Thomas, R. (1996) 'Statistics as Organizational Products', Sociological Research Online, vol. 1, no. 3, <>

Copyright Sociological Research Online, 1996


Statistics as Organizational Products

by Ray Thomas
Faculty of Social Sciences, The Open University

Received: 21/8/96      Accepted: 30/9/96      Published: 2/10/96


The paper argues that statistics should be seen as organizational products and that growth in the range and variety of statistics testifies to growth in the power of organizations. The paper emphasises the importance of identifying the functions of statistical systems, and recommends a genealogical approach to help identify the unwitting testimony given by the assumptions and motivations associated with the categorizations and data creation procedures used in the production of statistics. The paper examines the motivations, assumptions, and functions associated with statistical systems involving the Census of Population, the British National Food Survey, economic management, and unemployment. The discussion focuses on the evidence these case studies provide on the role of statistics in society and of the influence of organizational meanings on society.

Census; Organizational Coordination; Data Creation; Evidence; Informal/Formal Economy; Organizations; Statistical Systems; Statistics; Unemployment; Unwitting Testimony

Evidence from Statistical Systems

Most textbooks on social research methods have little to say about official statistics or other published statistics. There are usually sections on descriptive and inferential statistics, but, typically, these sections assume that the data to be used has been produced by the user. The prevailing attitude to official statistics among many social scientists is captured in the Penguin Dictionary of Sociology: 'The categorization of data reflects the interest of government and may not be meaningful to sociologists' (Abercrombie et al., 1984).

There is a significantly sized literature on errors in official statistics, see, for example, Jacobs (1984). Texts on secondary analysis (e.g. Dale et al., 1986; Stewart and Kamins, 1993) and other texts concerned with the use of statistics emphasize the necessity to examine the small print associated with the production of statistics (e.g. Fuller, 1977; Reid, 1987; Slattery, 1986). But the literature aimed at evaluating the nature of the governmental interests in official statistics is relatively thin. One of the most cited publications is that by Hindess (1973). Hindess argued that users of statistics should 'take account of the conceptual means of their production' (p. 45), basing this argument on contrasts between the categorizations used in the 1951 Census of Agriculture in India and those that might be required for scientific study.

The book Demystifying Social Statistics (Irvine et al., 1979) constituted a landmark publication in its insistence that statistical method as well as statistics as facts about society and statistical method should be seen as social products. The pseudonymously written Chapter 'How official statistics are produced: views from the inside' by the Government Statisticians' Collective gives a graphic picture of the ways in which the statistics 'imprison us in the concepts and concerns that dominate official life' (p. 138). Another notable contribution was Bulmer's 1980 paper 'Why don't sociologists make more use of official statistics'. Bulmer discusses the various theoretical positions that lead sociologists to neglect official statistics and summarized some of the strength of official statistics.

This paper supports the arguments of Hindess, Bulmer and Irvine et al.'s authors that social scientists should examine, and be critical of, the categorizations used and the data creation procedures followed in the production of statistics. The paper also supports the argument that, even if these statistics are systematically produced, they cannot automatically be considered as scientific. But the paper proposes that social science should go farther. The Dictionary of Sociology, quoted above, claims that the investigation of the social processes that lie behind criminal statistics have proved to be a fertile area of research for criminologists. The argument of this paper is that investigation of the social processes behind most kinds of statistics are a fertile field for social scientists. The growth in the range and variety of statistics as facts about society, discussed in the next section, gives importance to this argument.

The historian Arthur Marwick makes a distinction between the witting and unwitting testimony given by historic evidence (Marwick, 1980). To use statistics, critically or uncritically, is to pay attention to the witting testimony given by the statistics. This paper points to the unwitting testimony given by the assumptions, motivations, and functions associated with statistical systems - asking such questions as 'Why were these statistics produced?', and 'How are these statistics used?' The paper argues for examination of the assumptions and motivations underlying the categorizations used and the data creation procedures followed in the production of statistics, and of the functions of the associated statistical systems.

To anticipate a later example, it is one matter to consider the accuracy and meaningfulness of changes shown by trends in income per capita statistics. It is a different matter to consider the functions and cultural influences of national income statistics.

