I coached a client a while back on her decision of whether (or rather when) to move away from the big city she was living in. There were many unknowns in her decision space. The opening premise was that there were many uncertainties to her decision that we needed to address. But eventually it emerged that the key issue was overcoming insecurity. How did we get there?
My client knew she missed the countryside and the sea. She knew that ultimately she wanted to get away from the rushed lifestyle of the city. She knew where she wanted to move to. But she also knew that she valued all the friendships she had built in the city over the years. She liked that she could walk everywhere or rely on excellent public transport. She liked that all sorts of cultural treasures are within easy reach for her now. She knew that moving away from all that would be a big change. That she’d have to build new social networks. That she might have to get a car to get around.
Nonetheless, she was quite clear on wanting to move, but she was uncertain on when would be the best time. What made her uncertain was that she was not sure how it would work out to build a new social network; whether she would find it easy to make new friends. And she was uncertain about how it would be to drive through narrow country lanes after not having driven for a long time.
After coaching her a bit and analysing the decision parameters she was grappling with, it dawned on me. I was a bit unsure at first about how to express this to her. But fortunately, we had good rapport and I could just say it. I realised that what she was struggling with was not uncertainty, but insecurity.
That was a real lightbulb moment for her, and totally transformed how she held her decision problem. It also critically determined the help I was able to give her to work through what had held her back. It’s easy to confuse insecurity for uncertainty. But in order to overcome insecurity you need different tools than when addressing uncertainty. So let’s dive straight in and tease these two apart.
What is Uncertainty?
To be clear, “uncertainty” can mean different things depending on context. In the sphere of choice and decision making, though, uncertainty refers to situations where a future state is uncertain. There is more uncertainty about the weather in London on April 1st than in Los Angeles. Southern California has a more constant climate year round than England. Usually, when we speak about uncertainty in decision making, we have some idea of how things might pan out, but, well, we cannot be certain.
We know that there are several ways how something can work out, but we do not know which one will happen. If you flip a coin you know that it has to land either heads or tails up. But you cannot know for certain which one it will be. In other words, we are in the realm of dealing with probabilities.
Three types of uncertainty
I’ve been teaching decision making for over 10 years at Cardiff University. I helped countless students to formally analyse and work through real-life decisions. This led me to understand that there are actually three slightly different types of uncertainty. I am not sure whether anyone else uses the taxonomy that follows. I don’t remember seeing it anywhere else. So it may well be just my personal view. But I have found it useful, certainly from the perspective of how to help people work through their uncertainties.
Calculable uncertainty
What I call calculable uncertainty in many ways is the easiest, most tractable form of uncertainty. You are facing calculable uncertainty when you can calculate the probabilities of various outcomes and — here is the irony — can be certain that your calculations are correct. Sometimes these calculations are simple, other times they are complex. If we go back to the coin flip example, provided we are dealing with a fair coin (i.e. one that has not been tampered with), the probability of throwing heads is 50%. This is an example of a simple probability calculation.
Somewhat more complex would be to calculate the likelihood of winning the jackpot of the Euromillions lottery. In this lottery you pick 5 numbers between 1-50 AND also two ‘lucky stars’ which are numbers between 1-12. The lottery then draws 5 random numbers between 1-50 and 2 random numbers between 1-12. If all of these numbers match the ones you have chosen, you win the jackpot.
It is pretty obvious that the chances of that happening are pretty slim. It is also pretty obvious that we can calculate exactly HOW slim these chances are. Less obvious (unless you are a mathematician, or have recently studied maths) is how to figure it out. In fact, I had to look up my high school probability calculus to remind myself of how to do it.
So, here goes: To calculate this, we need what is called the binomial coefficient to calculate the total number of possible combinations of drawing 5 different numbers between 1-50. This is C(50, 5) = 50! / (5!(50-5)!). If you don’t know what the ! means, don’t worry (it stands for factorial). At any rate, C(50, 5) = 2,118,760. So there are over 2 million different combinations of 5 numbers between 1-50.
