Brexit Talk
Published in

Brexit Talk

Racism and the EU referendum

Did the EU referendum show the ugly face of the UK underground?

A question on Quora sparked off this thought. Can we crunch some numbers and look for indicators? Is there s correlation between Leave areas and the highest numbers of racist incidents? Seems like a nature assumption. So crunch we did.

Quora: Is the UK more or less racially divided since the June 23rd referendum ?

OK, so I’ve copied this here because I am going to get all nuanced on that question.

Racial division in this question has not been defined. This could mean geographical division, in which case nothing statistically significant has changed at all.

It could also mean that hate crimes, which make up the courser grained category racism falls in, has increased, which it definitely has. Or it could be racist incidents specifically, which it definitely has! I’ll leave the figures for perusal below this narrative.

What does this tell us?

Well, for rare some areas, the financial year 2015–2016 did not provide a maximal increase in racism over time. A couple of years ago, some areas had higher growths in racism than they did in 2016.

However, a positive percentage change is a growth in value in any event, and this does not correlate with any increase or decrease in people living in those areas (obtained from census estimates). This means there is a statistically significant, disproportionate rise in the number of recorded incidents of racial abuse in England and Wales. In any event, 2016 saw a maximal rise:

  • North West England (Leave, 53.65% — Oddly, even Manchester saw a 22% rise despite voting to remain)
  • North East England (Leave, 58.04%)
  • Yorkshire and the Humber (Leave, 57.71%)
  • South East England (Leave, 51.78%)
  • Wales (Leave, 52.53%)

Anyone see anything correlating with that? That’s right, all Leave areas of the UK! Indeed, when we take a correlation coefficient (how much Leave voting relates to maximal increases in racism, it comes out at 0.9313 — the maximum you can get is 1.0 — This means areas voting Leave was almost certainly positively correlated with an increase in racist incidents — it’s unlikely to get much stronger than that).

Now, a lot of Brexiteers (read all that I have tried to explain this to) won’t understand this next bit or refuse to. After all, it unequivocally links Leave voting with racism, but numbers don’t lie, people do. There is one single outlier. An area which had a drop in racism, yet voted Leave. Noting that all 12 regions regions bar Northern Ireland, Scotland and London voted to Leave the EU, and the increase or otherwise in racist incidents followed the trends or changed by an insignificant amount. Yet, the East of England, which voted to Leave, saw a drop in racist incidents of 12% in the year to April 2016. What could be going on here?

This looks great until you check the data more closely. As you can see above, it had a massive increase (a spike) the year before of 20% when all other years were low single digits. Something took hold that year and then disappeared this year. Not clear what. Could it be the BNP or UKIP were more active? Total speculation though. If anyone in Lincolnshire and surrounding areas can shed some light, I would be interested.

Here are the percentage change figures for UK regions as a whole stretching back to 2009/10.

North East

North West

Yorkshire and the Humber

East Midlands

West Midlands

East of England


South West


Which aggregate to the full UK racism trend of.

Police Figures — Current

It is dangerous to ignore police figures! One of the worst pieces of advice anyone can give anyone is to ignore police figures. Police figures give the experiences of people and it is almost always bigger than those numbers suggest. Many folk don’t come forward. So you don’t get 100% engagement. This is the same with many violent crimes too, including rape. Many women fear coming forward (a story for another day).

Let’s also not forget the police are a public service who have to make less money than ever stretch. Plus, we have to see what we are getting for our money.

There was a spike after 23rd June 2016 which saw an increase in the recorded level of racist hate crimes of 57%. This required reporting it to the police and it being booked. It causes a sustained year-to-date increase of 14% over the year June to September (note, this is just 3 months and it beat the previous year by 14%). This was a because of a “since June” increase settling at 44%. Unfortunately, those who have told you that it returned to close to normal levels are a little misguided, as the figures to date can’t reduce. It would have to undo/delete those previous racist incidents, which is not possible. When it’s recorded, it’s recorded.

Note, until prosecutions and the year have completed, it is not possible to say if any more people have been convicted of it.

What do these rises really mean? An example is probably needed to fully appreciate it. I am happy to provide this as it is a useful analogy and do so with no animosity. Not everyone is good at math.

To make the math simple, let’s say there are 1,000 such crimes each year, split 250 incidents each quarter, but growing at 7% a year (extrapolated down per quarter).

A spike of 57% in a year-on-year statistic actually results in 1,570 crimes of this nature appearing in an equivalent 12 months period including a spike in the quarter containing the vote. This really means that the “latest” quarter did not see 393 incidents (157% of 250 incidents or so) but an increase in 12 month figures, applied in one quarter. ie we see 57% of the 1,000 incidents (570 incidents). So 250 + 570 = 820 incidents.

This is why it looks like some of the figures suggested 400% or so. It’s a people/journalistic interpretation which compares apples with oranges. And I say that as a categorical “leftie” who would like that magnitude to be true :) It is simply not.

However, there is one more thing. The pertinent thing on this topic. Once those figures have been added, we need to see a negative number, equal in magnitude in the next quarter (just because 6 months have passed — two quarters. It can be longer in a long tail) to revoke that initial spike. That didn’t happen. The growth rates started to return towards “normal”, which is not the same thing.

For our example, we are now at 1,570 incidents. If it grows even 7%, what you’re then seeing is that “spike” push back through the 7% coefficient, resulting a higher number than even that!

For example, 1,570 x 1.07 (7%) = 1,680 incidents at the 12 month point. This is then higher than the 1,000 incidents of last year by 68%! Note how the compounding works (it’s like Net Present Value of mortgage repayments with a monthly variable interest rate — same concept). There is a difference between the growth rate (which is a concept of calculus) and the actual snapshot value (a concept in counting). It requires different “brains” to think that one out and most people fall into the latter over the former. And this is using a 7% rate, not the 14% consistent rate it actually has become.

On a personal note, anecdotally, I suddenly found myself subject to 3 incidents in as many weeks, having not been subject to any in 30 years and the Police have done little about it.


Like this article? Then please hit-the-heart to reach more people with it.



Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
Ethar Alali

Ethar Alali

EA, Stats, Math & Code into a fizz of a biz or two. Founder: Automedi & Axelisys. Proud Manc. Citizen of the World. I’ve been busy