Statistics are more commonly linked to lies than to spin. Lying with statistics can usually be dealt with. Contrary evidence can usually be found. But the term spin captures problems with statistics that are more widespread than lies − and are more difficult to deal with.

Spin in government statistics is insidious because it less noticeable than lies. Statistics gives us a picture of the world. Statistics are selective, but not obviously so. The categorisations used to produce the statistics are not usually clear and they are not usually specified in any detail. Usually there are no alternative sets of statistics that might give a different picture.

Statistics relating to work, for example, relate to an entity that is far from any generally accepted cultural meaning of the word work. Statistics relating to employment and unemployment are based on categorisations that in many of their features are significantly different from everyday perceptions of those regarded as in employment or unemployed. They may also be very different from the perceptions of the subjects, i.e. those who are actually counted as in employment or unemployed.

Statistics relating to work and employment generally cover only paid work and paid employment. Unpaid domestic work, caring, parenting, or any kind of voluntary work is not covered. This particular bit of spin emphasises the importance of the formal economy over the informal economy of family and cooperative activities.

The emphasis on the formal economy is purposeful. Official statistics focus on the formal economy because the formal economy produces the taxes that support the activity of government. That connection reinforces the point made by sociologists many decades ago. Government surveys and other official statistics do not necessarily tell us about society, they tell us about the governmental view of society.

Official statistics support international as well as national government organizations. The massive tome System of National Accounts (SNA) is published jointly by the United Nations, the World Bank, the International Monetary Fund, OECD, and the European Community. The SNA is the bible for economic statistics everywhere and is based upon on the statistical definition of work generally limited to paid work. These five organizations and many other international organizations have a vital interest in the SNA because it describes the basis of the financial contributions from their member states.

The statistical emphasis on paid work creates difficulties for governments in many policy areas. A good example is dealing with the costs of parenting. It is difficult to deny that bringing up children is real investment and real work. But caring for children does not count in the statistics as work. And bringing up or educating children is not counted as investment in economic statistics.

The outcome is a kind of governmental schizophrenia. There are government grants to single parents not just as a matter of social justice but also because it is recognised that parenting is real work and that parenting is a real investment in the future of society − even if this is not acknowledged by economic statistics.

Labour market statistics give a different picture. Why should this group be supported with grants that discourage them from making any contribution to the economy? So there is also pressure on single parents to enter employment that reduces the time they can spend parenting. The use of statistics that give such an unbalanced picture cannot make budget-time decision-making easier.

Statistics of employment have an air or precision about them − that is mostly justified. But the expression of that precision is often very different from that assumed by most users of the statistics. According to the standard International Labour Office (ILO) criteria, for example, anyone who is paid for working an hour or more in a week is counted as in employment.

ILO criteria involve the use of a labour force survey (LFS). The output includes statistics for the numbers in paid employment by average hours worked. So it is possible to get statistics for the numbers in employment who work, say, five hours a week or less. But such statistics give us no idea as to whether those who work less than five hours a week regard themselves as in employment, as unemployed, or simply as having a spare time job, a newspaper round, or a paying hobby.

The categorisations used for the production of statistics of unemployment are also taken out of the hands of the subjects of the statistics. If LFS respondents say they worked an hour or more in the previous week they are automatically classified as in employment. Respondents other than those categorised as in employment are asked if they would like to be in employment. If the answer is positive respondents are asked what steps they have taken to seek work in the previous four weeks. But at no stage in the standard LFS are respondents asked directly if they believe they are unemployed,

ILO unemployment statistics for unemployment are unusual in statistics not being based on any clear definition. Even if respondents have taken what are considered adequate steps to look for work they can be excluded from the statistics because they fail to meet availability criteria. Also excluded from the statistics are those who would like to work but are not seeking work because they do not believe that jobs are available; this group is categorised in the statistics as ‘discouraged workers’.

ILO criteria express international agreement on the way unemployment should be measured. But this disguises the awkward fact that there is there is no international agreement on the meaning of unemployment. Usually there is a concept underlying the production of statistics, but if we ask any statistical office about the concept underlying statistics of unemployment there is no real answer. The only answer possible is to say it is measured according to the procedural criteria of the ILO. The failure to explicitly specify what is being measured leaves the statistics open to ideological interpretation. (See, for example, the discussion in ‘Is the ILO definition of unemployment a capitalist conspiracy?’ In Radical Statistics, 88, 2005 at http://www.radstats.org.uk/no088/Thomas88.pdf.)

In the case of unemployment there are alternative series available for most European countries. Most of the alternative series are based on administrative statistics covering entitlement to unemployment benefits. Such administrative statistics are available in great detail but suffer from the disadvantage that they depend upon government rules governing entitlement to benefit. It seems likely that fall in claimant unemployment since 1997 in the UK, for example, is mostly attributable to tightening up of rules affecting entitlement to benefit.

In the 1980s changes in the rules affecting entitlement to unemployment benefit led to charges that the Thatcher government were fiddling the statistics. That controversy led to the inclusion of a pledge to create an independent statistical service in the Labour Party’s 1997 election manifesto. But eight years later it is not clear however that any progress has been made.

In November 2005 Gordon Brown announced that he would introduce legislation to support the creation of statistical service at “an arm’s length” from government. But it is not clear what this means in terms of independence. Nor is it clear how legislation might reduce government control exercised through the civil service code and the purse-strings of budgetary allowances.

Is Gordon Brown’s announcement just another bit of spin? Is it possible to have a statistical service independent of the Government of the day?

I expect to discuss such questions in a further posting.