National polls play an important part in the narrative around public policy and are often held up to be representative of the diverse make-up of the UK public. In this blog, Siobhan Morris, Professor Ann Phoenix, and Dr Olivia Stevenson consider how representative they really are and what impact this could have on the government’s Levelling Up ambitions.

With the recent elections held across the UK, polling of the UK public has once again been extensively deployed in the name of gauging societal opinion. Whether it be attitudes to ‘partygate’ or assessing priorities to inform levelling up, ‘public opinion’ is given significant weight by politicians, media, policy professionals, and academics alike. Yet too little open dialogue has occurred asking the question ‘just how representative of the ‘public’ polls really are?’.
When we commissioned YouGov polling into attitudes regarding ‘building back better’ after the Covid-19 pandemic, our aim was to investigate how opinions differed by geography, gender, ethnicity, age and other characteristics, with data weighted to match the UK population. We polled almost 7,000 people on the areas that they felt should be prioritised for investment, and whether they thought that “building back better” could be achieved.
Headline figures appeared to show widespread, unified support for the slogan, but examining the views of particular population groups told a different story. For example, attitudes towards the importance of “building back better” varied by age and ethnicity. 84% of Black, Asian and minority ethnic respondents aged 65 years and older reported it was either “very important” (39%) or “fairly important” (45%). In comparison, 53% of white respondents in the same age group agreed, with figures of “very important” (19%) or “fairly important” (34%).
This analysis, however, does not delineate the different population groups within the categories of Black, Asian, minority ethnic and white. The option to delineate was not available within the polling package offered, nor is it included ‘as standard’. Assessing and analysing public opinion data through intersectional lenses, rather than opting for broad-brush findings across the ‘population’ as a whole, is rare.
The Government Family Resources Survey shows that 1 in every 5 people are disabled in the UK and figures reported by Scope, suggest that a high proportion of this number are Black and Minority Ethnic. It is therefore striking that neither ethnicity nor disability are included as standard demographic measures in the vast majority of polling conducted at the national level.
Including a representative range of minoritised ethnic voices[1] in national polls – such as those polls used to assess priorities for levelling up – is described by pollsters, such as YouGov, as ‘not easy’. Likewise, whilst polls seek to understand and report on how people experience living with disabilities in the UK, disabled people are frequently only identified in analyses of ‘disability only issues’ and the study of disability, rather than as standard in general demographic breakdowns within polling.
Delineating the priorities of different ethnic groups or disabled groups is frequently lacking from national polling on levelling up.[2] The views of entire population groups are not being included in the evidence base informing and underpinning this key debate. The strategies for inclusion of people within polls affects the insights they return. So, the ‘voice of the people’, as reflected in polling data, is inseparable from the exclusionary choices made from the outset. Polls are highly influential in shaping public policy, but the population-level claims made about UK public attitudes are not often supported by the sampling frame.
As is noted in the Levelling Up White Paper, metrics to be explored include “restoring a sense of community, local pride and belonging, and empowering local leaders and communities”. Yet, when assessing priorities for addressing programmes of work in these areas, polling could (and often is) deployed. Indeed, national polling organisations frequently conduct polls that feed into policy development or public attitudes for the UK Government to draw upon.
The UK Government is bound to a number of conventions (for example, United Nations Convention on the Rights of Persons with Disabilities) requiring them to formulate and implement policies – using research – to combat stereotypes, prejudices, and harmful practices against people with characteristics protected under the Equalities Act 2010. It has also committed to monitoring public perceptions through the Office of National Statistics, in areas such as the effects of policies on equalities.
At the same time national polling organisations – YouGov, IPSOS Mori, NatCen etc – are governed by the British Polling Council, which aims to assure the standards of survey results by ensuring the reliability and validity of the results.
Whilst these offer protections around the production of evidence and the formulation of policy, if national polling frequently excludes certain groups within UK society, then discussions of ‘public perceptions’ and how they can inform the Levelling Up programme and its policy priorities are at best flawed and at worse, inaccurate. High-quality intersectional research on public perceptions remains limited.
It is no longer sufficient to accept the exclusion of groups that pollsters consider ‘hard-to-reach’. As polls are now accepted as constituting public opinion by politicians and the media, we need to be critically engaged in attempting to refine the polling of public opinion. Emergence from Covid-19, delivering the ‘levelling up’ agenda, and efforts to tackle the cost-of-living crisis, can only succeed if policymakers are accurately informed about how inequalities impact particular groups. Policy formed on the basis of public opinion data, therefore, needs to be examined, interrogated, and reported in all its intersectional diversity.
It is in recognition of this that the UK Statistics Authority launched the Statistics for the Public Good strategy in 2020, with one of its core principles being inclusive statistics. The aim is to ensure that “our statistics and our workforce reflect the experiences of everyone in our society so that everyone counts, and is counted, and no one is forgotten.”(our emphasis). The Inclusive Data Taskforce (IDTP) set up by the National Statistician reported in September 2021 and found that trust is a barrier to participation in data collection across all its consultation activities.
The IDTP marks a welcome start to representing the diversity and inequalities of the UK population. It is, however, crucial that in debates regarding levelling up, the evidence underpinning and informing policy is itself ‘levelled up’. Genuinely intersectional, inclusive polling that includes the views of all must therefore be a priority.
[1] ‘Minoritised ethnic’ (or the similar term ‘racially minoritised’) recognises that individuals have been minoritised through social processes of power and domination rather than just existing in distinct statistical minorities. It also reflects the fact that ethnic groups that are minorities in the UK are majorities in the global population. (Source: The Law Society)
[2] For example, https://docs.cdn.yougov.com/li2af8iulc/Internal_LevellingUp_merge_211214_W.pdf and https://www.centreforcities.org/blog/what-does-the-public-think-about-levelling-up/
The views and opinions expressed in this post are those of the author(s) and not necessarily those of the Bennett Institute for Public Policy.