The paper goes on suggest that it is appropriate to characterize statistics as organisational products. The paper explores this approach with four case studies - statistical systems associated with the Census of Population, the National Food Survey, economic statistics, and with unemployment. These case studies constitute a purposive rather than a random sample. The Census is the largest single survey in Britain, as in most other countries. The National Food Survey is Britain's longest running annual household survey. Economic statistics conform to internationally agreed definitions and, on most measures, account for about half of all British statistics. Statistics of unemployment are distinguished by being the subject of two independently collected series.

It would need a significantly sized research project to provide a comprehensive picture of the operation of these four statistical systems. This paper is selective, touching on some of the more obvious points in order to indicate the importance of the unwitting testimony given by these systems. The paper aims to identify the motivations, assumptions, and functions of these statistical systems. And it discusses some of the evidence these system provide about the role of statistics in society and about the influence of organizational values on society.

The Availability of Statistics

A growth in the range and variety of publicly available statistics in recent decades gives importance to the subject matter of this paper. The availability of statistics about society has increased substantially in recent decades as a result of both production and supply factors. Trend in the production of statistics in Britain is indicated by the number of large national surveys as shown in Chart 1.

In the early 1980s Sir Derek Rayner conducted a major 'value-for-money' review of the Government Statistical Service. The Government accepted Rayner's recommendation that 'information should not be collected primarily for publication (but) primarily because government needs it for its own business' (Great Britain, Privy Council Office, 1981). But, in spite of this policy, there was expansion of survey work later in the 1980s. A series of victim surveys, The British Crime Surveys, was instituted. The Labour Force Survey's coverage was increased dramatically. In the 1990 a new Dietary and Nutritional Survey supplanted some of the functions served by the continuing annual National Food Survey (of which more anon). An annual Health Survey started in 1991. In 1993 a new Family Resources Survey has supplanted some of the functions formerly dependent on the continuing annual Family Expenditure Survey.

Chart 1 also includes selected examples of non-government surveys at the national level. The quality of life surveys carried out in the 1970s have not been repeated. But the so called 'opinion' polls have become more frequent. The British Social Attitudes Surveys became firmly established in the 1980s. After the government got cold feet, the Wellcome Trust funded the National Survey of Sexual Attitudes and Lifestyles (Wellings, 1994). Nearly every important area of activity and attitude in the British Population has now been the focus of a major national survey. The only obvious gap is in religious activity and attitude.

The growth in the production of statistics has been compounded by developments in the new digital technology that has increased the availability of statistics through support of secondary analysis. The microdata from surveys such as those listed in Chart 1 are deposited with the Data Archive at Essex host University. Information technology developments mean that they are more accessible than would otherwise be possible. The Data Archive delivers the microdata in a variety of formats (favouring CD-ROM), but data is also available online with supporting programs for analysis through sources such as the Manchester University Computing Centre.

Another way in which information technology developments are increasing the availability of statistics is through record linking between surveys and administrative data. Chart 2 indicates growth in this area with a selection of some of the more important longitudinal studies conducted by central government. The source of the selection is a recently published Guide which states that 'for confidentiality reasons, not all the data are available on a general basis,' but invites enquiries to named contact points (see GSS(PH) Secretariat, 1996, p. 1).

This Guide shows that what used to be called 'The Longitudinal Study' which links Census data to administrative data which started with the 1971 Census, was actually preceded by the Offenders Index started in 1963 which links administrative data on convicted criminals. Chart 2 identifies three major new longitudinal studies started in the 1970s and three in the 1980s. Already six new longitudinal studies have started in the 1990s.

Does the increased availability of statistics means a focus on the world as represented by the statistics, rather than on the underlying societal problems? This question has been well articulated in the context of statistics used to control the activities of organizations (see Power, The Auditing Explosion, 1994). This paper asks parallel questions about the public role of statistics. To what extent do the categorizations of the statistics structure the social and political debate? Why are these statistics produced? How are statistics generally used?

Official, Social, or Organizational?

Thomas (1984a and 1984b) gives a general and theoretical answer to some of these questions in arguing that the principal function of statistics is to contribute to organizational co-ordination. Hindess (1973) and many later authors favour the term official statistics. But a lot of statistics are produced by non-governmental organizations. Statistics of the media are probably the biggest single category. Statistics produced by non-governmental bodies, where they are published, are usually published only in summary form - see Mort and Siddal (1985). But they do not otherwise appear to be different in character from government statistics.

