Decision Skills
Smarter COVID-19 Decision-Making
How to apply sound principles from decision science to your own life
If you’re not aware of COVID-19, you’ve probably been hiding under a rock for the last few weeks (good, stay there). The rest of you are probably asking yourselves, “What should I do, if anything?”
[This article is also available in: Arabic, Hindi, Italian, Russian, Spanish.]
First, here’s what NOT to do. Don’t form an uninformed opinion and then go foraging for media that confirms your views. You’ll find it, perhaps you’ll feel a bit better, but you may as well skip the whole exercise — it’s a waste of time. You already know you’ll just confirm whatever you wanted to believe. If this is what you’re doing, this guide is here to help!
A better way to approach things is:
- Focus your thinking on potential actions you might take.
- Select your decision criteria before you look at data.
Disclaimer: My expertise is in data and decision-making — not epidemiology — so this is a step-by-step decision-making guide to help you structure your thinking while you seek COVID-19 advice from scientists who know a thing or two about diseases.
Step 0: Face your irrationality
If your main objective is to feel better, perhaps you’ll find it soothing to read about two nasty psychological effects which might be ruining your ability to cope effectively with the information you’re getting:
If you’re interested in learning about how they’re messing with your head during the pandemic, check out my articles on those topics in the links above. Maybe it’ll be a sort of therapeutic moment courtesy of behavioral economics. (Said no one ever?)
Step 1: Understand yourself and set objectives
Start off by thinking carefully about your ethics and values. At this stage, try to think in more general terms than the recent pandemic. Ask yourself hard questions about what you consider to be your responsibility to your world, community, friends, family, and self.
If you’re not in the habit of investigating your own morality, now is as good a time as ever to have a grown-up moment and face unpalatable general questions like, “Under what circumstances, if ever, am I willing to put a stranger’s life at risk? How much risk?” (If you’ve ever driven a car around others or gone outside when you have the sniffles, I’m sorry to inform you that you’ve already risked strangers’ lives.)
Once you understand your priorities and what you consider to be right and wrong, you’ll start to form a view of your various objectives and how much they matter to you. For example, is the ability to eat at restaurants important to you? Is it more important to you than your family’s health?
This analysis assumes that your typical behavior is in service of existing objectives (e.g. visiting your favorite restaurant might bring you utility) and makes sense for you as long as it isn’t put into conflict with a different objective that you value more (e.g. staying healthy might bring you more utility). In the next few steps, we’ll structure your decision-making around signals that your old behavior is no longer consistent with your stated priorities.
Notice that at this stage in the decision process, you should avoid contemplating whether it’s safe to eat out or not. You’ll tackle that in Step 5. For now, stick with weighing the importance of things to you outside of the context of today’s specific virus.
Disclaimer: I should point out that since I’ve performed this exercise for myself, I clearly have opinions on what’s ethical, but I’m not here to impose any of my views on you. This post isn’t about telling you how much of a nice person to be (what am I? your government?), only how to make decisions wisely within your own moral framework… though I hope you’ll choose to be kind.
Step 2: Consider potential actions
Once you know where you stand, brainstorm a bunch of potential actions you might consider taking in the near future which deviate from what you’d do if there were no COVID-19. Everything from washing your hands a bit more to working from home to full self-quarantine to, I dunno, reading manuals on how to hunt bears with your bare hands.
Technically, the list of potential actions is infinite, so whenever one pops into your mind, you’ll save time by asking yourself:
- “Is making a careful decision about it important to me?”
- “Could anything convince me to do it?”
Answering “no” to either one means you can conserve effort by eliminating it outright so the decision process doesn’t take forever.
Example 1 — Not important: The decision to wash my hands a little more is not important to me because it doesn’t cost me much. Even if it turns out that it was a waste of time, I won’t be upset. I’m okay with just doing it. I’ll do it. No more thinking or research required. Cool, that was easy.
Example 2 — Not a real decision: There’s nothing (short of an utterly nutso sci-fi scenario) that could make me hunt a bear with my bare hands. Excellent. There’s no decision to make. I don’t have to think about this one anymore.
Once you’re done brainstorming, try to tackle your list in order of importance so that you get to the big ones before you get tired and go watch TV. Or, even better, set a timer for the amount of time you’re willing to invest into Steps 2–4. Each of the steps below should be done separately (in parallel unless they’re dependencies) for each action.
Step 3: Choose action triggers
For each potential action, think VERY carefully about which information you will accept as triggers for this action. Appropriate triggers (or “decision criteria” if you prefer additional syllables) are sufficiently strong signals that following your old course of action is inconsistent with your stated goals and priorities (thus triggering you to switch to the new action under consideration).
Let’s try an example. Perhaps you’ve got theatre tickets for tomorrow night and you’re considering the action of skipping the performance.
Default action: Go to the theatre as planned.
Alternative action: Change your behavior and don’t go.
By default, you’ll go to the show and you’re committing to not going if one of your triggers happens. (Choose them wisely — if you allow yourself to backpedal later, you’ve voided the whole point of structuring your decision-making. That’s why it’s important to really visualize that you’ve actually received this information and check that you’re content to respond with your new action.)
Throughout this section, you’ll need to be thinking in terms of your priorities — if you care only about your own short-term health, then your triggers will be mostly related to information about the personal risk of catching the virus and its severity in your demographic group. If you care about the wellbeing of others and/or about having access to medical resources in the next few months, you’d also need to consider triggers related to “flattening the curve” and information about the risk of spreading the disease (since it would be a disservice to your stated priorities to clog your nearest hospital with people who got sick because of you). It’s a relief that you took the time to weigh these priorities in Step 1, else how will you come up with appropriate triggers?
