Private or public? The Dutch debate about social insurance statistics (1900-1940). She discusses the implementation of the Workman's Compensation Act of 1900. How did employers manage to avoid becoming involved in the functioning of a bureaucratic state agency (the Rijksverzekeringsbank) and so avoid the state getting to know 'everything' about them? The solution that was found for some of them, private registration, was the reason why Dutch public health statistics remained very incomplete.

Were social statistics only created by the state? Certainly not. The early Dutch labour movement was convinced that statistics could be of great help to show the general public and Parliament in an 'objective' way how bad social conditions were in the Netherlands, and how ripe they were to be reformed. The example of the very emotional Dutch public debate on child labour statistics was a proof of this. Dutch middle-class advisors with social-liberal ideas and foreign socialist inquiries inspired the Dutch labour movement to conduct their own inquiries. Was this movement successful? Sjaak van der Velden explains in his Statistics and the early Dutch labour movement (1870- 1918) why it was not, and how welcome the CBS was, a government body, which from 1899 onwards developed an additional, impressive systematic programme of social and economic official statistics.

The same resistance caused another problem: the failure to hold a fully-fledged census of industrial firms. Although traditionally agriculture and trade had been the economic strong points, after 1870 industry became a very important production sector of the Dutch economy. Before 1930 a statistical picture of industry had to be composed from a large diversity of local administrative statistics, the registration of steam engines and occupational censuses. The sometimes rather intrusive labour surveys of the 1880s were not calculated to enhance the cooperativeness of the industrialists. In his contribution A statistical latecomer: Dutch industry in figures, Jacques van Gerwen explains why, despite strong persistent pressure by the CCS, the first business census only took place in 1930.

Little is known about the role of statistics in Dutch cost accounting and managerial control practices. The example of the United States shows that the rise of large transportation, production and distribution enterprises in the period 1850-1925 was an important stimulus for the systematic application of statistics in trade and industry. The processing of company information for internal planning and control was of great importance in the railways, but also in mass production enterprises. In his famous study of the managerial revolution in American business, Alfred D. Chandler Jr. shows how statistical reporting supported cost accounting, managerial control and forecasting in the railways, the Carnegie Company, Du Pont, General Electric, General Motors and Standard Oil. Later on the scientific management movement was at the forefront of the statistical analysis of cost accounting aimed at the improvement of productivity.

In industry in the Netherlands, companies did not develop on the same large scale as in the United States, and the bureaucratisation of society was manifested itself mainly in a growth of the number of accountants." While today statistics is often an obligatory subject in management courses, in the period 1850-1940 there was only pragmatic learning on the job. It is not known to what extent statistical analyses of accounts were used to carry through cost-saving measures and to make the right investments.

Using statistical expertise — even mathematical statistics — in order to market a profitable product was a well-known activity in insurance. Already in the seventeenth and eighteenth centuries the practice of Dutch insurance stimu-lated statistical innovations in the field of probability. Christiaan Huygens and Johan de Witt were active in the late seventeenth century, Willem Kersseboom and Nicolaas Struyck around 1740, Jan Hendrik van Swinden around 1800, and Adolphe Quetelet and Rehuel Lobatto around 1850.

There are three contributions in this publication that concern, respectively, the railways, the banking industry and the world of insurance. The Dutch railway companies needed statistical information for an optimal running of their networks. How did they control the highly decentralised activities of their business? How could they monitor the railway traffic and adapt the system to changes in the pattern of travel and the transport of goods? Augustus J. Veenendaal Jr. explains in his Dutch railways and statistics (1850-1920) how the companies collected incredibly elaborate data, which provided particulars of every aspect of the workings of the railways. How secret was this information and why were many of these statistics published in annual reports? Another question is if business statistics was more developed than in other European countries.