Emery (1993) uses the phrase social construction of statistics as a subtitle to his study of the production of vital statistics in Ontario. But the terms construction could be take to imply that the subjects of statistics play a significant part in determining the categorizations used. This paper refers to statistics as products rather than constructions because the definitions used are determined by organizations and are usually imposed upon the subjects of statistics irrespective of the meaningfulness of these categorizations to the subject.

The authors Demystifying Social Statistics (Irvine et al. 1979) favoured the term social statistics. Even economic statistics, according to the arguments of Irvine et al., should be seen as social products. But use of the term social seems unnecessarily broad. Administrative statistics, for example, which include vital statistics, most sources of economic statistics, and most educational and health statistics, are by definition, organizational in the sense that they are produced as a by- product of the day-to-day functioning of the organization. The Government Statistical Collective argued that statistics of wealth, unemployment and homelessness, for example, monitor governmental operations concerned with these social conditions rather than the social conditions themselves (Irvine et al. p 138).

The social survey method was pioneered as throwing light on social problems, but even social surveys have become administrative in character. The Rayner Report on the Office of Population Censuses and Surveys emphasized that the work of OPCS (Office of Population Censuses and Surveys) in Britain stems from governmental policies rather than being concerned with social conditions (Witzenfeld and Craig, 1980, paragraph 8.15). Starr (1987) points to the way statistics become 'cognitive commitments' taking their subject matters out of politics and putting it into the nexus of intra-governmental interactions (p 53). Prewitt in the same volume goes further and suggests that the production of statistics transforms social problems into problems of resource allocation (p 272).

The unwitting testimony given by statistical systems is that statistics should be seen as organisational products. It is reasonable to hypothesize that the growth in the production of statistics gives unwitting testimony to a growth of governmental and organizational power. The nature of particular statistical systems gives specific unwitting testimony about role of statistics in society and the influence of the meanings which they create. Consider some examples.

The Census of Population

The term census implies a full count, and it is easy to imagine that the main function of censuses is to estimate the total population. But it would be more accurate to describe the functions of the census as being concerned with the allocation of resources. The function of the earliest censuses was for governments to gather resources - conscripts or tax - systematically (see, for example, Brunt, 1971). Nowadays the typical function is for the distribution of resources. The leaflet entitled 'Why is this information needed?' distributed before the 1991 Census in Britain draws attention to this function. The count 'helps central government work out how much money to grant to your local authority and health authority'.

The leaflet points to the unwitting testimony given by the role of Census statistics in the governmental structure of Britain. For the past fifteen years, for example, it has been taken for granted that the market should play the major role in the allocation of resources. But the use of census statistics illustrates that Britain has a planned economy - as far as the allocation of resources to local government and to health authorities are concerned. The Revenue Support Grant System for local authorities and the Weighted Capitation Formula for health authorities are the counterpart to a centralized system of taxation.

The Revenue Support Grant System, formerly the Rate Support Grant System, uses Census statistics to distribute £18 billions to local authorities in the financial year 1966/7 (see Society of County Treasurers, 1996). The Weighted Capitation Formula, formerly the Regional Allocation Working Party system, uses the Census to distribute £20 billions per year to health authorities (Great Britain, House of Commons, 1995; Peacock and Smith, 1995). The way these allocation systems distribute resources to health authorities and local authorities is rarely subject to political debate, perhaps because it is so statistical in nature. (see Sheldon and Carr-Hill, 1992).

The Census Leaflet also points the resource allocation function below the local authority or health authority level. '...these authorities use Census information when planning services ... to help plan how much housing might be needed in the future . . help government and business to plan jobs and training, helps plan roads and transport, and help local and health authority plan services and facilities for long-term sick and elderly people'.

The Census Leaflet eschews use of the term 'planning' at a national level, but indicates the even after twelve years of Thatcherism the importance of planning at a local level is acknowledged. The Leaflet's use of the term planning in a positive way five times on a single page is surely unusual in British government documents of the 1990s? The unwitting testimony of the Census is that planning is a widely accepted and approved activity at that local level. Even Thatcherism apparently prefers the term 'planning' in this context to, say, 'spending taxpayers money' or 'allocating resources'.