You may choose multiple triggers for each action. These triggers should all be phrased as information that you receive. This information might be about:
- A law, e.g. “My government has banned gatherings larger than x.”
- An event, e.g. “The number of people who tested positive within 10 miles of where I live passes x.”
- A research publication, e.g. “Scientists say that asymptomatic people can spread the virus.”
- A social norm, e.g. “Nice people aren’t going to the theatre these days.”
- Advice, e.g. “It’s a good idea to avoid being in a confined space with more than x people per square foot.”
- An explicit change in cost, e.g. “All tickets will be refunded upon request.”
- A change in risk estimates, e.g. “If I hang out in crowds, the probability of getting sick in the next two weeks is at least x.”
- A category I haven’t put on this list, e.g. “My best friend will disown me if I go outside.”
[Listen up, data nerds — if you’re used to thinking in terms of statistical hypothesis testing, things might get weird for you here since I skipped some steps to keep things simple for casual readers. I recommend checking out the appendix below where I explain what’s happening via the framework you’re probably used to — the one with default actions and null hypotheses.]
Step 4: Choose minimum quality of sources
Choose the minimum quality of information sources that can serve as valid triggers. If a source of lower quality gives you the information, you will attempt to verify it with your chosen source and not act unless the quality is sufficient. Continuing our example, your minimum quality sources for each of the triggers might be:
- Laws: The official website of your government. When NYC banned gatherings of over 500 people yesterday, I checked it on the legit NYC website. Others might accept the same information from a reputable newspaper and not verify further. Your minimum information source quality is a personal choice that’s up to you.
- Events: A sufficiently (for your tastes) reputable newspaper. Personally, I like the ones which have a reputation of proper fact-checking.
- Research: A qualified epidemiologist with access to sufficient data to make the present inference publishes finding with moderate-to-high confidence in their conclusion. As a scientist myself, I might be on the more careful side when it comes to trusting scientific publications. While some folks choose to trust anything any scientist says, earning my trust takes a combination of experience/credentials, access to relevant data, and the strength of assumptions versus strength of assertions.
- Social norms: Someone on Twitter? Which sources of cues about social norms you’ll accept is something so personal that I can’t begin to advise you. For some, it’s enough that a random person says it within earshot. Others might wait until it’s in Harvard Business Review’s etiquette section.
- Advice: Your doctor or your country’s equivalent to the Center for Disease Control (CDC). I’m not going to tell you where to get your medical advice, but I do hope you have standards.
- Cost: The theatre website. If the theatre says they’ll refund my ticket, I’m more convinced than if my neighbor says she heard a rumor that tickets might get refunded.
- Risk estimate: The CDC again, an informed statistician working with relevant data, or perhaps your own simulation. Yes, many of us data scientists are writing simulations (toy models of the world to see how things play out with different assumptions). Our simulations change when we update assumptions based on publications by scientists we trust (or find that bug in our code).
- Other: Other. For all categories, I advise you to have some quality standards (how strict is up to you) and to be aware of your standards. Never panic in response to chatter by a source that does not meet your personal quality bar for each trigger. For example, if my dad tells me that my best friend will disown me if I go outside, I’m not going to panic — I’ll call my best friend to verify.
Remember, these examples (vaguely based on my tastes) are not suggestions. I wouldn’t presume to tell you how to live your life, except perhaps to encourage you towards self-awareness. Each of us must make our own peace with how we structure our decision-making.
Step 5: Gather information
Now that you’ve come up with the minimum signals the universe needs to send you to trigger various actions, you’ll gather information (and keep gathering it) from your high-quality sources. If the decision is important, do your homework thoroughly.
Notice that I’m asking you to think before you look at information. That’s why the data gathering bit is saved until Step 5. If you consider typical habits of the human animal, you’ll notice that most folks don’t do it this way. So why am I asking you to do it?
Thinking and setting criteria BEFORE looking for information helps immunize you against confirmation bias. To learn more about that, check out my article on data-driven decision-making. Of course, it’s possible that things you forgot to consider come up, so you’ll need to make existence-based decisions (learn more here). Even in those cases, having thought a bit about the minimum strength and quality of information that you’ll accept as a trigger for action is very valuable.
Step 6: Act (or don’t)
If no relevant information comes in, keep doing what you were planning to do. When a different action is triggered, do it. (Perhaps you’ll notice that some actions should already have been triggered and this exercise helped you catch up.)
The moral of the story
Take some time to do a bit of thinking now so that you’re wiser and calmer when information comes along.
If you read all the way to this point and you know someone who might find this useful, please share it with them. In the meantime, I’m going to resist making a joke about what it takes for advice to go viral. Seriously, though — barely anyone will see this unless you (and you and you and you…) share it. The version of the link that allows anyone to view this article for free is: http://bit.ly/quaesita_covid
If you’d like to learn more about structured decision-making, a good place to start is my Introduction to Decision Intelligence.
The Hypothesis Tester’s Appendix
If you’re used to classical statistical inference and you want to know what the hell happened to the null hypothesis, I’ve put the appendix here.
What to read next
Liked the author? Connect with Cassie Kozyrkov
Let’s be friends! You can find me on Twitter, YouTube, Substack, and LinkedIn. Interested in having me speak at your event? Use this form to get in touch.
And now for something completely different…
Thanks for reading! If you had fun here and you’re curious about AI, here’s a beginner-friendly intro I made for your amusement:
