Radical
Statistics
Making Gender Statistics empowering
Irene Bruegel
Most feminists in the United Kingdom both inside and outside academic circles are wary of, if not downright hostile to, statistical analysis. On the other hand, there are some who might be regarded as a having a touching faith in the power of statistical categories. In between are those who might be called the 'femostats': those professionally involved in the generation, analysis and dissemination of gender statistics, not as any old task, but as a political act of some kind. I was asked to speak at the conference on the basis of my involvement over the last two plus years with the Gender Statistics User Group (GSUG). The group is the youngest offshoot of the Statistics Users Council, but it can also be seen as the latest arena in which femostats have come together in Britain. I want to stress that I am writing personally and though I shall say something about the work of the group, I want to air broader issues about what such a group is, and what it might be.
Broadly speaking Gender Statistics enable us to track forms of inequality between men and women; as such we should all be Gender Statistics Users. Taking up the mainstreaming agenda, the GSUG's concern is with statistics across the range, including those on transport and communication where there is little tradition of gendered analysis, but where gender disaggregation is relatively straightforward. More challenging still are themes like economic power/economic relationships/income distributions (as distinct from the labour market) where there are real conceptual questions in deriving meaningful gender disaggregation. This may not be achieved with a simple contrast between men and women, and some contrasts hide rather than reveal differences. The most obvious being in making comparisons between male and female heads of household.
Having declared that we are all putative Gender Statistics Users, our second aim has been to take the concept of Gender Statistics User way beyond the realm of us femostats, i.e. 'normal' statistics users located in odd corners of Universities, research organisations and government who make use of gender disaggregation in their daily work. As can be expected, even the group of femostats ranges between those whose stance is first and foremost distanced and professional and those who are nearer the barefoot end of the spectrum. Beyond that we want to include feminist campaigning organisations and trade unions, indeed almost anyone involved in jockeying for resources and in pressurising for policy change in the community at large. This may indeed include those seeking resources and policy changes of benefit to certain groups of men. Men's health issues for example should generate demand for new types of gendered health statistics.
I start with a general discussion of the politics of gender statistics in Britain today, illustrating this with the question of who are and who should be/might become a Gender Statistics User. I then take the discussion beyond the concept of national statistics and national policy to draw on the work I am doing on regional development, local regeneration and community capacity building. This takes up the point that Perrons (1999) makes in her chapter on gender statistics. While the House of Commons, the Ministries and Whitehall can be seen to have opened up to women to a degree, power can also be seen to have escaped elsewhere. Upwards of course to global institutions, mainly private, where women are decidedly few and far between, but also outwards to quangos, agencies, and partnerships dominated by private capital. To my mind this means we have to look much more at non-official statistics, specifically to the way data are collected and analysed by the new institutions of governance. Radical Statistics as well as Radical Gender Statistics has to take on board the debates on monitoring statistics and performance measurement. I suggest that we can use spatial analysis of differences in performance as a lever for change. As I discuss, the issue of gender statistics is no longer simply about differences between men and women, if it ever was, but has to take on board differences among women.
The politics of gender statistics
The feminist critique of official statistics has followed the general pattern of feminist social analysis. It began, naturally, by challenging the invisibility of women, and did this in ever more sophisticated ways. Gender inequality was shown to be concealed variously by:
- the bias towards the collection of statistics on men (or on categories, like full time manufacturing worker, inhabited largely by men);
- by failures to disaggregate data by gender;
- through the classification of data according to schema that related poorly to women's experience; and
- the concentration of data collection on economic relationships and certain types of social problems to the neglect of others.
For the most part feminist social scientists viewed statistical categories - most particularly 'head of household' - as exemplars of the masculine bias of social science methodology and very few were drawn into the practical politics of seeking to improve the statistics. In doing so the Gender Statistics User Group nevertheless follows the path of some established tradition: an Equal Opportunities Commission (EOC) working party in the early 1980s chaired by Audry Hunt from the Government Social Survey and work undertaken by women's units in local government, specifically the Greater London Council (GLC), as well as by single issue campaigns on rape, domestic violence and child poverty. Many of the improvements, most especially in occupational categorisations and in the presentation of unemployment data have not, however, come from specifically feminist campaigns but wider debates about the efficacy of existing practice and from international pressures.
