sps1000055a Social Policy & Society 10:2, 205–218 C© Cambridge University Press 2011 doi:10.1017/S1474746410000552 Developing Key Indicators of ‘Fairness’: Competing Frameworks, Multiple Strands and Ten Domains – an Array of Statistics S y l v i a W a l b y ∗ a n d J o A r m s t r o n g ∗∗ ∗Department of Sociology, University of Lancaster, UK E-mail: s.walby@lancaster.ac.uk ∗∗Department of Sociology, University of Lancaster, UK E-mail: J.E.Armstrong@lancaster.ac.uk This article argues that it is possible to develop key indicators to assess the broad agenda of ‘fairness’, even in the context of competing frameworks (e.g. equality, capabilities), multiple equality strands (gender, race/ethnicity, disability, age, religion/belief, sexual orientation, gender identity and social class) and several domains (e.g. health, education). Indicators are required to summarise complex statistical information to make it accessible to a wide audience. Methods for selecting indicators are discussed: identifying overlaps in frameworks, finding commonalities and applying quality criteria. Synthesising in this way enables identification of ‘key indicators’ of equality, including: intimate partner violence, risk of poverty and pay gaps. I n t r o d u c t i o n The measurement of inequalities and the development of indicators to summarise complex data are a necessary part of the evaluation of policies to reduce inequalities. The existence of competing frameworks, several legally protected equality strands, multiple policy domains and a wide range of potential statistics means that identifying indicators to assess ‘fairness’ is challenging. There are disparate and overlapping frameworks which conceptualise matters of fairness or justice differently, of which equality is one amongst many. These include: equality (equal treatment, opportunity or outcome); diversity and difference; good relations; human rights; and capabilities. Are there sufficient points of overlap to draw out a lead framework to serve as the basis for key indicators? There are several ‘equality strands’ to consider: the ‘conventional’ three of gender, ethnicity/race and disability, plus the more recently legally protected grounds of religion/belief, sexual orientation, age and gender identity (transgender). Social class occupies an ambiguous position; it has been to the forefront of recent debates (e.g. NEP, 2010), intersects or cuts across all other strands and is the subject of some unimplemented measures in the 2010 Equality Act (Act of Parliament, 2010). In addition, the range of equality/anti-discrimination issues has moved beyond a concern with employment to access to goods and services, largely reflecting developments in EU legislation. More recently, consideration of inequalities expanded to include different ‘dimensions’ with the Equalities Review (2007), a report commissioned by the government preparing for the new single equalities body in Britain, the Equality and Human Rights Commission (EHRC). This report specifies ten such dimensions or domains (i.e. ‘freedoms and activities that 205 Sylvia Walby and Jo Armstrong people have reason to value’ (Equalities Review, 2007: 6)): longevity; physical security; health; education; living standards; productive and valued activities; individual, family and social life; participation and influence; identity and expression; and legal security. Is it possible to put forward a single list of indicators that fulfils the demand for cross-strand coverage across these multiple domains? Statistics which relate to ‘fairness’ are unevenly available across the strands and domains; for some domains, such as education and employment, data can be accessed for most of the equality strands, while for other domains, such as legal security, it is more limited. Nevertheless, there are still many statistics to draw on. This raises the question of what criteria can be used to identify those most appropriate to serve as key indicators? One task of the EHRC is to map out existing inequalities in the Equality Measurement Framework (EMF),1 the foundation of which was laid in the Equalities Review (2007). The scope of the reports on the EMF is testament to the enormity of this task. The report by Alkire et al. (2009), which lays out the full framework, provides a rich and detailed proposal for the myriad of pieces of data proposed for the multi-faceted EMF. The aim of this article is slightly different to that of the EMF. The priority here is to identify key indicators that can be used in assessing ‘fairness’. Indicators summarise complex statistics in a form that is meaningful and accessible to a wide audience, including policy makers and the larger public (Walby, 2005). The article briefly introduces the competing frameworks of fairness, multiple equality strands and domains and the range of statistics that could serve as potential indicators. The methods for selecting indicators are then discussed: identifying points of overlap in competing frameworks, applying criteria of quality and finding indicators that are meaningful to as many strands as possible, without neglecting strand specificities. It is argued that combining these methods allows for synthesis and the development of a short list of key indicators. M u l t i p l i c i t y o f c o n c e r n s C o m p e t i n g f r a m e w o r k s o f f a i r n e s s The Equality and Human Rights Commission (EHRC) proposals for the Equality Measurement Framework (EMF) invoke multiple frameworks of fairness. Fairness is the lead or overarching concept used in the foundational review that preceded the establishment of the EHRC, the Equalities Review (2007). Five specific frameworks are identified in the Equalities Review, three of them grounded in legislation, one in practice in business and Non-Governmental Organisations (NGOs) and another in recent philosophical thought. These are: • equality, • diversity, • good relations, • human rights and • capabilities. First is ‘equality’. ‘Equal treatment’ is the core concept used in legislation aimed at removing certain forms of discrimination against protected equality groups. It is a legal requirement that individuals that belong to these protected equality groups be given ‘equal treatment’ with other people in specific areas of life, predominantly economically related. 