The allocation of resources at a local level are not usually the subject of public debate. There is a contrast between the intensity of debate at the national level on the allocation of resources to, for example, the National Health Service, the relative lack of public debate on how these resources are distributed to hospital boards, and the paucity of debate on the distribution at a local level. The use of census statistics transforms what might have been political issues into matters of administration.

In the 1991 Census there was substantial under-enumeration. The preliminary demographic checks on what have become known as 'the missing million' are described by Marsh, 1993. The Census Validation Survey carried out in June and July 1991 was able to account for only a small proportion of the under- enumeration - see Wiggins (1993) and Heady (1994). For the first time since the Census was started the best estimates of total population have had to depend upon on other kinds of records.

Many of the implications of this situation are discussed by Simpson and Dorling (1994) who identify the main group most severely underrepresented - men in the 20-30 age range and old single people - and a variety of resource allocation problems. If the missing million are not picked up by other statistical systems such as the National Health Service Central Register, does this mean that resource allocations systems will systematically discriminate against the areas where the missing population lives? If so resource allocation systems might reinforce existing inequalities?

The situation also has implications for the academic community. If governmental machinery cannot make an accurate count of the population then they surely give support to others, such as social scientists, to investigate?

The National Food Survey

The National Food Survey, Britain's oldest regularly conducted survey, started as the Wartime Food Survey in 1940 and celebrated its 50th anniversary in 1990. In 1940 there was concern with nutrition standards associated with food rationing. The NFS asks about household purchases of food. In the 1940s it would have been reasonable to assume that household purchases would be closely related to consumption by individuals within the household. More than a half century later that assumption does not apply. The problems are seen as obesity, over-reliance on manufactured foods with a high fat content.

The weakness of the in NFS with regard to dietary matters was recognized by statistical models created in the 1970s that estimated individual consumption on the basis household composition expenditure data (see Cheshire, 1991). But a quite new cycle of surveys of diet and nutrition was started in 1990 that are designed to measure individual consumption directly rather than rely on data on household purchases. The reports already published include a survey of adults conducted in 1990 and a survey of children aged 1 1/2 to 4 1/2 in 1992 (See Gregory et al. 1990). Complementarily a new annual survey designed to monitor trends in the nation's health was started in 1991 (see Bennett et al., 1995 for a report on the 1993 survey).

The broad goals of these new surveys are not very far from the original purposes of the NFS. So why has the NFS proved to be so durable, and why is it continuing? Part of the answer is that the NFS contributes to government economic statistics such as the Retail Price Index and estimates of Consumers Expenditure (see Slater, 1991, p 79). But the main answer to this question seems to be that the NFS has proved to be a very useful source of information for the food manufacturing and distribution industries. The great strength, according to the Marketing Director of Sainsbury Plc, is that the NFS supports monitoring of long term trends that cannot be achieved through any other survey (Hunt, 1991).

The Rayner Review of Statistical Services conducted in 1980 refers to widespread use and purchases of special tabulations by outside users (MAFF, 1981, paragraphs 69 - 70) and proposed that the NFS and the annual Family Expenditure Survey be merged. The unwitting testimony given by the contents of the published reports of the NFS help explain why this merger did not take place. The published tables relate consumption, as measured in terms of household purchases, to social group. The NFS has usually used the social grading classification (A1, A2, B, C, D, E1, E2 and OAP) of the commercial world - not social class of socio-economic groups as used by the Office of Population Censuses and Surveys.

Up to the 1990s there were substantial appendices in the reports of the surveys were devoted to fairly sophisticated estimates of income, price, and cross elasticities of demand. To give an example, it was estimated that the cross elasticity of margarine to butter in 1990 was 0.46. (In other words and increase in the price of butter by, say, 10% would be likely to increase quantity of margarine bought by 4.6%). After 1990 these appendices disappeared from the published reports because the government found that it could make money by selling the information to commercial users.

The NFS may have contributed substantially to the efficiency and growth of the food manufacturing and distribution industries in Britain. But a healthy food industry does not necessarily mean healthy food. The products of the food industry when translated into diets are producing a population with an increasing tendency to obesity. It would not be unreasonable to put forward the hypothesis that the NFS has actually contributed to the health problems that the health surveys started to address in 1990.