The collection of official statistics both reflects and confers status and so it is hardly surprising that the increasing international significance of gender politics has brought in its wake an enormous improvement in the gender statistics available on a regular basis. What is less clear is how far the availability of statistics has fostered political action. Certainly the lack of measurement of domestic violence, the limitations of the data on rape and sexual harassment limited political campaigning, but statistics have never really been central to campaigns on gender politics, except in relation to employment. It is individual cases and individual stories that have galvanised action, however unrepresentative these cases might have been. Even much of the publicity given to the female forfeit - life-time earnings deficit as a result of sex segregation and discrimination - was couched in terms of Mrs Middle Qualifications, Mrs No qualifications, and the like.
The power of specific cases is not of course confined to gender politics, but feminists have made much of the contrast between subjective understanding, individual experience and empathy as distinct from supposedly objective data. Thus campaigns won through the deployment of statistical methods can be thought to play into the hands of the enemy, or at least to be taken out of women's hands. So feminist radical statisticians more than radical statisticians in general face battles on two fronts. As radical statisticians we have always sought to deconstruct that supposedly objective power of measurement, but we would not be in it if we did not also see statistical analysis as a valid medium of discourse, and see it as our task to open that medium up to others. Radical feminists, as against radical statisticians, have a very different agenda. They do well to warn us against our, possibly congenital, tendency to believe that policy development is rational. But, against them, I think engagement with the formal procedures of rational policy analysis can contribute to worthwhile change, for wider groups of women, not just the femostats.
Two early clashes in the short history of the Gender Statistics User Group illustrate the gulf. The first was predictable and not of great interest. The initial conference was designed to bring together statistics producers within the ONS (Office for National Statistics) and the wider GSS and campaigning groups such as Gender Statistics Users. The producers felt that one campaigning group in particular was misinformed and uninterested in the realities of data collection, while the group felt that their criticisms based on experience with women were given shorts shrift.
The second issue is more challenging. One group of campaigners put great store by getting a question on unpaid household work included in the Census. The femostats tried to suggest that the Census was not necessarily the best means of collecting such data, but that was not it seemed the point. The medium/campaign was the message; it became clear that the group saw the very inclusion of such a question in the Census as in itself furthering women's interests. The issue was not the provision of data for analysis, but the right of those without employment to declare them as working, and not be forced to tick a box marked 'economically inactive'. The Census question was of symbolic importance, pure and simple.
The compromise we made was of little interest to the campaigners, but I would like to air it here. This is to campaign against the use of the term economically inactive to refer to those engaged full-time on housework and/or family care in any discourse: in the publication of tables, the labelling of variables, the publication of labour market analyses and academic papers and in the media more generally. The issue would no longer be the wording on the Census form, but the erroneous conflation of the term 'economic' with 'employment' (and implicitly) with production. There are ranges of activities that have a bearing on wealth creation beyond employment: saving, investment, housework, and indeed consumption itself.
The point I am trying to make is, however, a different one, namely that the issue was not the data but the nomenclature. The campaigners identified the Census as a terrain of political battle. Given the high media profile of the Census, an appropriately worded question was seen as much more effective in furthering the recognition of unpaid housework as work and, therefore, of revealing elements of women's oppression than any analysis of data in reports that gather dust on the shelves. Though I started this article by suggesting that their stance was naive: changing questions alters very little in reality we femostats could equally be revealed as naive. Pursuing good data may well be far less effective than using the Census to make a splash in the media. Now that New Labour is tending to identify good research (and research that merits funding) as research that produces the 'right' results, I think I now have more sympathy with the confrontational/sceptical stance of radical feminism.
I certainly argue that femostats need to take on issues of interpretation with more vigour, and adopt a more sceptical stance than simply identifying 'gaps' to be filled. As measures of gender relations, of the distribution of power between men and women, statistical measures are bound to be limited, to be reductive, at best to be broad indicators. Statistics measure the outcome of decisions, not the whys and wherefores of the decisions, nor room for manoeuvre available. Even statistics on domestic violence may tell us more about the opportunity women have to voice their complaints than the power men have over them. Statistics on male violence can indeed be interpreted to reveal the inability of some men to get their way in sophisticated as against crude and violent ways. Certainly there is a case for seeing them as revealing male inadequacy as much as male power. I am not denying the abuse in domestic violence, but taking an extreme example to illustrate a more general point.
A key issue for gender statistics has always been to open up the household, household dynamics and relationships to political and policy analysis, to break with the division between statistics collected at an individual level and the household level. There has been considerable progress in measuring time-use, as one part of the prising open of household relations. We have also seen the development of a new series on women's individual incomes. The Gender Statistics User Group has contributed to the development of these series by commentating on these new statistics and lobbying for appropriate resources. That said, as with measures of labour market participation and indeed measures of women's presence on decision-making bodies, feminists are bound to point to the limited value of any one of these statistics as measures of gender relations of power. Thus the measure of time-use, to fit into the framework on statistical measurement cannot really measure how much 'sovereignty men and women have in structuring their time-use. It is possible to ask about satisfaction but that simply raises new questions of interpretation.