The banking sector in the Netherlands produced many statistics, but before 1936 gave them little publicity because of both formal and informal institutional constraints. Only De Nederlandsche Bank (DNB) provided detailed figures about its business process in tables and graphs. Joke Mooij explains the reason for this in her contribution Statistics: a means of communication. The Bank Act of 1863 marked a turning point in the publication of statistical information pertaining to banking. Later on the DNB was to serve as collector and intermediary in providing statistical data on the banking sector to the CBS. How did the DNB deal with two apparently conflicting requirements to the bank: one of transparency and accountability, and the other of maintaining bank secrecy?

Traditionally, calculations that formed the commercial basis for the annuities of life insurance companies were made by mathematicians. They functioned as actuarial advisors to these companies. In the 1920s it became clear that mathematicians and actuaries were growing apart. Why? Danny Beckers explains in his Actuarian science, mathematics and statistics. The Association of Mathematical Advisors for Life Insurance Companies (1889-1920) how during the period 1850-1940 some of the actuaries developed specific applied statistical tools. The demands of the competitive environment of business increased the sophisticated application of mathematics in the treatment of statistical information about insurances. Many mathematicians were not tempted by these developments and developed no interest in the mathematical treatment of statistics.

Volume II, Part 1. Statistics and sciences

In the first part of Volume II the 'statistical mind' in academia comes to the fore. In this part it is investigated how statisticians and other scientists extended the influence of statistics in academic disciplines. When the university system had been remodelled following the example of its German neighbour in 1876, a new style of scientific reasoning and practice could develop. Foreign influences were strong. In this section the various contributions show how, in an interactive way, a new scientific habitus of statistical reasoning developed in various sciences, although taking very different forms.

Around 1850 statistics was not accepted in the medical sciences, with the exception of the subdiscipline of social medicine, which focused on populations and not on individual patients. From 1880 onwards some research-oriented medical doctors began to criticise the traditional attitude of doctors, involv-ing the idea that each patient is a unique person and that the doctor's art is to relieve the patient's suffering. In The rise of quantification and statistics in Dutch medical research (1850-1940) by Paul Klep and Brand Kruithof the following question is examined: how was it possible that quantification and probability became so important in the medical sciences and that the habitus of research-ers was transformed from styles of thought like taxonomy and postulation to statistical analysis? Through quantification, the personal experience of the doctor and the characteristics of the unique patient could be objectified. In addition, 'objective' and testable statistical statements about the variation of bodily processes, types of diseases and optimal treatments became possible. Why did statistics win the day, and why, in the end, did doctors become com-petent and confident in the use of statistics as a style of scientific thought?

What happened with the Staatenkunde, which was discussed in the earlier publication on the statistical mind?" Originally the institutionalisation of Dutch geography predominantly reflected the Staatenkunde tradition: categorising, measuring and quantifying regional entities without much theoretical basis. Hans Knippenberg investigates The transformation of the statistical mind in (human) geography. How did it happen that, although statistics and geography originated from the same field of knowledge, they grew apart? Knippenberg's contribution discusses how specialisation and diversification of the discipline in the Interbellum gave new impulses to a transformation and deepening of the 'statistical mindset' in geography by introducing new measurement concepts and related statistical techniques, new research methods and more formal ways of theorising. After the Second World War, this development culminated in a limited version of the Anglo-Saxon inspired 'quantitative revolution'. Can this process be understood by the fact that Dutch human geography had already developed a relatively strong tradition of (applied) empirical research?

To what extent was the application of statistics in the sciences accompanied by the demise of certain scientific views? In her contribution The statistical mind moulding heredity: Hugo de Vries and Mendelian genetics, Ida Stamhuis argues that, for the study of heredity to be transformed and become statistical, it had to go through several stages. Its focus had to be narrowed to transmission genetics, ignoring questions of development. The feeling had to emerge that purely theoretical considerations were no longer convincing, if they were not supported by experimental evidence. In addition, the experimental results had to be quantitative and the relevance of experimental statistical designs and probabilistic interpretations had to be accepted. The case of Hugo de Vries gives an inside view of the working of the statistical mind in the emerging scientific field of genetics. Why did he consider the costs of moulding his field of study in a statistical framework too high? The acceptance of the statistical style of thinking in genetics was apparently not unproblematic.