The implications for social research are about sources of information. It would be naive of researchers to think of the published volumes of the NFS as crucial sources for matters concerned with nutrition and health, or demand for food. If the researcher is seriously interested in nutrition and health, the surveys by Gregory et al.(1990) and Bennett et al. (1995), cited above are the prime source. If the researcher is seriously interested in demand for food they would be advised to try to get access to the many unpublished studies conducted on the basis of NFS data, by organizations and firms in the food and agriculture industries or by MAFF on behalf of such organizations.

The NFS raises questions about government statistics policy. The statistical relationship between MAFF and the food industry has been beneficial to both sides. And it might reasonably be asked whether the similar benefits might accrue from statistical relationships with other industries. The regular estimation of income, price, and cross elasticities for the use of the public and private transport, would, for example, make a useful input to public policy making as well as provide information useful to the transport industries.

Economic Statistics

The origin of national income statistics goes back to the 1920s when a conference was held by the League of Nations to bring about comparability between countries. But by the 1940s Keynesian economic theory had become well established. The motivation for the production of economic statistics changed from comparability to being accounts that would serve economic management.

The publication of the seven hundred page length System of National Accounts 1993 by EUROSTAT as a joint product of the European Community, IMF, OECD, the United Nations and the World Bank is an indication of the dominant cultural influence of economic statistics. It is relevant to point out that that the revenue of these international organizations depends upon the use of the economic statistics as a comparative measure to determine the allocation of national contributions. But the categorizations used in economic statistics are more closely geared to the needs of economic management than to international comparability.

The unwitting testimony given by economic statistics is the emphasis given to the needs of economic management. Economic management requires that the categorizations used and the data production processes are geared to measuring changes over a month-to-month, quarter-to-quarter, or year-to-year basis. Economic statistics do not, for example, attempt to measure human capital. The principal reason is that the monetary returns on investment in human capital extend over a longer period than is relevant to management of the economy according to Keynesian principles. For much the same reason most monetary costs directly associated with education are not treated as capital investment but are classified to public or private expenditure. From any viewpoint, except that of the short term management of the economy, these are misclassifications. Other activities contributing to the formation of human capital, such as the time students or trainees spend studying, are not counted at all.

On the other side economic statistics perversely include activities with negative externalities such as congestion and pollution. Economic statistics can validly be used to measure income changes on a year-to-year basis, but it is not valid to use economic statistics as indicators of standard of living over longer periods - although this is commonly done.

In spite of the seriousness of these limitations it is difficult to escape from the cultural dominance of economic statistics. It is, for example, difficult to think of growth separately from the way it measured by economic statistics. There is no alternative. Hence many of what are labeled as attacks on economic growth may be more appropriately regarded as exposés of the limited validity of economic statistics.

In general economic statistics define the formal economy of paid employment, and other contractual arrangements that are known to governmental authorities. All other activities can be regarded as belonging to the informal economy. The formal and informal economies are closely interdependent. Domestic support for paid employees, whether carried out by spouses or by employees themselves, contribute to productivity in workplace where the employees. The labour force available to the formal economy is created in the informal economy. One of the major functions of the formal economy is to produce income for pensioners and others not in paid employment.

A large part of political debate is concerned with the status that should be accorded to the informal economy. Discussion of the level of taxation, for example, cannot and should not be separated from the contribution made by the informal to the formal economy. Within this area the taxation and financial benefits accorded to parent symbolizes the dilemma about the creation of human capital. To what extent should the creation of human capital be left as a wholly parental responsibility and to what extent should it be supported by transfer payments from the formal economy?

The contribution that social science can make to such debates is limited by the asymmetry in the information available. That related to the formal economy is rich to the point that it can be directly related to the salary of social science researcher. Information on the informal economy is fragmented and nugatory in comparison.


The cultural influence of economic statistics also dominates statistics of employment and unemployment. Organizational statistics generally recognise only paid employment and so exclude those without monetary contracts. Students, voluntary workers, housewives, househusbands, disabled, carers, the retired, and those having independent means of income support are all excluded. But the statistics also treat students, voluntary workers, housewives, etc. as occupied, not able to be in paid employment, and therefore not able to be unemployed.

