How can we ensure everyone in society is represented in data analysis and outputs?

A clear message that emerged from our consultation activities was the importance of data robustness, in terms of representativeness and statistical validity. Additionally, the relevance and depth of the data are essential aspects that should be adequately considered. For providers of local services, this might mean having data about the local population which their specific services cater to. For users involved in regional or national advocacy, it could be having data to enable understanding of issues that impact a particular equality area or group in a manner that allows for disaggregation by characteristics of interest.

Being able to explore how different characteristics intersect to create different experiences for individuals within society is crucial to ensure that we “leave no one behind“. Similarly, the availability of geographically granular data is critical to provide policy-makers at a local level with the necessary evidence to inform their decisions. This is particularly relevant when considering data to inform the UK government “levelling up” agenda but is also essential for specific locally located institutions and services.

Individuals and groups who participated in our consultation activities identified a wide variety of issues as impacting on their use of data and inhibiting the extent to which statistics reflect the experiences of everyone in society. There was a general recognition that organisations and individuals could access a high volume of data on some but by no means all topics. However, deriving actionable insights from the available data was perceived as problematic, due to data gaps or data aggregation inhibiting value and usability. The issues that were consistently reported across the different consultation activities follow.

Lack of granularity in data to enable an understanding of specific subgroups and the intersectionality of personal characteristics

This challenge has been explored by participants in terms of the extent to which the data are fit for purpose, recorded in standardised formats and held in a condition that is findable, accessible and usable for analysis.

Throughout consultation exercises, participants emphasised the issue of data not allowing for disaggregation around key characteristics and the importance of understanding the intersection of different characteristics and the influence this has on various outcomes. For example, organisations who took part in our engagement activities reported that it is not enough to see data of interest disaggregated by ethnic group, but other characteristics are also required, such as ethnic group by age, or ethnic group by sex.

The current practice of aggregating personal characteristics into broader age, ethnic, faith, or sexual orientation categories was seen both to severely inhibit the inclusivity of statistics and to potentially misrepresent the issues and needs of smaller and more marginalised communities. The CSOs participating in this research highlighted that even within the data that are considered of the highest standard, such as the Census, the lack of granularity meant they were unable to develop an inclusive understanding of groups across the range of personal characteristics. This was seen to impact not just their services but the efficacy of public services and public policy, and ultimately people’s life outcomes.

CSOs and those organisations who took part in our online consultation mentioned their efforts to address the lack of granularity, by developing other sources of data to fill the gaps. These included:

• relying on qualitative research
• using administrative data collected by local services
• conducting their own surveys with service users
• trying to obtain data through Freedom of Information requests
• utilising published data to extrapolate national level data to the local context

This process was seen as potentially introducing error and bias, which were acknowledged as impacting on the quality and accuracy of insights available. Such approaches also were recognised as placing a high burden on organisations and individuals who typically have limited skills and resources to collect, collate, and/or analyse data.

Other reported areas of particular concern were missing data on marginalised and minority groups or those where the relevant characteristic may be temporary but there may be a long-term impact. CSOs highlighted that data were particularly lacking for:

• more recently defined or less understood groups (for example, those categorised as neuro-diverse, and particularly neuro-divergent women)
• those described as “harder-to-reach” (for example, Gypsy, Roma and Traveller groups, ex-prisoners, asylum seekers, young African Caribbean men)
• members of some types of communal establishment, for example prisons
• groups where data are collected in a non-standardised manner (and therefore expensive to collate and publish)
• those whose characteristics or circumstances are temporary, making it harder to capture intersectionality with outcomes of interest (for example, the number of pregnant people in employment)
• groups where it is perceived that there is limited political interest in publishing data (for example, numbers of asylum seekers, short-term or female prisoners or people who are homeless)

It was also felt that statistical data on its own cannot provide a comprehensive understanding of lived experience. Given the complexity, richness and intersectionality of issues affecting people’s lives, those who participated in the consultation activities consistently viewed it as crucial that qualitative analysis is carried out in addition to the production of statistics. This was highlighted as a means for providing fuller and more in-depth understanding of individuals’ lived experiences, and participants felt that greater inclusivity could be fostered through developing a more holistic picture of communities through the use of a combination of qualitative and quantitative analyses. It was also reported that, due to the lack of timely outputs, many services are reliant on qualitative research and more anecdotal data to understand the needs and circumstances of relevant groups and populations. Such insights are valued by CSOs, though recognised as less robust data for service planning in terms of estimating absolute levels of service need.

Some organisations and individuals who participated in the consultation activities also reported that the lack of linkages between current administrative datasets could be a missed opportunity to better understand under-represented groups. This was seen to result from the lack of a common, unique identifier (such as an NHS number) to allow for individual and household comparisons to be made across different datasets, and a “siloed mentality” in government policy and services. In some cases, this was felt to be exacerbated by weak or missing legislation and some organisations expressed that the framework for sharing personal data introduced under the Digital Economy Act was not explicit enough to enable the dataflow into the Office for National Statistics (ONS), for example. Given the resource and skills that exist within ONS there was a call, for ONS to play a stronger role in collating and connecting data in a meaningful, General Data Protection Regulation (GDPR)-compliant manner. It must be recognised, however, that there is resistance to data linkage from those who fear the misuse of data could be used to harm particular groups.

Lack of resources and/or skills to find and analyse data

Dedicated research capacity and the skills needed to source and analyse data was a common issue raised across the different consultation activities, in particular by CSOs, individuals from relevant groups and populations, and local government participants. Organisations and individuals with more experience working with data recognised the tension between the desire to have access to more granular data and the implications this has for data users. For example, most CSOs and many smaller local authorities do not have sufficient budgets to be able to employ people with the data analysis skills they need and, therefore, they do not have access to research skills in-house. For this reason, they would prefer summative data split by key socio-demographics and personal characteristics at a sufficiently granular level. This is particularly the case for those working more directly with service users. Participants requested data be provided to enable users to examine the intersection of personal characteristics, allowing the data to be broken into smaller sub-groups (for example, showing religious differences within ethnic groups, age and sex breakdowns across ethnic groups and sexual orientation by disability status).

Lack of suitable and accessible outputs

In relation to the reporting of equalities data, one of the main points stressed by individuals, groups and organisations who took part in our consultation activities is the difficulty for relevant groups and populations to relate to data outputs, due to broad categorisations and lack of granularity, as mentioned in the previous sections. The lack of readily available and easily accessible data for analysis was also reported. For example, data presented in traditional data tables were said to be time intensive to convert into a usable format for analysis. While some organisations were seeking raw data available for their own analysis, others required data that was already analysed in particular ways to suit their needs. For example, organisations who took part in the online consultation described difficulties exploring data which are not already analysed and having to download and analyse raw data.

Larger national CSOs, particularly those with an advocacy remit, tended to have a good understanding of the key data available on the groups which they worked with, while smaller and more local CSOs reported greater uncertainty as to exactly what data were available and where and from which organisation.

The use of simple language to help the interpretation of statistics and reassurance that the figures are produced by trusted sources were noted as key elements in accessing and feeling confident about information. To ensure inclusivity, participants stressed that facts and figures should be available in a range of formats to enable access for diverse audiences. The dissemination of findings in an exclusively online format, whether as data tables, online articles or statistical bulletins, was seen as preventing access for those who are digitally excluded, to whom the data may relate, and who instead have to rely on third parties, such as the media, to disseminate findings by more traditional means.