Similarly the provision of data on individual incomes of people within households is an important step further, but is still an inadequate measure of differential spending power, and still less of autonomy. It gives us a series which shows what income is due to different members of a household, but that is far from being the same thing as showing differential consumption standards, still less the differential power of different household members to determine the household's way of living. We could ask questions about consumption, but the essence of a household is about reciprocity. Just as we cannot assume fair sharing across households, we also cannot treat households as random collections on unrelated individuals. This prompts a further point about the statistics on the female deficit. It is an important political act in asserting the importance of women's access to independent income; at the same time it neglects important aspects of real life; i.e. that for some women such earnings have to cover themselves and their children; for others their own individual expenditure, but others do share in some of the spoils of their husband's earnings. Who earns the income matters, but so does the size and share of the household income women have access to.
There are two points I would like to draw from this: Gender Statistics will always need a bedrock of theory, supported by in depth qualitative analysis, that can be more nuanced. Only then can the structuring of desire, aspiration and power begin to be made manifest. Gender statistics have their place, but they are necessarily partial, reductive, confined to provide data within a pre-existing discourse. The second point I would draw from this is that any evaluation of policy measures cannot simply rely on the basis of narrowing any given gap between women and men. Analysis of pay trends in the United States and the United Kingdom shows that some of the narrowing of gender pay differentials at the bottom of the income distribution reflects the rising inequality of incomes; women at the bottom of the full-time earnings profile are doing better relative to men at a similar point on the male distribution, but only because men's earnings in this group have fallen relative to the mean; relative to the male and female median, they are worse off. Similarly increases in the proportion of women working full-time, or indeed part-time, are not always to be interpreted as being empowering, whatever the Chancellor thinks, in some cases they will reflect additional constraints, rather than additional freedoms.
Devolving gender statistics
I want to argue that there is an unfortunate bias towards the national in the approach to measuring the gender impact of policy. It is unfortunate because regional and local politics have always been more open to women and many important policies are devolved to the local or regional level, though often to unaccountable bodies. A healthier approach would be to open up the new regional agencies to scrutiny by regional women's groups supporting them to become effective Gender Statistics Users, with a role in developing still more local gender policy analysis. As it is, since the loss of effective local authority women's committees, local economic development and urban regeneration policy has become dis-engendered to a remarkable degree. The loss of gender awareness in regeneration, and therefore in social inclusion, strategies is masked by the very welcome programme of childcare provision. This has, however, become the only gender equality policy at local/regional level. (It is possible to argue that childcare provision should not be viewed exclusively as a policy for women... men are also parents, and children are supposed to benefit). We have lost the specialised training provision and support for women's projects that was part of the old Urban Programme and ESF.
I have just done a preliminary analysis of the successful bids in the last two Single Regeneration Budget (SRB) rounds and the absence of women or girls as a category of people expected to benefit from the programmes is astounding. We are talking here of expenditure of £1.2 billion in regional programmes over the two rounds. Out of the 250 projects there's just one explicitly targeted at women. That is in Catterick, where - as a garrison town - job opportunities for women are restricted. Otherwise lone parents get the odd mention here and there as groups to benefit from social inclusion strategies, but even this was decidedly more frequent in 1997/8 than in 1998/9. Quite a number of projects mention childcare provision as one element of their programmes. But again numbers dropped dramatically from SRB4 to SRB5. The childcare places identified as key output targets totalled 3400 for SRB4 and 2500 for SRB5, and 95% of these were in just 3 of the 9 regions, with none at all in the East Midlands, London, the North East and the South West. The 2500 places promised contrast with 104,204 jobs promised; that works out at one child care place for every 41 jobs, when 20% of the workforce and 22% of those looking for work have at least one child of school age and below. This suggests that for every 5 people in work, rather than every 40, there may be demand for some type of childcare.
Women may well benefit from the range of training and job creation programmes in the SRB, but we may never know without extensive on the ground research. There is no requirement to monitor job creation/retention and training provision in SRB programmes by gender, only ethnicity. It is interesting to note, too, that job creation is to be measured in full-time equivalents, with no provision for recording how many part-time jobs might be involved, still less is there any reference to pay or any quality dimension of job creation, other than whether a job is permanent (i.e. expected to last at least 6 months). There is one sole output measurement that mentions gender: the proportion of women amongst those involved/benefiting from community safety initiatives, though these need not be initiatives targeted at women, like women's self defence classes.