In order to be classified as unemployed and individual has to qualify according to organizational definitions. But the two sets of definitions used in Britain - those of the Count of Claimant and those of the International Labour Office used by the Labour Force Survey - are different, they cover very different populations, and the populations covered are diverging. Chart 3 illustrates the divergence over 1984-95 in terms of Indexes of Similarity that compare what the size of the population the two series have in common with the size of the population excluded from one or other of the series. Chart 3 shows that overall the index fell from 58% in 1984 to 45% in 1995, that the similarity is much less for women than for men, and that the indexes for both men and women have fallen over the period.

It would be possible to explain the differences in coverage and the growing divergence in terms of the technical detail of the definitions used. But the differences in detail stem from the different functions of the two series. The Count of Claimants has a long history as a governmental series, and has predecessors going back to the 19th Century in the form of statistics produced by trade unions who at that time paid unemployment pay to their member (see Garside, 1980). The crucial categorization made by the Count of Claimants is entitlement or non-entitlement to unemployment benefit. The principal function of the LFS/ILO Series is to achieve comparability with other countries. The crucial categorization made by the LFS series is between economic activity and economic inactivity.

The Count of Claimants is necessary to government for control of unemployment benefit expenditure. The LFS/ILO Series is necessary to underpin regional policy in the European Community. Neither of these series aims to measure the unused supply of labour. The two series are diverging in terms of population partly because the definition of entitlement to unemployment pay has been substantially tightened over the past decade. With the tightening of the definition, the smaller the population covered by the Count of Claimants.

The two series are also diverging because of changes in the labour market. The growth of part time working, the growth in the number of non-employed men, and the growth in the number of women entering the labour force have all increased the size of the population capable of economic activity. With the growth in the size of population that is capable of economic activity, the greater the population included in the LFS Unemployment Series that is not included in the Count of Claimants.

The Report of the Working Party of the Royal Statistical Society recommended that the LFS unemployment series should be used instead of the Count of Claimants (Working Party on the Measurement of Unemployment in the UK, 1995; Steele, 1996). But though both series can be considered as adequate in relation to their functions, neither series is by itself adequate in relation to topics such as the potential size of the labour force nor of the reserve army of labour.

The Labour Force Survey of Winter 1995/6, for example, found that 4.6 million people said that they would like paid employment. But only half of these are counted as unemployed by the LFS Unemployment Series. The other half do not satisfy the availability or motivation criteria of the ILO definition of unemployment. (Labour Force Survey Quarterly Bulletin, June 1996, p. 6). The growth in the non-employed population capable of economic activity throws doubt on the validity of making a statistical dividing line between the economically active and economically inactive populations. It seems likely that a growing number of people move directly between employment and 'economic inactivity' and vice versa. If these 'go-getters' never confess to economic activity while not in employment, i.e. they do not say that they are looking for and available for work, they are never counted in the unemployment statistics.

It has been suggested that a combination of the Count of Claimants and the LFS Series into a single integrated series would give a more comprehensive picture than either series alone (See Thomas, forthcoming). But it is doubtful whether even such an Integrated Series would give an adequate picture of the flexibility that now exists in the labour market in Britain. It would be possible to investigate that flexibility on the basis of Labour Force Survey data. But both the LFS Unemployment Series and the Count of Claimants disguise that flexibility.

Unmet Statistical Needs and the Informal Economy

This discussion of the assumptions, motivations, and functions associated with statistical systems is not aimed at disparagement of these systems. These series are essential to government and to other organizations. The point of this discussion is identify the nature of these statistical systems and to use this unwitting testimony to explore what these system tell us about society and about the influence of these systems on society.

Each of the systems discussed provides evidence of this kind. The Census tells us important things about the degree of centralization and planning in the British economy. The National Food Survey illustrates the value of government statistics to private industry. Economic statistics and statistics of unemployment illustrate the importance given to the needs of economic management.

Economic statistics and statistics of unemployment also indicate the influence of organizational meanings on society and the ways in which these statistical systems shape our thoughts and perceptions. The message for the social sciences is not that these statistics should be ignored because of the distorted picture they sometimes give, but than these distortions should be investigated and exposed.

