Affirmative Action is damaging Civil Service Competence

An Englishman’s self-assurance is founded on his being a citizen of the best organised state in the world. - Tolstoy 1869

The Civil Service is a bit like a Rolls-Royce—you know it's the best machine in the world - Rab Butler Conservative Minister

In popular culture Britain’s permanent civil service has been imagined as a supremely competent organisation of intellectual men. This can be seen in fiction, with civil servants, like Sherlock’s brother Mycroft Holmes or Sir Humphery Appleby from Yes Minister, being defined by their wit. This impression lies in the historic use of tough and prestigious examinations to join the Civil Service. As such, it has employed the best and brightest. Despite being an excellent academic, John Maynard Keynes left his graduate study in Economics to apply to join the civil service. And even then, Keynes was second place in the exams forcing him to join the Colonial Office instead of His Majesty’s Treasury.

The origins of the modern civil service go back to the Northcote-Trevelyan report of 1859, commissioned by Gladstone’s liberal party to end corruption and improve competency. The report recommended that the civil servants be chosen by their success on examinations to select for ‘intelligence’, ‘fitness’, and to avoid the ‘evils of patronage’. However, the inspiration came from China, which used intelligence tests to pick scholars as bureaucrats from 605 to 1905. Britain's system of exams has been employed ever since the 19th Century and is seen in the current Civil Service Fast Stream. 

However, the classical liberal, meritocratic ideal of the civil service is being undermined. The Cabinet Office is continuously developing diversity policies in applications that make it easier for non-whites to become civil servants. But if you are selecting people for the colour of their skin, you are not selecting them for the content of their character - their merit. 

The first move was to create the Summer Diversity Internship in 2004. To get one of these internships you have to be non-white or poor or disabled. Crucially, diversity interns are allowed to skip examinations when applying for the full job. Although the legality of such a discriminatory and anti-white policy may have been tenuous in 2004, the Equalities Act 2010 legalised such racist hiring. In the mid 2010s the Government went further by removing screening by A level results and university attended. In 2014 all male shortlists for top civil service jobs were banned.

The big change currently underway is an accelerating move away from examinations which occurred after the 2016 ‘Socio-economic Diversity in the Fast Stream’ report by the Bridge Group. I should say the Bridge Group describes itself as a charity. However, the majority of its income comes from writing reports to organisations advising on how to improve diversity. As is the case with most consultancy work, a manager likely commissioned the report to justify and provide legitimacy to his diversity policies. Prior to the Bridge Group’s report, all applicants, except for diversity interns, had to undergo a numerical and verbal test. But the report recommended that diversity reforms:

‘should include exploring the removal of the online verbal and numerical reasoning tests in their current form or, at least, place increased weighting on the online competency questionnaire, and the situational judgement tests. These have been shown in the analysis to have a less adverse effect on candidates from under-represented groups, including lower SEB.’

In other words less weight should be put on measuring intelligence with examinations because poor and ethnic minority groups with low intelligence perform badly on these tests. It’s an accepted fact of psychology that some races tend to perform better on intelligence than others and, despite the taboo around the issue, in an anonymous survey, 85% of intelligence researchers agree that the black-white IQ gap is partially genetic. 

Unfortunately Bridge Group does not show their analysis so we’ll have to check it ourselves. Due to one of John Fuerst’s many Freedom of Information requests we have applicant exam results from verbal and numerical tests (if they had to do them) from 2018-19. The Cabinet Office says the data they provided are in percentiles, this is wrong since the results are far from uniformly distributed, but they appear to be some sort of results. I standardise results into IQs by assuming a mean IQ of 110 and standard deviation of 15. I find the mean by assuming only the top half of the normal distribution of IQ scores in Britain go to university and then only the top 90% get the minimum 2:2 result to apply. This gave me a mean IQ of 113, but my methodology was upward biased for mean IQ so I rounded down. 

As we can see, whites score much better than Africans on the numerical and verbal tests and slightly better than Asians. However, the Asians include the high IQ Chinese and the low IQ Asians from South Asia. Note how skewed the distributions are to the right and that there's a second ‘hump’ just below full marks. I’m afraid this is because the tests are online and without a time limit, meaning those that really want the job can get close to full marks. Also note how small the high scoring bump is for the Africans, not only did they perform poorly due to intelligence, most could not even be bothered to put in effort.

So what’s the Cabinet Office’s solution? Situational Judgement tests, which were implemented after 2016. These give candidates multiple choice questions on how to respond to problems of office life. The problem with these tests is that they are subjective with no clear right answer. Furthermore, these tests appear not to measure any differences in aptitude for the job. What matters for civil servants is that they can judge complex issues of policy design and implementation, not learnable office etiquette on when to ask colleagues for advice.