206 Developing Key Indicators of ‘Fairness’ ‘Equal treatment’ is thus a specific application of the concept of equality in specific circum- stances. It is a process that is intended to deliver ‘equal opportunities’ in these fields. The legislative basis of ‘equal treatment’ has a long history. Key developments included those in the European Union (EU), which made it a requirement on all Member States, including the UK, to provide for this. These rights to equal treatment are embedded in treaties, starting with the Treaty of Rome in 1957, expanding to grounds in addition to gender, and domains in addition to employment in later treaties, especially the Treaty of Amsterdam in 1999. A long series of legally binding Directives derive from these treaties and are the basis for most of the UK legislation on equality and equal treatment (Europa, 2010). Equality is usually understood as an outcome rather than as a process. It is often, though not always, intended to be an outcome of equal treatment. There are significant differences in approach as to what are the most appropriate and important issues to which equality of outcome should apply (Phillips, 1999; Blond and Milbank, 2010). The concept of equality is given a distinctive twist in the UK policy documents setting up the EHRC. The Equalities Review (2007: 16) defines equality as follows: An equal society protects and promotes equal, real freedom and substantive opportunity to live in ways people value and would choose, so that everyone can flourish. Second is diversity, a recent addition to practices concerned with fairness. This framework is introduced into the definition of equality in the Equalities Review (2007: 16) in the following sentence: An equal society recognises people’s different needs, situations and goals and removes the barriers that limit what people can do and can be. There are two issues here. The first is that different treatment is sometimes needed in order to achieve equal outcomes, as is recognised in the disabilities legislation. The second is that the goals people might choose are recognised as being potentially different. Equality is thus not defined as achieving the same things. The third framework of fairness is ‘good relations’, which has long been used in UK legislation concerning ethnicity and race, as in the Race Relations Act (Act of Parliament, 1976). As Johnson and Tatam (2009) observe, while the concept has been used primarily in terms of race and religion/belief, it may have wider applicability to other strands. ‘Good relations’ is closely associated with a number of other concepts, including social capital, integration and community cohesion (Johnson and Tatam, 2009). Fourth is the ‘human rights’ framework, which has a long philosophical history and, for the UK, a more recent legislative one. Human rights was introduced into UK legislation in 1998 (Act of Parliament, 1998) following the adoption into law by the EU of the European Convention on Human Rights. In practice, this means the adoption of minimum or ‘threshold’ standards in relation to how individuals are treated by public bodies and the state, especially in regard to civil liberties and the use of violence (Kelly, 2005; Walby, 2009). The fifth framing of fairness is that of ‘capabilities’. The EMF (Alkire et al., 2009) and the Equalities Review (2007) are heavily influenced by the concept of capabilities, drawing on a body of philosophical work initiated by Amartya Sen (1999). These ‘capabilities’ are the ‘substantive freedoms’ that enable a person to do, or be, as they choose; they focus on a person’s freedom to have the capability to do and be the things they have reason to 207 Sylvia Walby and Jo Armstrong value (Sen, 2009: 231). This component of choice allows the possibility that the preferred outcomes may vary between people for valid reasons, and they may be diverse. In this way, this approach locates diversity, difference and choice at the centre of the framework. Capabilities are different from ‘functionings’, which are what people actually achieve (outcomes). This means that there are close parallels between the distinction between capabilities and functionings and the distinction between equality of opportunity and equality of outcome. The existence of a variety of frameworks of fairness represents the first and perhaps the most important challenge in identifying key indicators: which to use as the underpinning framework? M u l t i p l e s t r a n d s a n d d o m a i n s Recent policy developments have meant an increasing diversity of equality strands and relevant policy domains. The ‘conventional’ three equality strands or legally protected grounds – sex, race/ethnicity and disability – have been extended, largely to comply with EU Directives, to include age, gender identity, religion/belief and sexual orientation. The new Equality Act (Act of Parliament, 2010) introduces an eighth strand into the equalities legislation, with the proposed public sector duty regarding socio-economic inequalities. Under the Equality Act (2006), the EHRC also assumes responsibility for monitoring human rights, although discussion of a separate human rights body did take place prior to the creation of the Commission (and Scotland has a separate institution, the Scottish Human Rights Commission). Early equality law, following EU Directives, was focused on the equal treatment of equality groups with others in the field of employment, broadly defined (e.g. the transfer of pensions between partners is included since pensions are treated as deferred pay). This has been widened to include the sale and distribution of goods and services. Initially this was uneven between the legally protected grounds, but is being levelled up for all groups in the 2010 Equality Act. These domains have been widened still further by the requirement on public bodies to promote equality, initially for race, then