Misusing statistics as an ‘end’ rather than a ‘means’

How we speak about equality, diversity and inclusion matters. The terminology we use to describe our identities. The use of words that include and exclude. History, power and privilege are tied-up in how we communicate with each other. The possibilities and limitations of language shape what we can change in our everyday realities.

Less discussed is how we speak about EDI data, particularly numerical data. We falsely assume that numbers, unlike words, are value-neutral.

Numbers never lie.

This is false. Most obviously, their calculation often depend on subjective decisions made along the way. Take the example of the gender pay gap (the percentage difference between average male and female earnings). When calculating the pay gap between male and female employees, which staff are counted: full-time, atypical, agency or all-staff? Is the data weighted? Does it include or exclude bonus pay? Are these mean or median figures? All of these decisions determine the number you reach at the end. Never take EDI data face value. Scratch the surface, check the workings and run your own calculations.

However, a darker (and more opaque) problem lies in the misuses of EDI data. When numbers no longer represent a reality but become the target and end in themselves. Rather than serving as an indicator of progress or measure of change over time, numerical data becomes the primary fixation and focus of the work – more so than the problem the data was intended to measure.

Take data on the disproportionately small number of women in senior leadership positions in certain sectors (such as Chief Executives of FTSE 100 companies). An increase in the proportion of female senior leaders is an indicator of change but it is not, assumedly, the end result. This depends on the reasons for diversifying senior leadership: to challenge patriarchal structures, make senior leadership more representative of the life experiences of staff or bring in fresh minds with different outlooks? None of these desired outcomes are apparent from a study of percentage point increases or decreases in isolation.

Rather than serving as an indicator of progress or measure of change over time, numerical data becomes the primary fixation and focus of the work – more so than the problem the data was intended to measure.

Consider UK Prime Ministers in the past 50 years: seven were men and two were women. A disproportionate reflection of UK society, which is generally split equally. However, the premierships of Margaret Thatcher and Theresa May were not/have not been indicative of advances for gender equality. A fixation with the numerical data in isolation, or without context, risks overlooking bigger questions of intersectionality (which types of women?) and institutional power (has their presence challenged or solidified existing structures?)

An unhealthy obsession with numerical EDI data leads to its misuses as an end rather than a means to an end. It also presents a dangerous trap for those working to advance EDI. Internal resources to bring about institutional change might be less forthcoming when numerical data alone makes it look as if things are getting better. Year-on-year percentage point increases, even if it is just one or two points, suck the oomph out of an activist and politically-charged agenda. EDI number crunchers: remain ready to disaggregate, ask people to explain their workings and question, question, question what an upward picture of ‘change over time’ really means for people’s lives experiences of EDI. Otherwise we risk EDI data being misused to mask bigger problems.

Dr Kevin Guyan is an equality and diversity researcher based in Edinburgh. He is writing in a personal capacity.