On top of that we need to match the 2 lucky stars. Here a similar calculation gives us C(12,2) = 66, so 66 different ways of drawing 2 numbers between 1-12. In both cases, only 1 of them is the winning one. This means the chance of getting the winning 5 is less than 1 in 2 million. The chance of getting the correct lucky stars in 1 in 66. To win the jackpot we need both of these, and putting them together gives us a total chance of winning the jackpot of 1 in 139,838,160 or approximately 0.000000715%. Pretty slim chances. But, at least you can know precisely how likely it is that yours is the winning ticket.*
The point here is that in situations of calculable uncertainty we know that there is a way how we can calculate the probability of a particular outcome, even if we may not necessarily have the maths at our fingertips. When we know the probability of something happening precisely, we can say that we have maximally reduced uncertainty. Although we still cannot know what will happen, we know precisely how likely it is that a particular thing will happen.
Uncertainty with multiple reference classes
The next type of uncertainty is less tractable than calculable uncertainty. Here we have multiple so-called reference classes that allow us to calculate the probability of an uncertain event, but it is not clear which is the most relevant reference class to use. But what are reference classes in this context?
Imagine you wanted to know how likely it is that [insert your favourite football team here] will win the European Champions League. You could begin by considering the total number of teams participating. In 2023-24 there are 32 teams participating in the group stage, so you could say that the chance of your team winning is 1 in 32 or just over 3% (provided it qualified). Alternatively, you could consider the last 10 years and ask how often your team won. Between 2012 and 2022, Real Madrid won 5 times. If you take the past 10 years as guidance, that would give you a chance of 5/10 or 50% for Real Madrid winning (if they happen to be your favourite club).
You can probably see that the former reference class (total number of teams participating) is not the best, because it treats all teams equally and does not take into account past performance. But using the last 10 years as reference class may bring other problems, if, for example, you know that a particular team has undergone tremendous change in the previous season (good or bad). In other words, the past 10 years may not be that useful as a guide to expected performance either.
They way I think about such problems is that by thinking about what appropriate reference classes might be, we get insight into what the probability space looks like. A range between 3% and 50% is vast, especially if you wanted to bet money on your team, but it is still better than not knowing anything at all. And if we were really serious about wanting to know how likely it was that team X will win the Champions League, we probably could refine our probability range further.
But this is not a sports or betting blog so we won’t do that. The take home message is that if we get interested in a particular uncertainty, we can probably narrow down which reference classes are useful and which are not. Doing this enables us to increase the precision and accuracy of our estimate and thus, in fact reduces the uncertainty we face.
Uncertainty with no proper reference class: Degree of belief
Often we encounter situations where there simply is no precedent. These situations are the hardest to penetrate. Many political decisions fall into this category. When the British parliament decided to implement the Brexit referendum result by triggering article 50 to leave the EU, they had no idea how things would pan out. There was no historical precedent. No other country had left before. Nor were there relevant similar scenarios elsewhere in the world. How likely was it going to be that customs delays would lead to massive traffic jams around Dover? To food shortages in supermarkets? What were the chances of finding a solution that allowed Northern Ireland to leave the EU, but the Irish Republic to remain in the EU without creating a customs border between them? Of violence flaring up again in Northern Ireland? And so on.
When there is no appropriate reference class, there are no frequencies we can use to calculate probabilities. Instead we enter the realm of degrees of belief. When operating in this sphere, the best you can do is to try and find out as much as you can about the situation. Perhaps talk to experts and take their views (and perhaps even estimates). A degree of belief probability by definition will always be subjective, but with a bit of effort, you will be able to arrive at an estimate that feels right for you.
A good way to test your degree of belief probability is to use an imaginary gamble. Let’s say I asked you how likely you thought it was that it will snow tomorrow where you live. Let’s say I paid you $100 if it snows and nothing otherwise. Or you can have $X now instead. Whatever value X is for you, will be a good indicator of your subjective degree of belief that it will snow tomorrow. So if you are happy to take $10, you most likely believe that there is a 10% chance. If you needed $80 to give up the prospect of winning $100, you’re 80% sure it will snow. And so on.
Interim summary : How uncertainty matters
From a decision making perspective, dealing with uncertainty means that we do not know how things will turn out. As we have learnt above, we may be able to get a pretty good handle on the probability that a given outcome will occur. This is particularly the case when we have past instances that can guide us. The more carefully we think about the situation, the clearer it will become which instances make useful reference classes and why.