disability and gender, and all grounds in the forthcoming public duty (Act of Parliament, 2010). A concern with inequalities is however no longer confined to those areas where there is legislation to provide for equal treatment, but is much wider, encompassing most areas of governmental policy making. The Equalities Review (2007) included a list of ten domains (capabilities or freedoms), drawing on background papers written from the perspectives of Sen’s capabilities and international human rights (Burchardt, 2006; Burchardt and Vizard, 2007). These domains are: longevity; physical security; health; education; living standards; work; individual, family and social life; participation and influence; identity and expression; and legal security (Alkire et al., 2009). Thus, a second challenge in identifying key indicators is to ensure adequate and appropriate coverage of all the strands across the ten domains. A n a r r a y o f s t a t i s t i c s Several reviews of data on equalities in Britain, including the Office for National Statistics (ONS) review (2007b, 2007c), the ONS (2003) Brief Guide to Gender Statistics, and Breitenbach’s (2006) review for the Equal Opportunities Commission, have revealed the availability of a wide range of statistics. A review of the data was also conducted by 208 Developing Key Indicators of ‘Fairness’ the authors of the current article (Walby et al., 2008), commissioned by the Equality and Human Rights Commission, upon which this article draws. The starting point for the review was the Equalities Review (2007), including the contributions of the background papers (e.g. Burchardt, 2006; Burchardt and Vizard, 2007; Vizard and Burchardt, 2007). The review included an examination of the work of the legacy commissions, especially, but not only, their research and policy work on key statistics and indicators (e.g. Commission for Racial Equality, 2006; the Disability Rights Commission, 2007; Equal Opportunities Commission, 2007a). The work of relevant government departments was considered, including the Department for Work and Pensions (e.g. the disability equality indicators, Office for Disability Issues, 2008a), and Department for Communities and Local Government (indicators for local authorities, DCLG, 2007a; the race equality in public service indicators, DCLG, 2007b). Government performance indicators were reviewed, including those of the UK ‘Public Service Agreements’ (e.g. HM Treasury, 2007) and the Scottish Government National Performance Framework Indicators. Key publications and reports by civil society organisations, other policy bodies and academics were reviewed especially, but not only, for those strands which did not previously have a Commission (i.e. age, sexual orientation and religion/belief), and for those dimensions (e.g. violence against women) that were not previously central to the work of those Commissions. In addition, a systematic review of the major surveys (e.g. the Labour Force Survey, the British Crime Survey) and sets of administrative data (e.g. housing statistics) was undertaken to assess the extent to which they provided relevant data. The review generated a list of around 200 key statistics on equalities (see Walby et al., 2008). More recently, the review has been further developed in the report by Alkire et al., (2009), which consists of a vast range of statistics to be used in populating the EMF. As these reviews evidence, the availability of data is somewhat uneven, both in terms of the equality strands for which information is available, and in terms of the varying coverage of the ten domains. In part, this reflects the history of the treatment of various groups (e.g. the different timing of legal protection for each social group) and the way in which some domains clearly map onto well-established policy areas (e.g. living standards), while others (e.g. identity and expression) do not. Nevertheless, the range of statistics is substantial, which raises a third challenge: how to select those to serve as key indicators? M e t h o d s f o r s e l e c t i o n : o v e r l a p s ; c o m m o n a l i t i e s a n d s p e c i fi c i t i e s ; a n d q u a l i t y c r i t e r i a The process of selecting key indicators involved three challenges: the attempt to reconcile disparate frameworks, the appropriate treatment of multiple strands and domains and choosing the best statistics to serve as indicators. Other difficulties associated with selection, in particular those concerned with data availability and the choice of technical measures, are discussed in greater depth elsewhere (Walby et al., 2008; Walby and Armstrong, 2010). I d e n t i f y i n g o v e r l a p s i n t h e f r a m e w o r k s o f f a i r n e s s The operationalisation of the notion of fairness is contested at least partly because there are rival conceptualisations, both internal to the notion of equality, such as equality of outcome or opportunity, and also due to the increased policy significance of adjacent 209 Sylvia Walby and Jo Armstrong frameworks, including human rights and capabilities. The range of frameworks raises a question of whether it is possible to suggest a single list of indicators or whether multiple lists are required. There is tension between the concept of ‘equality’ with those of diversity or choice. If there is diversity, then it is hard to know what standard to use in measuring equality. In its review of equality statistics, the Office for National Statistics (2007b: 74) comments that there is a need to clarify the relationship between equality and diversity. One aspect of the issue identified by the ONS is that by including diversity and choice so centrally, it makes the specification of the standard or outcome to be measured rather complex, since the implication might be that there can be multiple desired outcomes on any given issue, but the operationalisation of the concept of ‘outcome’ in statistics requires just one per issue. One illustration of the potential implications of accepting diversity is found in long-running feminist debates regarding sameness, difference and transformation