Example of a Civil Service SJT:

But that’s the real use of situational judgement tests - they don’t measure ability. Allowing unintelligent candidates to have a chance at getting the job. Racial differences on this test almost disappear. Whereas whites score 7.7 IQ points higher than blacks on the numerical reasoning, they only score 3.6 ‘IQ’ points higher on the Situational Judgement Test. Also note that there is no skew in these results despite the unlimited time allowed. No matter how hard you try getting a top score you can’t improve your results, it’s just luck.

Gillian Smith, then head of the civil service put it starkly in 2016:

‘In particular, the Bridge Group challenged us on two things, One was the cognitive loading of the initial online test – particularly the verbal and numerical reasoning tests, which we knew had some adverse impact on diversity. So we've taken those out for this year… Now the online test is a situation judgement report’ 

What’s so brilliant about this quote is that the woke bureaucrat accidentally rediscovered the notorious Spearman’s hypothesis - the better a test measures cognitive ability the larger the racial gaps in intelligence. One paper provides similar results - the g-loading of SJTs correlates 0.77 with the size of the black-white gap on the test.

Whilst Gillian Smith said she wanted to replace numerical and verbal tests with SJTs in 2016 the process is not yet complete. First SJTs were merely introduced as another way to judge candidates. Then in more recent years entire streams of the Civil Service Fast Stream have dropped the tests. In the 2019 data up to ⅓ of all applicants only did Situational Judgement Tests. Trying to find the facts on who is and isn’t doing the tests is extremely difficult. Currently on the government’s websites all civil service streams mention the SJTs, but in 2021 none of them mention the verbal or numerical tests. 

I reached out to applicants to see what was going on. I’ve been told that now only certain specialist streams such as Finance take the numerical test or verbal test. The most revealing information I received was from an insider. He said that many streams had dropped the tests and the ones that still had the test were putting less weight on them year after year. But why isn’t this information public and easy to find? Apparently there is such a taboo around the issue that recruiters are afraid to talk about what they are doing and why. Moreover, they are afraid that advertising the tests for specialist streams would put off minorities, or worse, publicising the practice could lead to criticism - not from the likes of me, but from others that will ask why tests were ever being used for which minorities performed poorly in. 

Of course, the devaluation of testing aptitude is happening across the world. Universities are removing the Graduate Record Examination (GRE) and, from what I can see, most graduate jobs in top areas, scuh as finance, are using situational judgement tests. The scandal appears to be everywhere, but no one has noticed. Even the British police are using situational judgement tests.

But what does it matter if organisations hire people less for the intelligence? Well many times over IQ has been found to robustly predict job performance, even in apparently cognitively easy manual labour tasks. IQ is associated with a 1% increase in income and in efficient labour markets, workers are paid according to their productivity. Thus 1 IQ point suggests a 1% change in productivity. If instead of taking the average white applicant, the average african applicant is chosen, then the Civil Service's productivity will be approximately 7% lower. This is bad. 

But don’t just take it from me, let’s see what the government has to say about the verbal and numerical tests:

“The Verbal Reasoning Test (VRT) and Numerical Reasoning Test (NRT) are measures of general mental ability, which is recognised as the strongest single source predictor of job performance at all seniority levels, across sectors of the workplace.”

As we can see, Situational Judgement Tests are being brought in across the public sector to decrease the number of white employees and increase the number of ethnic minorities. This blog is called Anglo Reaction for a reason. Our society is rejecting the classical liberal virtue of meritocracy and disgust at the ‘evils of patronage’ which serves political ends over effective administration. We need less of the Bridge Group report (2019) and more of Northcote-Trevelyan (1859). We need a government worthy of being operated by the likes of Lord John Maynard Keynes.

So what is to be done? First of all we need to know exactly which and how many candidates still get the numerical and verbal tests. And we need to know how these numbers are changing. If you are brave enough to use your real name you can send in a Freedom of Information Request. These important changes to the civil service also need to be publicised and discussed in the media. Brave backbench MPs should be putting questions about civil service recruitment to Ministers and justifications should be sought.

Lastly, it is not entirely impossible that civil service recruitment can be changed by judicial review. As is described in the Bridge Group report, the Constitutional Reform and Governance Act 2010 requires that the selection of civil servants must be “on merit on the basis of fair and open competition”. Where merit means the appointment of the best available person judged against the essential criteria for the role. If the government is eschewing what they have called the best measure of competency for the job (verbal and numeracy tests), they cannot possibly claim that the job is being “offered to the person (or people) who would do it best”. 

If the lawsuit is framed as protecting white and well-to-do people from discrimination it will lose. But the lawsuit could be framed as protecting the Asian applicants from a discriminatory practice. The lawsuit against Harvard’s discrimination took this approach and it had good optics in the media, even though it lost. The key advantage of such a victory is that it would be a blackmark against Situational Judgement Tests. If their use is found to be discriminatory and anti-meritocratic then other employers will shy away from their use.