Immigrant selection and crime in Britain
Do less-selected groups have higher incarceration rates?
Written by Noah Carl.
The number of Albanians residing in British prisons was in the news last year. According to the latest figures, they comprise about 1.5% of all inmates, despite being only 0.05% of the population – which means they’re overrepresented by a factor of 30. And in fact, the British and Albanian governments recently made a deal under which hundreds of Albanian convicts will be sent home to serve out the rest of their sentences.
One Albanian who won’t be sent home is a man named Gjelosh Kolicaj. The Home Office had sought to deport him after he was jailed for smuggling £8 million of his gang’s profits out of the UK in suitcases. Yet British judges blocked Kolicaj’s deportation on the grounds that the Home Office had “failed to take sufficient account of his human rights”.
What could possibly account for such a large degree of overrepresentation? The answer of course is immigrant selection: a disproportionate number of the Albanians who’ve come to the UK in recent years have been criminals or people with a predisposition for crime.
Indeed, if we look at homicide rates (the most internationally comparable crime statistics), Albania’s is only 1.7 per 100,000 – putting it below Estonia, Israel, New Zealand, Canada and the US. Of course, Albania may have higher rates of other kinds of crime like theft or fraud. But you still wouldn’t expect Albanians to be overrepresented in British prisons by a factor of 30.
Albanians are just one group. To study the role of selection in immigrant crime more generally, we can use the IAB Brain Drain dataset. This provides estimates of the number of immigrants from each country of origin who are low, middle and high-skilled, respectively. If selection matters, we’d expect groups with more low-skilled immigrants to have higher incarceration rates – which we calculate as the number of prisoners in each group divided by the total population of each group in Britain.
Since the brain drain data are from 2010, I took prison populations and total population estimates for 2012 (the earliest year available). Note that the prison and population data correspond to England and Wales, while the brain drain data corresponds to the whole UK. But this shouldn’t matter as England and Wales comprises 88% of the country’s population and an even greater share of its immigrants. Data were available for 136 nationality groups.
12 nationality groups for which population data were available were not represented in the prison population dataset. Hence I assumed they had incarceration rates of zero. Excluding these groups from the analysis did not substantially affect the results.
The average incarceration rate across 136 groups was 0.42%. The five groups with the highest incarceration rates were Congolese (4%), Vietnamese (2.3%), Sudanese (2.1%), Albanians (1.9%) and Jamaicans (1.6%). Note that Albanians’ current incarceration rate is around 4%.1

The chart above plots incarceration rate against percentage low-skilled. There is a moderately strong positive association (r = .29, p < 0.001), with the points being clustered below the diagonal. Most groups have relatively low incarceration rates, but some have much higher rates – and these tend to be ones with a large percentage of low-skilled migrants.

Interestingly, national IQ taken from Lynn and Becker has a somewhat stronger relationship with incarceration rate (r = –.37, p < 0.001), as shown in the chart above. The pattern is much the same, only reversed left-to-right.2
When I entered the two predictor variables into a linear regression model of incarceration rate, they both reached statistical significance, though national IQ had a larger coefficient (β = –.31 versus β = .20). And when I Winsorized incarceration rate at the value for the second highest group (thereby limiting the influence of Congolese) and re-ran the model, the coefficient on percentage low-skilled was slightly larger (β = .24).
These results suggests that the two predictor variables’ effects are partly independent. Groups with more low-skilled immigrants have higher incarceration rates, as do those with lower average IQs.3
Another predictor variable that presumably matters here is age. Groups with more young men will have higher incarceration rates. Unfortunately, I was unable to obtain the relevant data.
It’s worth noting that percentage low-skilled is a noisy measure of immigrant selection, given that qualifications from different countries aren’t always comparable and there’s a lot of variation even within skill levels (two people with a university degree can have quite different IQs, for example).
Similarly, the incarceration rate is a noisy measure of criminality insofar as incarceration itself is a rare event. For 48 out of 136 groups included in the analysis there were fewer than 10 prisoners, and for 30 of the groups there were fewer than 5 – leading to substantial sampling error.
One final caveat is that the data on immigrant selection were for country-of-birth groups, whereas the data on incarceration rates were for nationality groups. While immigrants’ country of birth and nationality are usually the same, they aren’t always.
Despite these limitations, immigrant selection helps to explain differences in incarceration rates across country-of-origin groups in Britain, even when controlling for national IQ. This reinforces my claim that it is meaningless to talk about the impact of immigration per se.
Noah Carl is an Editor at Aporia.
Consider supporting Aporia with a paid subscription:
To chat with fellow Aporia readers and attend meet-ups, join our Telegram. You can also follow us on Twitter.
This is when using the number of Albanians in England in December 2019 (i.e., before Covid). If the number of Albanians in June 2021 (i.e., after Covid) is used, their incarceration rate is around 6%.
The Spearman correlation between incarceration rate and percentage low-skilled was ρ = .33 (p < 0.001), while the Spearman correlation between incarceration rate and national IQ was ρ = –.41 (p < 0.001).
Together, the two variables explained 17–19% of the variance in incarceration rate.
> Yet British judges blocked Kolicaj’s deportation on the grounds that the Home Office had “failed to take sufficient account of his human rights”.....could you get a more poetic line on just how far the British establishment has disappeared up its own backside. And yet anyone who is not a politics obssessive or a Substacker will know little of such things because their take on the 'issues' facing their country will be some sort of BBC-type Sunak or Starmer blah blah. Weep for us.
Thanks for an interesting article.
Immigrants should not be selected based on nationality but on desired traits.