in gendered divisions of labour (e.g. Fraser, 1997). These discussions have highlighted that an acceptance of difference and diversity can embed rather than challenge inequality. For example, research and campaigning by the legacy Equal Opportunities Commission (2007b) demonstrated the risks (e.g. homogenisation and stigmatisation) in using ‘difference’ arguments to explain the low participation of some minority ethnic women in employment; the EOC’s work instead pointed to the barriers and exclusions (e.g. discrimination in the labour market) experienced by these women, where ‘difference’ is actually indicative of inequalities in access to paid work. At the same time, however, in working towards equality it is important to avoid a position which ignores difference. For example, the Disability Discrimination Act (1995) places a duty upon employers to make reasonable adjustments in order that people with impairments are not disadvantaged or disabled by obstacles associated with the employment. In these circumstances, different treatment (i.e. making adjustments to remove the disabling barriers) may be required to achieve equal outcomes (House of Commons Work and Pensions Committee, 2009). The concept of ‘good relations’ is parallel to the concept of equality, but not the same. There are potential divergences where there are actions that might promote good relations that are additional to or other than equal treatment (e.g. inter-group and inter-community initiatives). But there are potentially significant overlaps between good relations and equal treatment in that equal treatment is likely to be perceived as constructive of good relations, while disproportionately worse treatment is likely to undermine good relations. As Johnson and Tatam (2009) observe, there are strong arguments in favour of extending the applicability of the concept of good relations beyond a conventional reference to race/ethnicity or religion/belief. Their review of the research aims to provide the basis for a measurement framework for good relations, to sit alongside the EMF. However, given the close links of the concept (and the causal processes implicated) with equality, the necessity of an additional measurement framework is questionable. Indeed, the Equalities Review (2007: 18) implied the need for consensus on a single framework for measurement, and the very existence of multiple frameworks may also be seen as potentially divisive. There is both tension and overlap between the concepts of equality and human rights. There is overlap in the sense that all human beings are regarded as equally entitled to human rights. There is tension in two respects. First, the measurement of equality is 210 Developing Key Indicators of ‘Fairness’ usually through the investigation of gaps; while the identification of whether there is a human rights abuse usually involves the application of a minimum threshold. Second, the policy domains where the concepts of equality and human rights have been traditionally applied and have achieved consensus status are different: equality (or more often equal treatment) is more often used in relation to economic and material matters, as in equal treatment in employment and the sale of goods and services; human rights are more often used in relation to excessive violence and to free speech and association. The use of a minimum threshold as opposed to the measurement of gaps, while effectively used in some indicators is limited in certain respects. For example, reaching a certain standard of living is regarded as a ‘human right’, in that no one should live in extreme poverty; however, beyond this standard such a framework becomes increasingly redundant. The relative disadvantage experienced by a group cannot be compared with the privileged conditions of the dominant groups. The gap between the two may be small, but it may also be large. The use of a threshold omits this information. It is possible that a majority of the population reach the threshold, but with large disparities between those who just reach it and those who far exceed it. Nevertheless, statistics on thresholds can be usefully transformed to measure such gaps. For example, the percentage of women living at or below the poverty line can be compared with the percentage of men, and so forth. This shifts the framework used though, from a framing of poverty as a human rights issue, to a framing in terms of inequality. Using the framework of human rights can extend the range of domains considered, for example, to include violence. But the addition of this framing may not always be necessary. Violence against women is commonly conceptualised as both a cause and consequence of gender inequality (HM Government, 2009). In the case of violence though, the goal is not to narrow the gap (since overall levels of violence can increase at the same time gaps between groups decrease), but to reduce violence against all groups. There is a tension between equality and capabilities. In part, this is a consequence of the varied ways in which Sen’s work has been interpreted. An important question is whether or not in practice capabilities and opportunities can be measured separately from functionings and outcomes. In their work on the EMF, Alkire et al. (2009: 2) distinguish outcomes from ‘processes’ (inequalities in treatment through discrimination or disadvantage) and ‘autonomy’, which is closest to notions of capability and choice (i.e. inequality in the degree of empowerment people have to make decisions affecting their lives). Despite their capabilities framework, Alkire et al. recommend using outcomes and processes, not autonomy, noting that measures of autonomy are not available and thus cannot be used. While they suggest further work on autonomy, it is questionable whether this will be fruitful. Social science has long critiqued notions such as choice; since choices and preferences are shaped by circumstances, which are likely to be affected by power and inequality, those in disadvantaged circumstances are likely to adapt their preferences to what is likely to be possible (Qizilibash, 1997). As a consequence, the preferences of the poor are likely to be more limited in ambition than the preferences of the rich. Such problems lead most to favour inequalities in outcomes as the best reflection of inequalities in capabilities (e.g. Robeyns, 2003). Most of the statistics Alkire et al. propose are measures of outcomes; some are measures of perceptions of processes. However, the use of perceptions as evidence is notoriously suspect in social science, being volatile, highly sensitive to wording of 211 Sylvia Walby and Jo Armstrong questions and with a varying rather than predictable relationship to the actual phenomena. For example, greater perceived discrimination is reported by the white population than by minority ethnic groups in access to housing (Walby et al., 2008), a finding at odds with the actual allocation of social housing (Rutter and Lattorre, 2009). Data on perceptions are thus rarely a reliable or robust measure to serve as the basis for key indicators. Utilising a capabilities framework therefore tends to produce many convolutions in practice, suggesting that it is perhaps better to retain this framework as a philosophical background and use equality in outcomes as the most robust measure. This reflects the conclusion reached by the Equalities Review, that measurement of inequality in outcomes actually attained in many cases will be a good indicator of whether people have the real freedom in question (Equalities Review, 2007: 130). The extent of overlap between an equalities framework with the other frameworks, together with its strength and robustness in terms of measuring fairness, suggest this framework as the lead contender for underpinning a single list of key indicators. This is not surprising given that there is a body of law, government and institutions that have sought for more than 30 years to operationalise the concept of equality and to embed it in legal, governmental and institutional practice. The many dilemmas and tensions in the equality agenda have been subject to democratic process, contestation between rival stakeholders, interpretation and reinterpretation in law, public debate and compromise. The institutional context for equalities has included the development of government units, several Commissions, a diverse and thriving NGO sector, new developments among trade unions and centres of research excellence in universities. In this rich and diverse context, there have been increasingly substantial attempts to specify, codify and operationalise the concept of equality. Debates on these statistics and indicators have led to increasing sophistication in their formulation. This has been a process involving simultaneous expertise and deliberative argumentation among the many stakeholders involved. S e a r c h i n g f o r c o m m o n a l i t i e s , a n d a t t e n d i n g t o s p e c i fi c i t y : s t r a n d s a n d d o m a i n s As noted above, the number of strands which are legally protected has expanded over recent years, largely following EU Directives. This expansion means that it becomes harder to find a limited number of indicators that are relevant to all strands. This is without considering the additional challenge of tackling the ways equality strands intersect. The equality legislation, which has traditionally concerned equal treatment in economic areas of life, has addressed the equality grounds that are protected under it in a similar way. This has largely concerned anti-discrimination measures. However, beyond the economy, the equality issues of most concern to each strand tend to diverge as a consequence of the ontological dissimilarity of these regimes of inequality (Verloo, 2006; Walby, 2009). Thus, while within the economy it is not too hard to find indicators that work across all strands, it becomes more of challenge beyond the economy. Extending legal protection and expanding the domains is clearly a welcome step in moving towards greater equality, but it poses challenges in terms of identifying indicators. One solution to this complexity is to search for commonalities. Some disadvantages and inequalities appear common across the strands: e.g. pay gaps, gaps in employment rates 212 Developing Key Indicators of ‘Fairness’ and gaps in risk of poverty. Their wide applicability makes them suitable contenders for key indicators. Thus, pay gaps, a well-established indicator of gender inequality, can be meaningfully applied to other equality strands, such as disability and ethnicity. Likewise, the measure of self-reported health can be used as a valuable indicator across different social groups. A search for commonality in the inequalities experienced by different groups for the purpose of selecting key indicators however should not be confused with implying that all regimes of inequality are the same, nor that all can be reduced to one. If one strand becomes dominant as an explanation of all inequality, and everything else is reduced to this, then it will become impossible to accurately identify causal paths, explain or tackle (and may even worsen) other separate, but interrelated, inequalities. In domains less related to economic matters, the task of identifying indicators appropriate to highlighting inequalities across the strands becomes a little more difficult. If the method of searching for commonality is applied without regard for strand specificity, then it risks becoming a somewhat meaningless exercise, resulting in fragmentation and a loss of the most urgent and pressing matters. This means that for some domains, different issues facing each strand group require emphasis by selecting different indicators. Where prioritisation by commonalities is not possible, such as in the domains of physical and legal security, then it becomes more appropriate to select different indicators for different strands (e.g. the specificity of violence against women: intimate partner violence; and hate crimes or targeted violence: racially or religiously aggravated violence, targeted violence against disabled people and homophobia). Hence, there is a need to select