In the days when I promoted Mosaic, our team used to share what we thought was an amusing anecdote. Marketers were always intrigued to know which Mosaic category we categorised them by. When we told them, it was common for them not to be perfectly happy about our accuracy of them, but they invariably told us it was an uncannily accurate description of their neighbours.
I believe that it is this anxiety about being classified by others that lies at the heart of the current uproar over the allegations surrounding Cambridge Analytica. Had these just been that the company obtained Facebook data improperly and used it without permission, the issue might have covered in the business section. The allegation that the data was used to develop emotional profiles is what made it become a headline issue. Only the data “subject” is entitled to describe him or herself.
Such an assumption is well worth exploring, particular in relation to issues which are very much front of mind with politicians and social commentators, efforts to stamp out discrimination on the basis of race and debate about the entitlements of transsexuals.
In the early days of immigration to Britain, it was a person’s physical features that were at the root of discrimination. The native population was not used to engaging people whose skin, hair and facial features were different from their own. Minorities that were most subject to discrimination were defined by colour. They still are.
In the Internet era the feature that most identifies a person from a minority group is no longer their physical appearance but their name. When checking CVs, recruiters don’t discriminate on the basis of people’s self-identification. They seldom see a photograph of potential candidates. What they see is their name. The same applies to customer service centres and any customer-facing staff who deal with customers over the Internet, mail or telephone. What other people think you are is a more potent force for discrimination than what you think you are.
This is relevant to our business since we use people’s names to help organisations identify the effectiveness with which they cater for the needs of minority groups. Which are the groups that a theatre venue has most difficulty attracting to its productions? Which groups are most or least likely to become members of a charity?
Quite understandably our clients’ business analysts ask us how accurately names match with people’s self-identification. This is on the assumption that self-identification represents the gold standard of accuracy.
It is not just in relation to ethnic background or religion that questions arise as to who is entitled to determine which category is apply to the data subject.
Curiously, although many of us when asked to describe ourselves would use terms such as “middle class” or “working class” – and that class is such a key component of our identity – few have any problem with our social class being inferred from the occupation that we give when completing a Census form.
As for disability, such are the entitlements to be gained from being registered as disabled, that this categorisation requires the State to judge us using a complex set of pre-determined formula. By contrast to age, which we are happy to have inferred from our date of birth, though, if asked, many of us feel and behave much younger or older than our actual age, there is a general presumption that we should be entitled to specify by which gender we wish to be classified.
But our gender is in many ways like our ethnicity – we have a gender with which we self-identify but this may or may not correspond to the gender by which we are socially perceived. This is why it is not entirely reasonable for men to demand the right of entry to the women-only swimming pool on Hampstead Heath or to demand, on conviction, to be remanded to a women’s-only jail.
Thus, in many respects, it might be more reliable for us to possess multiple forms of identity even within a single domain. The ethnic group I self-identity with may be white British, to others, on account of my name, I may be considered Turkish; to my medical practice especially in an era of personalised medicine, it may be the country of my ancestry that is important; to the dietician it may be the cultural background of the person who does the shopping in the household to which I belong.
Ethnicity and religion are not the only fields to which multiple forms of categorisation may be appropriate. For older people, their life expectancy might be a better indication of their “age” than the years since they were born. In respect of social status, the type of postcode in which a person lives may be a more relevant measure than the occupation of the head of their household.
So what do we conclude from all this? I would argue that as professionals it is easy for us to mistake the system of measurement for the characteristic we are trying to measure. Age, social class, self-declared ethnicity, gender, disability – the methods we use to measure them are only approximations of the characteristic whose behaviour we are trying to measure.
We need to consider the context in which different forms of classification are appropriate and be open to new ways of measurement that become possible and, indeed, necessary as a result of technological change.
Professor Richard Webber is co-founder of Webber Phillips