When no past instances are available, or when we are unsure about which reference classes are most appropriate, we turn to degree of belief approaches to probability. These are, by definition, subjective. If you have an important decision to make, you might feel uncomfortable relying on your subjective judgment of probability. That’s understandable. You might feel you ought to gather more evidence, talk to experts, do more research. All this will inform your subjective belief. Eventually, you will reach a point where you feel you have as good of a handle on this particular probability as you can possibly have.
That is all you need to make a good decision. It is important to watch out for analysis paralysis when dealing with uncertainty. There is probably always more research you can do to reduce your uncertainty. But at some point the extra mileage you gain by refining your probability estimates will not be worth your effort and time. And of course there is opportunity cost — if you delay a decision because you are stuck in information gathering mode you may lose access to some of your options.
As I explained in this article, the quality of a decision is not (purely) defined by its outcome, especially when uncertainty is involved. What matters is how clearly you as the decision maker have conceptualised the goals of your decision, how comprehensively you have delineated the various options, and how well you have appraised the relevant probabilities. If you’ve genuinely done the best you can to estimate the probability (and a tool like the imaginary gamble described above is a good way to measure this), then you can relax and know there wasn’t more you could have done.
What is insecurity?
You will not find any reference to insecurity in any decision making literature. The decision folks usually only deal with uncertainty. But coaches, of course, work with insecurity all the time. So what is the difference?
To me the key distinction is that if a situation involves uncertainty, I don’t know what will happen. I might know that there are multiple different ways how things can happen, how the coin might land, who might win the Champions League etc. But I do not know which of these things will happen (though I have an idea about how likely a particular outcome is). To me, uncertainty almost always refers to situations that are outside of my control. The weather, the coin flip, the state of the economy, the football score. Uncertainty is about not knowing a future state of the world.
Insecurity, in contrast, refers to one’s internal personal state. It is all about feelings towards the future. My client who wanted to move knew that she had to build a new social network. She knew that there would be less on offer in terms of culture. But what worried her is how she would feel about these things. No amount of probability calculus would help her with that.
Well, strictly speaking, you could approach these things probabilistically as well: You could ask, for example, how you felt in the past when you did similar things. Or you could even make predictions of how likely it is that you will be happy or sad. But there is a limit to how useful it would be to try to shoehorn your emotions into a probability matrix.
Overcoming insecurity
A far more promising approach is to lean into the insecurity. Unfortunately, society has conditioned us to view insecurity as a weakness. To be insecure is often equated with being immature. But we all feel insecure about all sorts of things. So if we want to overcome insecurity, we have to fist of all acknowledge it. Become interested in it. What is it about this situation that is scary? When else have you felt like this? What strategies have you used in the past to overcome fear or discomfort? And so on. A good coach is invaluable when addressing insecurities.
There is a pernicious trap that we can fall into when we encounter insecurities. This trap springs up when we mistake insecurity for uncertainty. Mistaking what ultimately is an emotional issue for a cognitive one, we go about resolving it in the wrong way. We might get increasingly obsessed with finding more information about how the future might turn out. We do more and more research. We may think that our information sources are not good enough. If only we knew more, then we would be in a good place to make this decision! We try to think our way out of a problem when what we need to do is work with our emotions and feelings.
If what is holding us back is insecurity, then no amount of additional information will ever be enough to make us feel ready to proceed. If we approach insecurity the way we would approach uncertainty, we are likely to get stuck in decision paralysis, ever searching for more information in a never-ending quest for clarity that will never come. Once we recognise and give space to the insecurity, we will be able to identify what is needed to assuage it. That is the shift that happened for my client, and there is no reason why it would not happen to you as well.
If you found this information useful, you may also be interested in my free eBook on what makes some decisions so difficult. You can download it below.
* Geeky Footnote
So you could in theory purchase just under 140 million lottery tickets and play every single combination. If you did that you would be guaranteed to win the jackpot, because one of your tickets will definitely be the winning combination. The problem is that each ticket costs £2.50. So you’d have to invest almost £350 million. The jackpot never has been that high and never will be, because the lottery is set up in such a way that they always make a profit.