indicators that have the widest relevance for all groups wherever possible, but with recognition that some differences remain and are too important to neglect. Q u a l i t y c r i t e r i a : r e l e v a n c e a n d c l a r i t y By contrast with the requirement to cover a multiplicity of concerns in the EMF, the task of identifying key indicators requires a selection of a very few measurements on which to shine a spotlight in order to focus and direct high-level policy attention. Some criteria of quality apply to both the EMF and key indicators, for example the seven criteria of statistical quality developed by the Office for National Statistics (ONS, 2007a) – relevance, accuracy, timeliness, accessibility and clarity, comparability and coherence – but some are particular to each. While the criteria of quality for the EMF include the provision of many hundreds of detailed measurements in order to fill the cells of its multiplicity of concerns, the criteria of quality for the selection of key indicators include simplicity for the purpose of clarity and focus. The most powerful indicators are those that are the simplest. For example, the effectiveness of the UNDP’s Human Development Indicator, which challenged the established indicator of economic growth, was at least partly due to its being a single figure, according to the Director of the Human Development Reports (Fukuda-Parr, 2003). Building on and developing the work of others, Walby (2005) identifies a list of requirements necessary for an indicator to be fit for purpose. Indicators should: summarise complex data, be unambiguous and easy to interpret, enable assessment of whether improvement or deterioration has occurred, be meaningful and relevant, be capable of 213 Sylvia Walby and Jo Armstrong being supported by robust and reliable quantitative data, available at regular intervals with comparability between countries and population groups and be neither so many as to confuse, or so few as to mislead. The application of these criteria means the exclusion of many potential statistics, in particular measures that lack robustness and clarity. For example, crime reported to the police is not considered a reliable indicator of levels of violent crime since it is affected by people’s willingness to report incidents and varying practices of recording; likewise, data based on people’s perceptions is frequently considered inadequate since it is affected by the views of others and fluctuates independently of real changes in observed trends. In selecting indicators that are relevant and meaningful, priority is given to those measures previously and currently used by both governmental and civil society groups, with awareness of points of contention between the various stakeholders. This gives priority to well-established indicators, such as the pay gap and political participation, with assessment of their relevance for strands where their use has been less common. Thus, the pay gap has long been used to assess inequalities by gender, but it also serves as a valuable indicator for other equality strands, including ethnicity and disability. S y n t h e s i s i n g : t h e s u g g e s t e d k e y i n d i c a t o r s The selection of key indicators invoked a number of methods: quality criteria were applied to ensure that the statistics could be appropriately used as indicators; points of overlap between competing frameworks were identified to reconcile disparate concepts; and common measures relevant across the strands were sought, balanced against the need to attend to strand specific issues. It was also necessary to work within the limits of the data available. Indicators are to be deployed to work towards equality, but they are formulated in conditions that are unequal and thus a degree of pragmatism and compromise is required in their development. Using these methods simultaneously leads to a synthesis where it becomes possible to identify ‘key indicators’. Equality emerges as a lead framework, not least because when operationalised in terms of outcome it offers the most robust measure of fairness. This can be seen in work of practitioner communities, including the legacy commissions for equality, which tend to adopt and refine statistics, such as pay gaps, gaps in risk of poverty and levels of violence for use as key indicators. Applying the criteria of quality meant seeking measures that are both relevant and easily understood. Again, previous work on indicator sets by government, the legacy commissions and NGOs is useful in pointing to the most relevant issues confronting the various equality strands. Together with a demand for cross-strand relevance, this led to a search for common indicators, but not at the expense of those indicators that are necessary to highlight strand specific issues. Following these processes suggests a list of key indicators in the ten domains; these are presented below together with a short note regarding their selection (for further discussion, see Walby et al., 2008): 1 Longevity: gaps in infant mortality rates This statistic was selected over other measures, such as life expectancy, since it has wider strand relevance, in particular serving to highlight inequalities by ethnicity. 2 Physical security: incidents of intimate partner violence, and hate crime, as measured by crime surveys. 214 Developing Key Indicators of ‘Fairness’ The selection of two measures reflects strand specificity in issues of violence. 3 Health: gaps in self-reported general health. This measure covers a wide range of issues in the domain, has relevance across the equality strands and is unambiguous and easily interpreted. 4 Education: gaps in qualifications of adults, and in educational attainment at the end of compulsory schooling. These indicators were selected in particular for their relevance and cross-strand coverage. 