This raises the question: how did other experimental sciences become con-nected to statistics? Harro Maat discusses in Statistics and field experiments in agriculture. The emerging discipline of inferential statistics the close connection of the design of field experiments in agriculture with the emerging discipline of inferential statistics. His chapter analyses this process for the Netherlands and the Netherlands East Indies. In agriculture inferential statistics dealt with the question: how can the effect of one factor be measured in an experimental field situation where many factors are at play? To investigate this successfully, a specific mathematical procedure as well as an organisational structure was required. How could both elements be integrated in an existing experimental practice. Why did it turn out to be easier to implement a new statistical approach in the colonies than in the mother country? The answer, it is argued, is that an aspect of experimental practice was that the links between trained agronomists and (lay) farmers were closer in the Netherlands than in the colonies.

Did the statistical mind penetrate all the sciences? Why, as Leo Noorde-graaf demonstrates in Statistics and economic historiography in the Netherlands (1850-1940), did sometimes only very elementary quantification find its way into a discipline? Historical derivation is an old and quite distinct scientific style of its own, as Alistar Crombie has shown, which focused more on the genetic understanding of the unique person, case or institution than on the interpretation of large numbers of observations. This does not of course, necessarily mean that the effects of statistics in economic historiography are unimportant. Noordegraaf shows that in Dutch economic historiography the introduction of statistics in the nineteenth and the first half of the twentieth century was a slow process and that the statistical style of thought did not gain a significant position. There were many economic-historical studies but in a large majority of them quantitative techniques did not play an important role. What was the connection with the phenomenon that only a limited statistical repertoire, mainly the analyses of figures, was used?

Volume II, Part 2: Statisticians at work

To what extent did the practice of the statisticians change? The second part of Volume II discusses statisticians and scientists at work, extending the influence of statistics into various sectors of science and society. The first paper Why did the Dutch Statistical Society abandon Statistics? written by Ida Stamhuis, sketches the foundation, history, and demise of the Dutch Statistical Society, and compares some of its characteristics with the English Royal Statistical Society. Why did the society change its name and why did it terminate its involvement in statistical matters once a central, national statistical organisation had finally been established in the Netherlands in 1892? The lack of well-organized, state-generated statistics alone does not seem reason enough to explain this influential choice. Can a comparison with its English counterpart help to explain it?

Henk de Gans, in his Demographic forecasting as a statistically controversial affair (1920-1940). A case study demonstrates another way of extending the influence and scope of statistics. This case study shows that there were differences of opinion between prominent official statisticians of different countries about the content of statistics. Leading Dutch statisticians were of the opinion that statisticians should refrain from making population forecasts, because of the inherent speculative nature of extrapolations into the future. Statisticians had to provide for accurate statistics that were reliable. The image of statistics had to be one of trustworthiness and population forecasting did not fit this expectation. The prominent Dutch statisticians were successful in giving statistics the image of a trustworthy discipline. How was it possible that, notwithstanding their negative attitude towards population forecasting, the leading Dutch statisticians made considerable contributions to the innovation of population forecasting methodology? In this chapter the dilemmas these statisticians faced are discussed: the situation they were in, the contributions they made, the influence they exerted and the consequences of their policy in the Netherlands after the war.

The following three contributions discuss the introduction of statistical instruments: the method of least squares to measure errors, graphs, and index numbers. Instruments are 'material tools the human investigator uses to disclose, probe, isolate, measure, represent, or otherwise bring to attention the objects of investigation'." Although the statistical instruments discussed here are not material but methodological, they have the same characteristics. Afterthoughts 4 discusses this more thoroughly.

