Race disparity: in data
By Rebecca Ghani and Anne Edimo
Data has the power to shine a light on inequalities in our society, and at a time when the world is fiercely protesting the killing of George Floyd, police brutality, and institutional racism, we highlight some statistics drawn from data showing how race affects outcomes in key areas, in the UK today
The killing of George Floyd brings racial discrimination into sharp focus, and pushes us to look not only at police brutality, but also beyond the tip of the iceberg, to the wider systematic inequalities that negatively affect life chances of black, Asian and minority ethnic (BAME) people, from cradle to grave.
While killings by police, death in custody and police brutality against black people are arguably the more visible and reported-on elements of racial inequality (although there is still variance in the perceived ‘newsworthiness’ of black deaths), there are countless less-visible, long-term, often-buried consequences of systemic racism in society. These impacts are felt from education and employment, through to the justice system and healthcare.
Data, and the statistics derived from that data, can help highlight these issues and can give researchers, the public and policymakers the impetus to act.
Deaths from Covid-19
The stark difference in Covid-19 deaths between BAME people and white British people was highlighted in a Public Health England (PHE) report published last week. It found that, in England, people of Bangladeshi ethnicity had around twice the risk of death than people of white British ethnicity. And people of Chinese, Indian, Pakistani, other Asian, Caribbean and other black ethnicity had between 10 and 50% higher risk of death when compared to their white British counterparts.
The report found that BAME people have a higher likelihood of contracting coronavirus due to contributing factors such as overcrowded households and living in urban and deprived areas. It also noted they are more likely to have jobs that expose them to higher risk.
As well as a higher likelihood of contracting the disease, the data shows that once BAME people acquire the infection, they are also more likely to die, partly explained by factors such as co-morbidities which are more common among certain ethnic groups.
The report does not further explore why BAME communities are disproportionately affected, and a Health Service Journal article disclosed that a key section of the review — which included information about the increased risk being linked to discrimination and life chances — was removed before publication.
The PHE report has also received criticism for omitting critical information about air quality. A European study found that long-term exposure to pollutants may be ‘one of the most important contributors to fatality caused by the COVID-19 virus in these regions and maybe across the whole world’; and a study by Imperial College London reported that ethnic minorities and deprived communities in England are the ‘hardest hit by air pollution’.
The Office for National Statistics (ONS) reports that men working in the lowest skilled occupations had the highest rate of Covid-19-related death, with 21.4 deaths per 100,000 males. ONS data also shows that the percentage of workers in the lowest skilled occupations was highest (16%) in the black ethnic groups.
In healthcare, citing UK and international data highlighting the increased risks, NHS England has advised that all NHS workers from BAME backgrounds should be risk assessed because the evidence suggests they may be at greater risk from Covid-19, and this advice is being echoed in the teaching professions and others.
And a Liberty study found that, under coronavirus laws, BAME people are more likely to be fined than white people, with the charity stating: ‘People of colour are paying the price of arbitrary policing.’
Death in custody
Death in custody is also an area where there is overrepresentation in the BAME population. ‘Death in custody’ includes death while under arrest in a police station, while being detained for the purposes of a search, or following any other ‘contact with the police where there may be a link between the contact and the death’.
According to figures from the Independent Office for Police Conduct, in the category ‘deaths in or following police custody’ from 2004/05 to 2018/19, black deaths (excluding other minorities) in custody were overrepresented in the number of deaths (8%), compared to the national demographic breakdown (3.3% at the last census).
The charity Inquest echoes this, with data showing similar risk factors for all BAME people while in police custody — in cases where there is restraint, use of force, or where there are mental health-related issues.
The Lammy Review, published in 2017, found that although BAME people make up 14% of the overall population, they comprise 25% of the prison population. It also reports that over 40% of young people in custody are from BAME backgrounds.
Death in childbirth
Data from MBRRACE-UK (Mothers and Babies: Reducing Risk through Audits and Confidential Enquiries across the UK) found that black women are five times more likely to die in childbirth than white women. The data shows the number of women who died during or up to six weeks after pregnancy, from causes associated with their pregnancy, from 2015 to 2017.
It reports that there were 38 deaths per 100,000 in the black ethnic group, compared with seven per 100,000 in the white ethnic group.
‘At present we don’t have enough information to explain the complex and multiple factors at play. Understanding these disparities needs urgent research and action,’ the report states.
This issue was explored further on Radio 4’s Woman’s Hour: Black women are five times more likely to die in pregnancy, birth or postpartum than white women. The programme explains that the risk has been increasing year on year, and the discussion includes interviews with black women about their experiences of giving birth.
—
These are just some areas where data highlights systemic inequalities and discrimination. This data, and data like it, should be published openly, in accessible formats, to allow the public, researchers and policymakers to examine the evidence, expose the inequalities and push for action on these issues.
Is the data itself a problem, and does AI help or hinder?
While these datasets and statistics highlight the inequalities in our society, and their destructive outcomes, is the data itself also a problem? There are inherent intended and unintended biases in data collection, treatment and use which can lead to incomplete and skewed data and unfair outcomes in areas ranging from car insurance through to planning applications.
And it seems AI and algorithms are susceptible to the same biases. The Ada Lovelace Institute’s Director, Carly Kind has warned against the use of algorithms when the data that feeds them is inherently unrepresentative. In an interview for Raconteur, she said that potential biases in facial and voice recognition technologies can originate from unrepresentative datasets and this can occur when there ‘isn’t a proper representation of ethnicities, genders or social classes.’ The Open Data Institute has also warned about the risks of automated decision making.
In July 2019, an interim report produced by the Centre for Data Ethics and Innovation established that the rise of AI-enabled technology in recruitment ‘was predicted to be a growing trend over the next few years’. With the ongoing Covid-19 pandemic forcing all sectors to adapt, HR professionals have inevitably had to rely on algorithmic recruitment tools more than ever. Although some experts claim that AI has the potential to remove biases from recruitment and help to increase diversity, others believe it does the complete opposite. More work needs to be done to ensure decision algorithms are regulated, so that fairness and accuracy in the use of AI-enabled technology are guaranteed.
So while the datasets and associated statistics shine a light on inequality, data practices also need to be examined and scrutinised, to help ensure that the data itself is representative, unbiased, comprehensive and trustworthy.