5 Standard of living: gaps in the risk of living in a low-income household, and the Gini measure of inequality of income distribution. The former indicator is widely used to highlight inequalities by gender, disability, race/ethnicity, age; the second, again widely used, offers useful contextual information for the first, particularly in examining changes in inequality over time. 6 Productive and valued activities: pay gaps and employment gaps, and vertical occupational segregation. These indicators are selected to assess long-standing inequalities in the field of paid work; their widespread use attests to the significance of the issues they highlight for most if not all of the equality strands. 7 Individual, family and social life: gaps in the extent of independent living. Selecting a key indicator for this domain was problematic due to a lack of appropriate statistics. Nevertheless, data are becoming available as the concept of ‘independent living’ is being developed (Office for Disability Issues, 2008b). 8 Participation, influence and voice: gaps in the proportion of Members of Parliament and in councillors/council leaders; gaps in membership of the boards of top companies. The first indicator is widely used to assess inequalities in political participation and has cross-strand relevance; the second extends the concept of participation to highlight inequalities in economic governance. 9 Identity, expression and self-respect: perceived discrimination in employment, and perceived discrimination in accessing services. These indicators, based on perceptions of discrimination (measures which, as detailed above, are not robust), were selected in the absence of suitable alternatives and may be considered as temporary until more appropriate measures are developed. 10 Legal security: justice gaps, including disproportionality in the treatment of alleged offenders in the Criminal Justice System, in the number of people stopped and searched; and attrition rates for targeted violence against equality groups, including rape, intimate partner violence and hate crime. These indicators were selected to highlight the most relevant inequalities within the limits of the available data; given the strand specificity of the issues, more than one indicator is selected. C o n c l u s i o n s This article set out to identify a simple list of key indicators of fairness, taking into account the challenges of competing frameworks, multiple strands and domains and an array of statistics upon which to draw. It has been argued that a combination of methods (identifying overlaps, selecting robust measures, prioritising relevance and clarity, finding 215 Sylvia Walby and Jo Armstrong areas of commonality without neglecting specificity) allows the development of a single short list of key indicators. Synthesising in this way brings together and draws upon separate though adjacent fields of work: the philosophical underpinnings of different frameworks are juxtaposed against attempts to operationalise the concepts; the issues confronting different strand groupings in various domains are compared and contrasted; and detailed and complex statistical information is evaluated using criteria developed by expert practitioners in order to assess its relevance and clarity. The disparate knowledge bases and expertise of statisticians, social scientists and equality practitioners are therefore simultaneously drawn upon. Both the use of indicators and their development necessarily involves a loss of nuances and complexity; communicating detailed statistical information in a way that is accessible and meaningful would otherwise be impossible. Detail is sacrificed to realise the benefits of making inequalities visible, enabling assessments of progress, and to provide the tools for practitioners to intervene strategically and effectively in policy fields. N o t e 1 The change of government in May 2010 occurred after this paper was completed and before this Review using the EMF was due to be completed. R e f e r e n c e s Act of Parliament (1976), Race Relations Act, Chapter 74, http://www.opsi.gov.uk/acts/acts1976/pdf/ ukpga_19760074_en.pdf [accessed 08.04.2010]. Act of Parliament (1998), Human Rights Act, Chapter 42, Office of Public Sector Information, http://www.opsi.gov.uk/acts/acts1998/ukpga_19980042_en_1 [accessed 08.04.2010]. Act of Parliament (2010), Equality Act, Chapter 15, http://www.opsi.gov.uk/acts/acts2010/ukpga_ 20100015_en_1 [accessed 14.04.2010]. Alkire, S., Bastagli, F., Burchardt, T., Clark, D., Holder, H., Ibrahim, S., Munoz, M., Terrazas, P., Tsang, T. and Vizard, P. (2009), Developing the Equality Measurement Framework: Selecting the Indicators, Manchester: EHRC. Blond, P. and Milbank, J. (2010), ‘No equality in opportunity’, The Guardian, 27 January, http://www.guardian.co.uk/commentisfree/2010/jan/27/inequality-opportunity-egalitarian-tory- left [accessed 08.04.2010]. Breitenbach, E. (2006), Gender Statistics: An Evaluation, Equal Opportunities Commission Working Paper Series no. 51, Manchester: EOC. British Crime Survey, see: Economic and Social Data Service, http://www.esds.ac.uk/government/bcs/ [accessed 20.12.2010]. Burchardt, T. (2006), ‘Foundations for measuring equality: a discussion paper for the Equalities Review’, CASE, London School of Economics, http://archive.cabinetoffice.gov.uk/equalitiesreview/ upload/assets/www.theequalitiesreview.org.uk/foundations4measuring.pdf [accessed 02.04.2010]. Burchardt, T. and Vizard, P. (2007), ‘Definition of equality and framework for measurement: final recom- mendations of the Equalities Review Steering Group on Measurement’, Paper 1, The Equalities Review, CASE, London School of Economics, http://archive.cabinetoffice.gov.uk/equalitiesreview/upload/ assets/www.theequalitiesreview.org.uk/paper1equality.pdf [accessed 02.04.2010]. Commission for Racial Equality (2006), Factfile 1: Employment and Ethnicity, London: CRE. 216 Developing Key Indicators of ‘Fairness’ Department for Communities and Local Government (2007a), The New Performance Framework for Local Authorities and Local Authority Partnership: Single Set of National Indicators, London: DCLG. Department for Communities and Local Government (2007b), Improving Opportunity, Strengthening Soci- ety: Two Years On, a Progress Report on the Government’s Strategy for Race Equality and Community Cohesion, London: DCLG, http://www.communities.gov.uk/publications/communities/twoyearson [accessed 05.04.2010]. Disability Discrimination Act (1995), http://www.opsi.gov.uk/acts/acts1995/plain/ukpga_19950050_en# pt2-pb2 [accessed 24.06.2009]. Disability Rights Commission (2007), Celebrating the Journey: Impact Report 2000–2007, http://www.celebratingthejourney.org/files/lowresPDFimpactreport.pdf [accessed 02.04.2010]. Equal Opportunities Commission (2007a), The Gender Equality Index: Completing the Revolution, the Leading Indicators, http://www.train2000.org.uk/research-reports/pdfs/GenderEqualityIndex.pdf [accessed 20.12.2010]. Equal Opportunities Commission (2007b), ‘Moving on up? The way forward, Report of the EOC’s investigation into Bangladeshi, Pakistani and Black Caribbean women and work’, Equal Opportunities Commission, Manchester. Equality Act (2006), http://www.opsi.gov.uk/acts/acts2006/ukpga_20060003_en_1 [accessed 24.06.2009]. Equalities Review (2007), Fairness and Freedom: The Final Report of the Equalities Review, London: Cabinet Office, http://archive.cabinetoffice.gov.uk/equalitiesreview/upload/assets/www.theequalitiesreview. org.uk/equality_review.pdf [accessed 05.04.2010]. Europa, ‘Summaries of EU Legislation: Employment and Social Policy’, http://europa.eu/legislation_ summaries/employment_and_social_policy/index_en.htm [accessed 20.12.2010]. Fraser, N. (1997), Justice Interruptus, New York: Routledge. Fukuda-Parr, S. (2003), ‘The human development paradigm: operationalizing Sen’s ideas on capabilities’, Feminist Economics, 9, 2–3, 301–17. HM Government (2009), Together We Can End Violence against Women and Girls: A Strategy, London: Home Office, http://www.homeoffice.gov.uk/documents/vawg-strategy-2009/end-violence-against- women2835.pdf?view=Binary [accessed 31.03.2010]. HM Treasury (2007), ‘PSA Delivery Agreement 15: address the disadvantage that individuals experience because of their gender, race, disability, age, sexual orientation, religion or belief’, http://www.hm- treasury.gov.uk/media/E/8/pbr_csr07_psa15.pdf [accessed 05.04.2010]. House of Commons Work and Pensions Committee (2009), The Equality Bill: How Disability Equality Fits within a Single Equality Act, London: TSO, http://www.publications.parliament. uk/pa/cm200809/cmselect/cmworpen/158/158i.pdf Johnson, N. and Tatam, J. (Institute of Community Cohesion) (2009), Good Relations: A Con- ceptual Analysis, Manchester: EHRC, http://www.equalityhumanrights.com/uploaded_files/research/ research_report_42_-_good_relations_a_conceptual_analysis.pdf [accessed 05.04.2010]. Kelly, L. (2005), ‘Inside outsiders: mainstreaming violence against women into human rights discourse and practice’, International Feminist Journal of Politics, 7, 4, 471–95. Labour Force Survey, see: Economic and Social Data Service, http://www.esds.ac.uk/government/lfs/ [accessed 20.12.2010]. National Equality Panel (2010), An Anatomy of Economic Inequality in the UK, Report of the National Equality Panel, London: Government Equalities Office, http://www.equalities.gov.uk/ pdf/NEP%20Report%20bookmarkedfinal.pdf [accessed 31.03.2010]. Office for Disability Issues (2008a), Disability Equality Indicators, London: ODI, http://www. officefordisability.gov.uk/research/indicators.php [accessed 05.04.2010]. Office for Disability Issues (2008b), Independent Living: A Cross-Government Strategy about Independent Living for Disabled People, London: ODI, http://www.odi.gov.uk/docs/wor/ind/ilr-executive-report. pdf [accessed 24.05.2010]. Office for National Statistics (2003), Brief Guide to Gender Statistics, London: Office for National Statistics. 217 Sylvia Walby and Jo Armstrong Office for National Statistics (2007a), Guidelines for Measuring Statistical Quality, Version 3.1, London: ONS, http://www.statistics.gov.uk/downloads/theme_other/Guidelines_Subject.pdf [ac- cessed 05.04.2010]. Office for National Statistics (2007b), Report from the Review of Equality Data, London: ONS, http://www.ons.gov.uk/about-statistics/measuring-equality/equality-data-review/index.html [ac- cessed 05.04.2010]. Office for National Statistics (2007c), Report from the Review of Equality Data, Companion paper to the final report containing detailed information collected through a data audit, Lon- don: ONS, http://www.ons.gov.uk/about-statistics/measuring-equality/equality-data-review/index. html [accessed 05.04.2010]. Phillips, A. (1999), Which Equalities Matter? Cambridge: Polity Press. Qizilbash, M. (1997), ‘A weakness of the capability approach with respect to gender justice’, Journal of International Development, 9, 2, 251–62. Robeyns, I. (2003), ‘Sen’s capability approach and gender inequality: selecting relevant capabilities’, Feminist Economics, 9, 2–3, 61–92. Rutter, J. and Latorre, M. (2009), Social Housing Allocation and Immigrant Communities, Manchester: EHRC, http://www.equalityhumanrights.com/uploaded_files/4_social_housing_allocation_and_ immigrant_communities.pdf [accessed 08.04.2010]. Sen, A. (1999), Development as Freedom, Oxford: Oxford University Press. Sen, A. (2009), The Idea of Justice, London: Allen Lane. Verloo, M. (2006), ‘Multiple inequalities, intersectionality and the European Union’, European Journal of Women’s Studies, 13, 3, 211–28. Vizard, P. and Burchardt, T. (2007), ‘Developing a capability list: final recommendations of the Equalities Review Steering Group on Measurement’, Paper 2, The Equalities Review, Centre for Analysis of Social Exclusion, London School of Economics. Walby, S. (2005), ‘Improving the statistics on violence against women’, Statistical Journal of the United Nations Economic Commission for Europe, 22, 193–216. Walby, S. (2009), Globalization and Inequalities: Complexity and Contested Modernities, London: Sage. Walby, S., Armstrong, J. and Humphreys, L. (2008), Review of Equality Statistics, Manchester: Equality and Human Rights Commission, http://www.equalityhumanrights.com/uploaded_files/ research/1__review_of_equality_statistics_241008.pdf. Walby, S. and Armstrong, J. (2010), ‘Measuring equalities’, International Journal of Social Research Methodology, 13, 3, 237–49. 218