The idea of economic growth, for example, has been savaged from a number of different perspectives. Mishan in The Costs of Economic Growth pointed to negative externalities - the costs imposed by economic activity on members of society. Douthwaite in The Growth Illusion makes a sustained and wholesale attack on economic growth, accepting Mishan's diagnoses, and adds other arguments, such as that growth causes unemployment, and effects on family life that do not clearly fit under the negative externality heading. Both Mishan's and Douthwaite's arguments on the manifestations of economic growth are inseparable from the way economic growth is measured. Externalities are the product of statistical boundary between economic activity and other activities. Douthwaite's arguments about unemployment and the impact of growth on family life are a product of the way in which economic statistics define a boundary between the formal and informal economies.

Jacobs in The Green Economy is more circumspect about statistics. He argues that quality of life should be measured in terms of personal income plus quality of life factors not measured by personal incomes - including benefits from public expenditure and factors like crime levels and quality of the physical environment. Jacobs argues that standard of living is subjective and that a programme of environmental would require reduction of real disposable incomes in order maintain standards of living,

The existing statistical systems are the problem not the solution. Social science cannot deal with the arguments of Mishan, Douthwaite, or the suggestions of Jacobs, which involve both the formal and informal economy on the basis of the imbalance in the statistical information available. There needs to be investigation that would produce statistics to counterbalance the influence of economic statistics and that of other organizational categorizations such as the Count of Claimants and the LFS/ILO unemployment measures.

One such example is the Quality of Life surveys noted in Chart 1. If economic statistics are used as an indicator of quality of life then it is appropriate to check on their validity for that purpose. A QoL statistical series could provide such an independent assessment.

There is also a need for information on work done in the informal economy and the extent of dependence of the formal economy on the inputs from the informal economy. The under-utilized Labour Force Survey could provide much of the information needed. One direction could be to explore the complexities which would be involved in investigation of labour resources which are under-used in the formal economy. Another direction could be to investigate the scale, the nature, and the value of work in the informal economy such as parenting, caring, and simply 'looking after family/home'. It is pleasing to note that the Economic and Social Research Council has under consideration a Programme on informal economic activity. Such a Programme might help redress the imbalance in the information available on the formal and informal economies.


Statistics as Organizational Products is a revised version of a paper of the same title originally given to the Fourth International Conference on Social Science Methodology, University of Essex, England, July 1996.


ABERCROMBIE, Nicholas (editor) (1994) Dictionary of Sociology. London: Penguin (Third Edition).

ALONSO, William and Paul STARR (1986) The Politics of Numbers, New York: Russel Sage.

BENNET, Nikki, et al (1995) Health Survey for England 1993, London: HMSO, pp. 518.

BRUNT, P. A. (1971) Italian Manpower 225 BC - AD 14. Clarendon Press

BULMER, Martin (1980) 'Why don't sociologists make more use of official statistics', Sociology, vol. 14, pp 505 - 523.

CHESHIRE, Andrew (1991) 'Household Composition and Household Food Purchases' in J. M. SLATER (editor) Fifty years of the National Food Survey. London: HMSO.

DALE, Angela, et al. (1988) Doing Secondary Analysis. London: Unwin Hyman.

DALE, Angela and Cathie MARSH (1993) The 1991 Census User's Guide, London: HMSO

DEPARTMENT OF EMPLOYMENT GROUP (1996) Labour Force Survey Quarterly Bulletin, No 12, June 1996.

DOUTHWAITE, Richard (1992) The Growth Illusion. Devon: Green Books.

EMERY, George (1993) Facts of Life: The Social Construction of Vital Statistics, Ontario 1869-1952. Montreal: McGill-Queen.

EUROSTAT (1993) System of National Accounts, Brussels: ECE, pp. 711.

FULLER, M. F. and Lury, D. A. (1977) Statistics Workbook for Social Science Students. Oxford: Philip Allan.

GARSIDE, W. R. (1980) The measurement of unemployment. Oxford: Blackwell.

GREAT BRITAIN, HOUSE OF COMMONS (1995), Public expenditure on health and personal social services, HC21. London: HMSO.

GREAT BRITAIN, PRIVY COUNCIL OFFICE (1981) Government Statistical Services. (Report of the Rayner Review), Cmnd. 8236. London: HMSO.

GSS(SPH) Secretariat (1996) Longitudinal social statistics; A guide to official Sources. London: Office for National Statistics.