This is just one of the many failures of New Labour to join-up its policies. Gender mainstreaming is on the agenda and with it the development of suitable statistical series, but policies for regeneration and new job provision are to be the responsibility of regional development agencies. We can therefore expect little improvement in the gender awareness of the SRB. Not only is the Regional Development Agency (RDA) membership drawn from the business community and almost exclusively male, there are no national gender equality guidelines to inform their work. As a result none of the Regional Development Strategies developed by the new RD Agencies make reference to gender inequalities. Nowhere is a record expected of the degree to which inward investment strategies provide family-friendly forms of employment, or even the extent to which enterprise development supports the start-up and survival of women owned enterprises. Equality is treated exclusively an issue of ethnicity; at best social exclusion is seen to affect women as single parents on welfare. And although all the programmes contain community capacity building measures, the capacity for women to represent each other does not figure as a necessary element of community capacity building.
Although gender impact analysis and the development of gender monitoring statistics was pioneered at the metropolitan authority level, the production and analysis of gender statistics at the sub-national level is now woefully thin. This is partly an issue of sample size, as the excellent publications from the Northern Ireland EOC demonstrate. The Labour Force Survey (LFS) which is so useful for gender analysis (of pay for example, compared to the truncated New Earnings Survey sample) is not yet of a sufficient size to provide for a range of regional indicators, still less local indicators. The local data available by gender from NOMIS are severely restricted. The example of published unemployment statistics illustrates the point well.
When the government agreed to use the LFS to generate official monthly statistics of unemployment, based on self definition of job seeking rather than the claimant count, one of the key feminist campaigns for better statistics appeared to have been won. So national unemployment counts are now less biased against women than before. But the local and regional unemployment series as published in Labour Market Trends is based on the claimant count. This excludes by definition any one looking for a part time job and anyone else who cannot meet the stringent Job Seekers Allowance (JSA) requirements for job seeking, effectively to have childcare in place and to be willing to travel long distances and work employer determined rotas. The gulf between the gender breakdown of the claimant count and the Independent Labour Organisation (ILO) unemployed is as great today as it ever was. The claimant count covers only 48% of the female ILO unemployed as against 83% of the male.
This is not a symbolic issue. Once Census data are considered outdated the claimant count is the basis for identifying areas of employment need. The published local figures are not gender-differentiated. None of this would matter very much if the sex composition of the ILO unemployed was fairly constant across different local labour markets, but the county level data in the LFS suggest that they are not. In table 1 I have created a set of 30 counties and county amalgamations in England where the LFS sample is large enough to produce statistically valid comparisons and have ranked the counties on different measures of unemployment. Table 1 ranks counties by the ILO male unemployment rate, the female ILO rate and the undifferentiated claimant count rate for residents.
Click HERE for Table 1: County level unemployment
What this shows is that the claimant count is a reasonably good approximation to the spatial distribution of total male unemployment (with a Pearson correlation coefficient for the 30 counties of 0.98), it is not for women. The correlation for women was only 0.64.
The argument here is not so much that men are getting the giant's share of regeneration resources - though the emphasis on drugs and crime prevention rather than child care or even the prevention of teenage pregnancy in regeneration programmes certainly suggests that they are - rather it is that the differences in the needs of women in different areas are not reflected in the distribution of resources for regeneration. At a more pragmatic level, one can also suggest that the bias in the statistics produces a distorted account of the size of the local labour pools that might attract new investment.
Conclusion
My argument is that the gender statistics that we should be giving priority to improving are those that feed more or less directly into policy choices. We are not then talking about abstract statistics but the material effects of the way statistics are constructed. There is an enormous job to do in deconstructing such statistics, and in making them accessible so that they can become part of policy debates at all levels in society.
There are policy choices about how much emphasis is put on male as against female unemployment - and there are indeed ways in which the construction of unemployment rates de-emphasises the scale of the problem amongst men. Monica Threfell, for example, shows that the move to a more female friendly definition of employment - to include everybody working for at least an hour a week - underplays the scale of male unemployment in Britain. The argument is that radical statistics are statistics for a democratic society. In principle the shift towards community involvement in the design of SRB programmes, offers opportunities for policies to reflect local people's priorities. One of the things inhibiting this is the lack of good local statistics and the lack of knowledge and confidence that people have in using statistics. I am very pleased to see that community statistics are being discussed at the conference and I hope I have made the point that these also need to be gendered community statistics.
Irene Bruegel
School of Urban Development and Policy
South Bank University
i.bruegel@sbu.ac.uk