What was the best way to deal with the problem of the measurement of errors? Marcel Boumans discusses in Measurement and error problems (1800-1900). Buys Ballot and Landres critique on the method of least squares how measurement and error problems could be tackled. The method of least squares was devel-oped in astronomy to deal with measurement errors. Boumans demonstrates that in other fields different considerations played a role and different choices were made. The method of least squares is most appropriate for dealing with measurement errors produced by precise measuring instruments under similar circumstances. To reduce the influence of measurement errors the choice of the arithmetical mean is then the best choice. However, in meteorology and actuarial science is dealing with unreliable instruments or with observations only available for different moments in time. What consequences did this have? This contribution emphasises that to make statistics extend its domain into new fields of knowledge, statistical concepts specific for those fields must be introduced.

Another way of extending statistical thinking by means of new instruments is the use of graphs based on statistical data. Henk de Gans and Harro Maas argue in Making things visible: the development and use of the graphical method in the Netherlands (1870-1940), that by means of the development and use of the graphical method, things were made visible that would otherwise easily have escaped notice. The paper discusses why the graphical method in population dynamics and business-cycle research first became successful but later declined in importance. What aims did it serve and why was this approach no longer satisfactory later on? Starting from the observation that graphs are not just tools of persuasion, but can also be investigative tools, it is examined how graphs in population dynamics as well as business-cycle research served as investigative heuristics to unveil dynamic processes that were not easily discerned if the same data were presented in different ways, such as in tables. The inherent limitation of the graphical method to depict complex simultaneous processes made demographers and business-cycle researchers in the 1930s move to an explicit modelling approach.

How could complex, unstable social phenomena be brought into the domain of the 'statistical mind' and thus be objectified? What instruments were developed with that aim? In Measuring unemployment in the Netherlands (1900- 1940). The operationalisation of an elusive concept Peter Rodenburg discusses the measurement of unemployment in the Netherlands between 1900 and 1940, which he characterizes as the operationalisation of an elusive concept. He investigates two measurement procedures — one based on trade union unemployment insurance records, the other on labour exchange data — that were established by the Dutch official statistical office in order to measure the elusive concept of unemployment in the pre-World War II era (1900-1940). How did these measurement procedures help the understanding of unemployment as a social concept? The paper argues that measurement and conceptualisation of unemployment were not different stages in the measurement process, but went hand in hand and turned out to be mutually constitutive.

Another way to extend the influence of statistics is to try to make it easier to produce statistical results and make them available, to produce more statistical information and to make statistics more reliable and cheaper. The last two contributions discuss two entirely different methods which have, however, both these consequences: random sampling and mechanization. Jelke Bethlehem, Ida Stamhuis and Jacques van Maarseveen discuss in Complete enumerations or sampling? The historical debate about sampling for surveys the laborious transition in the official statistical community to move from doing surveys by complete enumerations to the much cheaper random sampling. How can the laboriousness of this process be understood? It appeared to be helpful to take into account the educational background of the statisticians. Why and when did random sampling become acceptable? When official statisticians realized that completeness was no longer possible, why did they prefer purposive to random sampling? The attitude of the newly founded private companies was entirely different. They could do nothing else than use samples. Why, in their discussion about how these samples had to be composed, did randomness only play a limited role?

To produce more statistical information with less effort, mechanization was more successful. Its success probably even delayed the use of the sampling technique. Jan van den Ende, Jacques van Maarseveen and Jan Atsma focus in Mechanization and statistics. The punch-card method in the processing of statistics on the punch-card method in the data processing of statistics between 1899 and the mid-1960s. How can it be explained that the CBS was the first Dutch organization to use punch-card machines? It is striking that the increasing replacement of manual data processing by these machines was not stimulated primarily by their technical improvement, but by specific changes in the bureau's statistical programme. A result of the cheaper and easier production of statistical data was often not that less money was spent, but that more statistics were produced. In addition, mechanization increased the accuracy and therefore the quality of statistics. What was the influence of the labour market, of the organization of labour in data processing, and of the establishment of a special data processing department?