When faced with the question of “when will we be done?”, the most factually accurate answer we can give is one from a probabilistic forecast. Yet counter-intuitively, despite being the most correct answer, it’s usually not the one many people want. What they want is a deterministic answer, even if it’s less accurate.
Let’s start by defining those two terms:
- A deterministic answer is precise: “It will rain at 3:05pm today.”
- A probabilistic answer contains some imprecision, despite being more correct: “There is a 80% chance of rain after 3:00pm today.”
Or in the case of the typical projects we’re working on:
- Deterministic answer: “We’ll be done on March 1.”
- Probabilistic answer: “There is an 85% chance that we’ll be done on or before March 1.”
The deterministic answer is far more satisfying to our brains, despite being both less accurate and often misleading. Our brains are optimized for energy conservation and want to use as little energy as possible for any decision. The precision of the deterministic answer is attractive as we can just use that without doing any processing of the answer.
Our brains have many shortcuts built-in to optimize for that energy usage, and these are generically called cognitive biases. We tend to lean on these even when we’re in a calm state and then when we’re under pressure, or feeling unsafe, we use them even more.
Precision Bias is when we believe that something more precise is automatically more accurate than the less precise answer. Stating that we will deliver on March 1 appears more reliable. It feels like a better answer, even when objectively, it’s not.
An 85% chance of meeting a date means that there’s a 15% chance that you won’t. The cognitive bias of Loss Aversion tells us that we’re more likely to weigh potential losses more heavily than equivalent gains, leading to biased decisions. We want certainty and even a 15% chance of failing will be given disproportionate focus. A deterministic answer does not trigger loss aversion.
Probabilistic forecasts make variability visible, and Illusion of Control wants us to pretend that we are in full control. In an environment of low psychological safety, an uncertain answer can be interpreted as weakness and therefore dangerous.
An interesting variation on Illusion of Control is that many managers would prefer to have a deterministic answer that they KNOW is wrong because feel that they still have control over that. “I know they won’t make the first scheduled date, but they might make the second.”
Probabilistic forecasts are inputs to decision-making, not decisions themselves. When leaders expect a single “right answer”, probabilistic data can feel like a refusal to decide, and that can backfire on the person providing the forecast. It’s often safer to provide a deterministic answer that’s wrong.
So am I suggesting that you should give up on probabilistic forecasting and provide deterministic answers instead? Not at all.
What I’m suggesting is that we should understand what’s getting in the way of us making effective decisions from accurate data. If we don’t understand our own biases, then they will continue to control us. In this case, not understanding will lead to poorer decisions.
We should also strive to increase safety across our organizations, for many reasons, but in this particular case so that we can make better decisions.