This essay, on the ways we categorise people, was my Observer column this week. (The column included also a short tribute to Paulette Wilson.) It was published on 26 July 2020, under the headline ‘Don’t call me BAME. We need a new political language.’

‘I hate being described’ as one, tweeted Channel 4’s Krishnan Guru-Murthy. Sunder Katwala, founder of the thinktank British Future, doesn’t identify as one. And last week Coventry’s Belgrade theatre promised never to use the description again.

What they are all talking about is ‘BAME’, that soulless acronym to describe black, Asian and minority ethnic people that is creeping into fashion in policy circles and journalism. It has been around for a while, but the Covid-19 pandemic and the discussions around Black Lives Matter protests have given the term a higher visibility. The majority of people, though, have no idea what it means. And most people who might be described as BAME loathe the term.

The problem is not simply that BAME is a clunky description. The debate also points to deeper questions about how people are categorised and what such categories tell us.

It is impossible to analyse society or the impact of social policies without distinguishing between categories of people. In France, the nation’s universalist ethos means that ethnic and religious data is rarely collected. The resolve to treat everyone as citizens, not as bearers of specific racial or cultural histories, is valuable. In practice, though, many people are denied equal treatment and racism is deep seated, but the lack of data makes it difficult to gauge discriminatory practices.

In Britain, there has been much debate about the unequal impact of Covid-19 on ethnic minorities. Not so in France, largely because there’s little information in which to root a debate. But if lack of statistics is a problem, possessing data is no panacea. Decisions on what data to collect and how to interpret it may themselves mislead.

Take the question of street violence. Twenty years ago, the major issue of concern was not knife crime but street robbery. Then, as now, black people were over-represented in the statistics, leading to the claim that there is something about black culture leading to criminality.

The criminologists Marian FitzGerald, Jan Stockdale and Chris Hale analysed the data. They showed that street crime was much more likely in areas with a high population turnover and a combination of young people living in poverty alongside others who were both more affluent and trendy enough to own gadgets such as mobile phones. Young black people lived disproportionately in such areas. But where such areas included large numbers of poor white people, they, too, were involved in robberies.

The category ‘lives in an area of high population turnover with a mixture of poor people and affluent trendies’ is not politically salient. ‘Black’ is. So street robbery became associated with black people. The result is what FitzGerald calls ‘statistical racism’.

People belong to many categories and categories overlap. African Caribbeans and Bangladeshis in Britain, for instance, are disproportionately working class, compared not just with white people but with other minority groups, such as Indians, Chinese and black Africans. But while discussion of the white population routinely takes class into account, discussions of minorities rarely do.

Consider school exclusions. Black pupils are disproportionately excluded from school. Look more closely and you see the problem is in particular with those of Caribbean descent. Pupils of black African descent are less likely to be excluded than their white peers. Figures also show that pupils claiming free school meals (FSM) – a proxy for poverty – are three times more likely to be excluded than the average pupil; 40% of all school exclusions are of FSM pupils.

School exclusion, then, is a major issue facing white working-class pupils, too, and class as well as race may play a role in the disproportionate exclusion of black pupils. But to say so is to invite the accusation that one is downplaying the significance of racism. And so, more nuanced accounts of discrimination are often ignored.

For analytical reasons we often need to disaggregate groups. The categories we use for analysis would depend on the questions we want answered. In a political context, however, we need more inclusive categories defined by aims and values rather than by narrow identities.

When I was growing up, I saw myself, and was seen as, ‘black’. In the 1980s, it was a political term, denoting a sense of a common struggle against racism. Over time, as identities have become less political, more ethnic or cultural, so the meaning of ‘black’ has changed.

Some see BAME as a means of articulating that feeling of commonality that ‘black’ once denoted. It’s not. It’s an administrative, not a political, term. What we still lack is a political language that can both encompass the varied experiences of particular groups and imbue a sense of solidarity to struggles for social change.