GREGORY, Janet, et al (1990) The Dietary and Nutrition Survey of British Adults. London: HMSO.

HEADY, P. (1994) 1991 Census Validation Survey. London: HMSO.

HINDESS, Barry (1973) The use of official statistics in sociology. London: Macmillan.

HUNT, Ivor (1991) 'A View from the Food Industry' in J. M. SLATER (editor) Fifty years of the National Food Survey 1940-1990. London: HMSO.

IRVINE, John, et al. (1979) Demystifying Social Statistics. Pluto Press, London.

JACOBS, Herbert (1984) Using Published Data: Errors and Remedies. Newbury Park and London: Sage.

JACOBS, Michael (1991) The Green Economy. London: Pluto Press.

MAFF, (1980) Review of Statistical Services, Rayner Review. London: Ministry of Agriculture, Fisheries and Food.

MARSH, Catherine (1993) 'The Validation of Census Data' in A. Dale and C. Marsh (Editors) The 1991 Census Users Guide. London: HMSO.

MARWICK, Arthur (1980) The Nature of History (Third Edition). London: Macmillan.

MISHAN, E. J. (1967) The Costs of Economic Growth. London: Staples Press.

MORT, David and Siddal, Leona (1985) Sources of Unofficial UK Statistics. London: Gower.

PEACOCK, Stuart and SMITH, Peter (1995) The Resource Allocation Consequences of the New NHS Needs Formula, Discussion Paper 134. York: University of York, Centre for Health Economics.

Prewitt, Kenneth (1986) 'Public Statistics and Democratic Politics' in ALONSO, William and Paul STARR (1986) The Politics of Numbers, Russel Sage. New York.

POWER, Michael (1994) The Auditing Explosion. London: Demos.

REID, Stuart (1987) Working with statistics. Oxford: Polity Press.

SHELDON, T. A. and CARR-HILL, Roy (1992), 'Resource allocation by regression in the National Health Service - a critique of the Resource Allocation Working Party Review' JRSS, vol. 45, part 3, pp. 403 - 420.

SIMPSON, S and D DORLING (1994) 'Those Missing Millions: Implications for Social Statistics of Non-Response to the 1991 Census', Journal of Social Policy, vol. 23, no. 4, pp. 543 - 567.

SLATER, J. M. (Editor) (1991) Fifty years of the National Food Survey 1940-1990. London: HMSO.

SLATTERY, Martin (1986) Official Statistics. London: Tavistock.

SOCIETY OF COUNTY TREASURES (Annual) Standing Spending Indicators. Chester: Cheshire County Council.

STARR, Paul (1987) 'The Sociology of Official Statistics' in William Alonso and Paul Starr (1986) The Politics of Numbers. New York: Russel Sage.

STEELE, David (1996) Options for Producing Monthly Estimates of Unemployment According to the ILO Definition. London: Central Statistical Office.

STEWART, David W. and KAMINS, Michael A. (1993) Secondary Research: Information sources and methods. London: Sage.

THOMAS, Ray (1984a) 'Why Have Government Statistics? And How to Cut Their Cost', Journal of Public Policy, vol. 4, part 2, pp. 85 - 102.

THOMAS, Ray (1984b) 'Statistical Information Systems and Management' Organization Studies, vol. 5 no. 4, pp. 345 - 358.

THOMAS, Ray (forthcoming) 'Integrating measures of unemployment', (to be submitted to the Journal of the Royal Statistical Society).

WELLINGS, Kaye (1994) Sexual behaviour in Britain - the National Survey of Sexual Attitudes and Lifestyles. London: Penguin Books.

WIGGINS, Richard D. (1993) 'The Validation of Census Data - Post Enumeration Survey Approaches' in A. Dale and C. Marsh (Editors) The 1991 Census Users Guide. London: HMSO.

WITZENFELD, S and CRAIG, J. (1980) The (Rayner) Review of the GSS: Intitial Study of the Office of Population Censuses and Surveys. London: Central Statistical Office.

WORKING PARTY ON THE MEASUREMENT OF UNEMPLOYMENT IN THE UK (1995) 'The Measurement of Unemployment in the UK', Journal of the Royal Statistical Society, vol. 158, part 3, pp. 363 - 418.

Copyright Sociological Research Online, 1996