Can we trust self-administered tests?

It would be easier and cheaper to have mass testing at home but is it reliable?

Welcome to Plugging the Gap (my email newsletter about Covid-19 and its economics). In case you don’t know me, I’m an economist and professor at the University of Toronto. I have written lots of books including, most recently, on Covid-19. You can follow me on Twitter (@joshgans) or subscribe to this email newsletter here.


If we want to suppress the pandemic within the next year, widespread and frequent rapid testing is our best option. There are many logistical issues associated with doing this but one looms large: who will administer the tests?

At the moment, pretty much all tests, even rapid antigen ones, are approved by regulators only if they are administered by a health professional. Even under the most generous interpretation of that term, it clearly means one thing: not you. That pretty much rules out the option of administering tests at home. Instead, you end up with a model that relies on testing as you enter specific places like offices, schools and such. That creates logistical challenges as those places have entryways that are not set up for the type of processing let alone where to put people in the meantime. This is one of the reasons I explored testing after you had entered rather than before. But regardless of how you cut it, once you need to have a specialised person administer the test, it is going to take time for each individual. Let’s imagine that just administering the test required an individual to be physically present for at least 5 minutes. Then in a 1,000 person facility testing each person twice per week, that is, 400 people per day. To get that done, in say, three hours (with staggered entry), would take 11 health professionals. Have you ever seen a security lane at an airport with 11 lanes? You start to appreciate that getting through that in under an hour is fairly heroic. It is going to take considerable innovation to get this down to levels people can easily live with.

For this reason, it has been suggested that testing should be administered at home. What this means is that you turn your 1,000 people into testers even though the total time to self-administer a test is likely to be the same as other methods. What you get is convenience.

Why isn’t this the obvious way to go? The answer is trust. Can we rely on people to properly self-administer a screen and then do the right thing afterwards? We already face this challenge with symptom checking and have chosen to test self-administration there. Ironically, we do this by asking people to fill out a form (on paper or an app) confirming they have checked their symptoms. Did they actually do it? Nobody knows for sure. But they did do the paperwork. In reality, it is a nudge to spend a few seconds thinking about Covid symptoms before heading out. Then if you do seem to have symptoms, to think about whether it is worth going out or if anyone will notice them once you arrive.

What is the difference if someone were to self-administer a Covid test? One issue is that you might perform the test incorrectly. It turns out the main risk there is not getting a false positive (see this recent paper) but, instead, a false negative. You might not pick (literally) enough up of the right type of sample to detect the virus. That means you might confidently act thinking you are negative and pose a risk to others. That said, you are still at risk and the activities you do to protect others are not dissimilar from the ones to protect yourself so there is reason to believe we won’t be empowering spreaders unknowingly. (I should add here that there is a George Costanza issue in that you may claim you are positive even if you test negative to get out of going to work or school. But the risk is you can only claim that once so what happens if you actually get Covid-19?)

The larger issues seem to me when you test positive. Just like a symptom check, that is when you face a dilemma. But unlike a symptom check, there are no excuses (i.e., it could be the cold or the flu) and you have Covid-19 and should isolate and/or receive a confirmatory PCR test and a visit to a medical professional to work out next steps. So let’s do the economist thing here and weigh up that individual quandary.

  • Need for treatment: Covid-19 is a scary disease. You want to make sure you get ahead of it and if you are testing frequently that means you are likely in the early stages and can get ahead of it. If you get a positive test, you have an incentive to go through the official process.

  • Protecting others: If you are positive, going out there will potentially infect others. If you care about them (or the consequences associated with not caring about them), then you will want to isolate at the very least. (This is the route emphasised in a recent paper by Thomas Hellmann and Veikko Thiele.)

  • Protecting yourself: If you have Covid-19, you can’t get more of Covid-19 (at least not any time soon). So if you were holding back your activity because you were worried about getting it, those worries are now over. In that case, there is an incentive to party on.

Putting it this way, the trust issues come down to a belief that the first two options will be outweighed by the last one. The first option is a private motive. The second is an altruistic one. While the third is a divergence between private motives and social consequences — in other words, the failure to internalise an externality. Not surprisingly, it is the one many economists worry about.

Here’s what is interesting: we kind of know something about the people who might weigh the third reason above the first two in their decision-making — they are the people who are most fearful of going out in the absence of a test. Why are they fearful? Because they also care about getting treatment and/or protecting others. Indeed, a recent paper found that with respect to Covid-19, those who were at most risk of poor health outcomes from Covid-19, were more likely to socially distance. These are presumably people who would also be motivated to seek treatment if they found they were positive.

In other words, the population is sorting. There are unlikely to be people who are driven by the third reason but not by the first two and vice versa. In broad terms, you either care about all three or your care about none of them. Either way, it is far from obvious that someone is going to take a positive test and decide they are clearly safe to party precisely because they already chose not to party or are partying already. A positive test is not going to change their behaviour drastically in terms of risk to others. Thus, my hypothesis is that people worrying about trust are worrying about the type of person who is unlikely to exist in large numbers.

All that said, we don’t have to implement widespread testing in a vacuum. We could rely primarily on self-administered tests with occasional check-ins to assess compliance. What this discussion suggests is that a blanket ruling out of self-administration does not seem warranted especially given the speed and potential logistical advantages to operating that way.


What did I miss?