There is a temptation to use official statistics without due consideration to the assumptions and concepts employed in the construction of the ‘social facts’. Clearly, researchers need to understand how such data is constructed and for what purpose?

Government agencies are responsible for the design and collection of information covering such topics as the economy, crime, education, health and the environment. Sometimes compiled with the assistance of non-government organisations, government information agencies aim to provide a factual overview of government policy and developments. Geographical region bodies, such as CIS and Central Asia Economic Cooperation, produce data covering a number of areas such as labour migration, capital flows, intra-regional trade. Their tasks are to provide statistical information for devising, managing and assessing common policies (e.g., single trade policy), to supply the general public with statistical information (e.g., taxes, unemployment, etc.), and to offer technical cooperation with the rest of the world.

As well, there are one-off surveys relating to a specific topic that is of current policy interest (e.g., agriculture and mining, women and safety). Often such data charts trends within society, and informs policy-makers about their decisions as well as means to forecast and evaluate the impact of new social policy.

In addition, marketing agencies collect information on the tastes, habits and opinions of the population for commercial reasons.

We shall discuss how statistics are not mere facts, but socially constructed, reflecting deep social relations, assumptions, norms and conventions.

Statistics provide policy makers and researchers with indications of the types of incident, act and practice (say, crime or health) being undertaken, and the extent to which the different practices are increasing and decreasing according to the implementation and impact of government policies (say, criminal justice or medical policies).

To have confidence in using official statistics, we must be sure that they fulfil the criteria of validity (produce true knowledge) and reliability (be repeatable). To achieve these, the following conditions must hold. First, to categorise a similar incident or act in the same manner, so that there is little room for discretion in how to record such information (e.g., all hospitals must agree on how to categorise emergency and non-emergency operations). Second, our statistics must be mutually exclusive so that two different acts cannot be categorised in the same way (e.g., surface and deep surgery). Third, the classificatory system must exhaustively categorise all incidents so that there are no acts left outside the classificatory system, otherwise it is incomplete and unsatisfactory for recording all the information.

Two important points to note when defining an act in a particular way (e.g., criminal or medical): definition and detection. For instance, for an act to be ‘medical’ (distinguishing it from social or psychological), the medical community must define it as such. In addition, someone must be have identified and detected it. Otherwise, the compilation of the medical statistics will be inaccurate.

Yet, it must be noted that definitions are problematic, since they vary across societies, culture and history. For instance, is mental fatigue and tiredness of limbs experienced by office workers a medical condition or psychological problem? Until recently, the medical community had considered that it was all in the mind, and still the community is deeply split about its causes. In other words, a definition is a diachronic, not synchronic and static concept.

Also, the issue of detection is problematic. The decision to report an incident or act will depend on a number factors, including a sense of obligation by the citizen, the perceived seriousness of the incident, its worthiness, potential benefits and harms from reporting it, and institutional practices. For example, women who are victims of domestic violence perpetrated by their partner may be reluctant to report it to the police. Studies have shown that women tend to conceal such experiences from the police for fear of repercussions (physical, emotion and material). It also depends on police practices and their willingness to see it as a legitimate part of their normal duties (e.g., some police officers may regard domestic violence as a ‘private’ and ‘family dispute’, and therefore trivial in comparison to other types of crimes). In addition, the routinisation of harassment and assault may lead women not to report the incident in the first place. It seems, therefore, that there is a gender dimension to the construction of statistics (viz., the difficulty of policing the private sphere of the home results in an under-representation of domestic violence in official crime statistics).

Moreover, there is a class dimension to the construction of statistics. Certain types of incidents and acts may be prioritised and targeted over others, or more visible than others, and therefore over-/under-represented in statistics. For example, middle-class, white-collar crimes (e.g., fraud and tax evasion) are less likely to detected and reported than working-class, blue-collar crimes such as mugging and burglary. This may reflect an in-built class bias and prejudice in the police system.

Furthermore, there is an ethnic dimension. Certain types of incidents and acts may be given legitimacy and recognition than others. For example, hospitals are more likely to treat white (or American or Slavic) people than black (or Chinese or Kyrgyz) people, so that the number of black patients may be under-represented in the medical statistics.

Thus, matters of detection, definition and institutional (e.g., police and medical) practices affect the production of statistics. The compilation of official statistics depends on a set of discretionary procedures (e.g., actor’s willingness to report an incident, and institutional readiness to record the incident and to take it seriously); institutional practices. Here, officials’ (such as police officers and medical staff) decision to identify, recognise and record incidents and acts depends on a wide range of circumstances (such as officials’ sympathy, and the organisational culture - organisational policies instructing officials how to behave). Clearly, official statistics do not ‘speak for themselves’, but rather tell us more about the organisational practices and power relations within society, revealing the depth of societal class-discrimination, sexism and racism.

The key elements in the process of compiling official statistics are interpretation, discretion and differential application and enforcement. Between the occurrence of the incident or act and being recorded as a statistic, there lies definition and detection issues, and individual and organisational practices of interpretation, application and enforcement. For example, there is room for interpreting a death as a murder or manslaughter; or an illness as medically related or induced by poor eating and exercise. Furthermore, the process of reporting, detecting and processing incidents (such as crime or health) is not socially neutral, since discrimination exists along class, gender and ethnicity. However, despite this, such discrimination can be overcome by professionalism, independent review bodies, etc. This produces an outcome that is variable and uncertain.

There are three broad schools of thought on the perspectives on official statistics.

• The realist school (informed by positivism and empiricism) is characterised as taking official statistics to be objective indicators of the phenomena to which they refer.
• The institutional school (informed by idealism) considers official statistics as neither valid nor reliable indicators of objective phenomena. Official statistics tell us more about an organisation’s behaviour and practices, or the discretionary actions of individuals, rather than the phenomena itself.
• The radical school (informed by critical theory) locates institutional and discretionary practices within a wider theory of the dynamics and structure of society. For instance, the government compiles social statistics on the crime, health and income of the nation that can facilitate the order and regulation of the population. Here, statistics are also analysed in terms of their effects as products, as well as in terms of being constructions, because the definitions used by organisations are usually imposed upon the subjects of statistics irrespective of the meaningfulness of these categorisations to the subject. What are the effects of this production and construction of official statistics? For instance, initially many governments had constructed AIDS as something affecting only the homosexual community, and the official statistics had produced the idea that AIDS was a gay disease. This led to health programmes targeting only the gay community. However, AIDS was spreading to heterosexual community and the drug-users. The simple picture was hopelessly wrong, and a product of homophobia , rather than careful considerations of the theoretical debates and evidence.

While some social scientists regard official statistics in need of demolition and demystification, there are those who employ such data for the purposes of exposing the ways in which they reflect power relations. Here, the emphasis is on the process of constructing statistics, rather than as a product.

While statistics seem to reflect the assumptions, conceptions and priorities of the state, corporate interests, and the social order, they are still useful for research because they produce interesting and descriptive findings of the world.