The backbone of Diversity Maths, this is the tool you use to assess a company’s DEI recruitment programme, and how it compares to a meritocracy.
It works by extending a technique used in DEI recruitment. DEI uses the assumption that recruiters are racist/sexist/trans-gayphobic by requiring that all job applications are stripped of all characteristics that might identify the race, etc, of the applicant.
Let’s start by applying this to, say, a high-profile professional firm recruiting graduates, and the characteristic is being BAME.
While the UK BAME population is around 13%, I calculate our top graduate population is about 24% BAME, based on recent graduate statistics. We can test for other assumptions later.
The process is now to TOTALLY anonymise all applications. So, if we selected, say, 100 recruits from a large population of applicants, what would we expect to see?
Intuitively, we would believe that having 24 BAME offers was the least unlikely result. We would also believe that, the further our result was away from 24, the less likely the result would be.
What intuition does not tell us is how rapidly the likelihood falls the more we move away from 24. We find that the results of 14, or 34, and anything outside those figures are so unlikely few people would believe them.
In fact, anything further than FIVE percent from 24 is a warning sign. Especially if such a pattern is repeated in consecutive years.
This is shown in the following graph:
Barring the emergence of “super candidates”, it shows the likelihood that recruitment has been based on race rather than merit. This will have several negative consequences:
- The company breaks employment law. However, any significant sanction is currently unlikely.
- The company attracts adverse publicity. However, some think that such practices are OK, as long as the racism is ”in the right direction”.
- Potential quality white recruits are deterred when “word gets around”
- Current staff lose morale when they realise recruitment/promotion are not for them.
- Race based recruits are likely to face silent ostracisation and opprobrium if their performance is not up to par. If they have not been selected on merit, this is likely to the degree that race has trumped merit. This will hit staff turnover.
- Some failed white recruits will be smart enough to work out the probabilities described above, and issue Employment Tribunal claims. Large organisations, desperate not to have their DEI regimes examined, will almost certainly buy these off with Non-Disclosure Agreements thrown in at a price. The extent of this practice is unknown – no sane company would ever admit it ever happened.
Even if a company is determined to continue such practices, knowledge of the Curve of DEIth will be valuable for internal assessment. Even if HR departments ignore it, Finance Directors will be interested in the credibility of figures appearing in, eg, the annual Report and Accounts, even if they are not “on their watch” – ie, not audited.
A technical description of the Curve of DEIth follows in the Appendix. However, the easiest practical way to utilise it is on a simple spreadsheet. Once this is set up, an hour’s tuition should mean that even non mathematicians can assess their company’s DEI recruitment programmes. It is then a matter for senior management – risks can be assessed and rational decisions made.
Although UK regulators have currently put Ethnic DEI reporting “on hold”, many companies are doing this voluntarily — so there is peer pressure to produce these figures. Also, as gender recruitment is often disclosed as part of (compulsory) gender pay gap disclosure, the Curve of DEIth can be useful here – we just change our “24” to a population gender split.
Appendix (Mathematical)
What are the chances of recruiting 34 or more from a large 24% population?
Chance of recruiting exactly 34 is A/B, where
A= (0.24)34x(1-0.24)(100-34)
and
B –Fact(100)/((Fact(34)xFact(100-34))
Where Fact (n) = nx(n-1)x(n-2)x(n-3)x…….x3x2x1
To get our answer, we do this calculation for 34, 35, 36, etc, up to 100, and add all the numbers together.
The results here are as follows:
34 or more BAME recruits – 1.5% chance, or one in 65.
14 or fewer BAME recruits – 1.0% chance, or one in 100.
Both are pretty unlikely – “Beyond reasonable doubt” perhaps?
(Note, the curve is not symmetric – this is correct – but it is pretty close to symmetric).