Professor Jon Deeks on what happens when a test is used for a purpose for which it wasn’t designed

Transcript of Radio 4 Today interview on 22 December 2020.

Radio 4 Today presenter Justin Webb: It is 7 minutes to 7. We’re promised Covid tests in schools in the new year — lateral flow tests available very cheaply and simply and easily across the board. That is why the government wants quite a few children to go back later than normal to get that system up and running. But there are concerns about the quality of those tests. We’ve heard them on the programme before, and those concerns appear to be intensifying after further work — further work in particular from Professor Jon Deeks from the University of Birmingham’s Institute for Applied Health Science. And he is on the line. Good morning to you again, professor. …

Why I didn’t — and Nir Eyal thought I should — sign the open letter calling for this

Image for post
Image for post

Yesterday, 1 Day Sooner posted this Open Letter calling for human challenge vaccine trials for Covid-19. As they describe such trials:

Human challenge trials deliberately expose participants to infection, in order to study diseases and test vaccines or treatments. They have been used for influenza, malaria, typhoid, dengue fever, and cholera. Researchers are exploring whether human challenge trials could speed up the development of a vaccine for COVID-19, saving thousands or even millions of lives.

Below is an email in which I explain why I declined to sign this letter. Further below is an exchange with Nir Eyal, who is one of the initial signers of the letter, to which my email gave rise. …

A study just published in Nature reveals the following: even for someone with no underlying health conditions, the increased risk associated with being 45 years of age, rather than 30, is greater than the increased risk associated with various health conditions the NHS deems sufficient to render a person clinically vulnerable to serious illness from Covid-19.

I. Quantifying the risks to the clinically vulnerable

“Coronavirus (COVID-19) can make anyone seriously ill. But for some people, the risk is higher”, according to an NHS webpage entitled “Who’s at higher risk from coronavirus”. The NHS lists the following people as at “moderate risk (clinically vulnerable)”:

Image for post
Image for post
“Who’s at higher risk from coronavirus”

The aforementioned study in Nature — which is entitled “OpenSAFELY: factors associated with COVID-19 death in 17 million patients” — quantifies the risks associated with most of the above health conditions. It finds that, when one adjusts to control for age, gender, level of income deprivation, and other health conditions, the listed NHS conditions are associated with increases in one’s risk of death from Covid-19 by the following factors (see righthand column of Table 2 on p. …

[In an earlier version of this blog, I mistakenly claimed that the UCU Left statement maintains that “UCU Left members of NEC voted against” the levy. It has been drawn to my attention that this sentence refers to the cutting of strike pay, not the levy. I apologise for this mistake and for my inference based on it that the quoted statement was misleading and dishonest.]

UCU Left has issued this linked statement strongly condemning the £15 levy to replenish the strike fund. …

A response to USS’s claim that it is

As discussed in a previous blog post, UCU and UUK have asked USS to model “an option involving a higher-return (and higher-risk) investment strategy”, in which the scheme would remain continually invested for the long run in a portfolio which is weighted roughly 65% in “growth assets” (equities and property). In that earlier post, I responded to USS’s objection that the “short-term risk” of such an approach was too high.

In this post, I respond to USS’s distinct objection that such an approach is intergenerationally unfair:

We believe that this option does not align well with Principle 3 [regarding intergenerational fairness]…. This is because in effect more risk is being taken to meet pensioner liabilities and, if that risk materialises, the cost increase would be split between employers and active members under the cost sharing rules in the absence of an alternative JNC decision. This is challenging in terms of intergenerational fairness. (‘Methodology and risk appetite for the 2020 valuation’, March 2020, p. …

[Update 9 September 2020: Here’s a brief Twitter thread in which I raise a similar, but simpler, objection to USS’s latest version of its short term risk metric.]

The joint UCU/UUK Valuation Methodology Discussion Forum has asked USS to model “an option involving a higher-return (and higher-risk) investment strategy”, in which the scheme would remain continually invested for the long run in a manner benchmarked against a “reference portfolio” weighted roughly 65% towards “growth assets” (equities and property). This is what is known as the “no de-risking” option, since it would involve no systematic shift of the portfolio over time from growth assets into bonds and other “liability matching” assets. (‘Methodology and risk appetite for the 2020 valuation’, March 2020, pp. …

Two questions to USS for their video on the 2020 valuation

In response to this invitation, I would like to submit two questions to which I hope USS will provide replies in a video they release on Thursday the 9th of April. The first question is about USS’s measure of short-term risk and the second about their measure of long-term risk:

Q1. Why should it matter how expensive it might become to purchase a self-sufficiency portfolio within a year, given that it would make no sense to purchase one so soon even if it were possible to shift the assets so swiftly?

Q2. What is USS’s justification for measuring the self-sufficiency deficit in 20 years’ time on the assumption that the scheme will be precisely 100% funded at that time (i.e., assets = technical provisions), rather than in any surplus that is regarded as 67% likely — where 67% reflects the level of prudence, above and beyond a 50% best estimate, that USS builds into their expected returns on assets in setting the discount rate for the technical provisions?

Some recommendations

If you’re a UCU member, please have a look at the following thread on Twitter:

…and then click through to these recommendations for the NEC ballot:

No. This figure is based on an error in arithmetic.

According to the material accompanying UCU’s pay and equalities strike ballot, “The value of university salaries has fallen against inflation by over 20% since 2009”. The union’s General Secretary recently reiterated that “We believe the true decline over the past decade is over 20%….” The grounds for this claim are spelled out here:

The union says that staff pay has actually fallen by around 20% in the last decade as pay awards in higher education have resulted in a cumulative increase of 11%. In the same time period, the RPI index has increased by 31.8%. Meaning staff in higher education have seen the value of their pay decline by 20.8% …


Michael Otsuka

Professor, Dept. of Philosophy, Logic & Scientific Method, LSE

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