Bolt 06 — Confidence Is Not Calibration

Bolt Series · Human Performance Calibration · Article 06

What if you are completely confident — and completely wrong?

We often talk to children as though confidence is automatically good.

Believe in yourself.

Be confident.

Do not doubt yourself.

There is something valuable in that advice. A learner who is paralysed by fear may never attempt work they are capable of doing.

But confidence has another possibility.

You can be extremely confident that an answer is correct.

And the answer can still be wrong.

You can be certain you are ready for an examination.

And discover that you were not.

You can doubt an answer that is, in fact, excellent.

So there are two different questions:

How sure are you?
How often is that level of certainty justified by reality?

The first is confidence.

The second is calibration.

Sport science gives us a useful warning

Confidence is genuinely associated with sporting performance.

But the relationship is not remotely strong enough to justify the simple slogan that more confidence automatically produces better performance.

A 2022 systematic review and meta-analysis covering 41 studies and 3,711 athletes found a positive relationship between self-confidence and performance, but the overall association was relatively small: about r = 0.25.

A separate meta-analysis of pre-event self-efficacy found a somewhat larger average relationship, about r = 0.31. Importantly, its prediction intervals included situations in which high self-efficacy did not lead to better subsequent performance.

That is exactly what Bolt needs.

Confidence matters.

But confidence is not a receipt.

Four students can have four very different confidence states

Imagine four students facing the same difficult Mathematics question.

Student A: confident and correct

She is sure the method works, can explain why, and repeatedly produces correct solutions under varied conditions.

This is useful confidence because it is supported by evidence.

Student B: confident and wrong

He recognises the surface of the problem, launches immediately into a familiar method and never notices that the method does not apply.

High confidence has not protected him.

Student C: doubtful and correct

She solves the problem correctly but distrusts every step, checks it repeatedly and nearly changes the right answer into a wrong one.

Capability is present, but the internal estimate is too low.

Student D: doubtful and wrong

He does not understand the method and correctly senses that he does not understand it.

That may sound negative, but it is actually a useful starting point.

If you accurately know that you do not yet know, you can ask for the right help.

Accurate uncertainty can be stronger than false certainty

Before a London race in 2015, Usain Bolt had been training hard after an interrupted season. He had enormous experience, extraordinary past performances and every reason to believe he could still compete at the highest level.

Yet he also said he would not really know how ready he was until he raced.

That sentence is useful because it contains confidence and uncertainty at the same time.

He knew what he had trained.

He knew which part of the race was weaker.

But he also recognised that competition would provide information he did not yet possess.

That is not weakness.

It is calibrated uncertainty.

The dangerous educational shortcut

If a student lacks confidence, adults often try to repair the situation by adding encouragement.

Sometimes that is exactly what the child needs.

But encouragement without evidence has limits.

“You’re amazing. You can definitely do it.”

What happens if the child attempts the task and fails repeatedly?

The encouragement can begin to sound disconnected from reality.

A stronger form of confidence is built differently.

“You solved eight of the last ten questions of this type correctly. The two errors came from the same step. Let’s repair that step, then test it again.”

Now belief is attached to evidence.

The learner does not merely hear that they are capable.

They can see why the claim is reasonable.

Confidence should have resolution

Students are often asked:

“Are you confident for the exam?”

That question may be too coarse.

A better conversation sounds like this:

Which topics are you highly confident about?
Which are uncertain?
What evidence supports each judgement?
When was the last time you performed them without help?
Can you do them when the question changes?
Can you do them under time pressure?
How certain are you that your answer is correct when nobody tells you?

Now confidence is becoming a map instead of a mood.

Why calibration matters more than simply raising confidence

A learner who is confidently wrong may stop checking.

A learner who is unnecessarily doubtful may avoid challenges they can handle.

Both students can make poor decisions for opposite reasons.

The educational target therefore cannot be maximum confidence.

It should be something more precise:

Be as confident as the evidence justifies — and be willing to update when the evidence changes.

Sometimes that will mean becoming more confident.

Sometimes it will mean becoming less confident.

Both can be improvements.

Because the goal is not to feel certain.

The goal is to know when certainty has been earned.