Possible Is Not Probable: How to Think Better About Chances and Base Rates

One sentence can derail an otherwise sensible decision:

"Yeah, but what if...?"

What if the investment crashes?

What if the new employee turns out to be terrible?

What if the trip gets cancelled?

What if the business idea suddenly becomes enormously successful?

What if this one unlikely thing happens?

All fair questions.

But there is a problem hiding inside them.

The moment something becomes possible, our minds can start treating it as though it deserves the same attention as something likely.

It doesn't.

A one-in-two chance and a one-in-ten-thousand chance are both possibilities.

They are not remotely the same decision problem.

And the reverse matters too:

Something with an 80% chance of happening is still not guaranteed.

Probability is uncomfortable because it refuses to give us what we really want:

certainty.

Instead, it gives us something much more useful:

a better way to be uncertain.

If you keep reading, you'll learn:

• why "it could happen" tells you surprisingly little;
• what a base rate is and why ignoring it can distort decisions;
• why a convincing story can overpower boring but important statistics;
• why one success or failure tells you less than it seems;
• why percentages often become easier to understand when converted into actual people or cases;
• how to distinguish probability from consequence;
• and a practical seven-question method for thinking more clearly when nobody can tell you exactly what will happen.

The goal is not to turn you into a statistician.

The goal is much simpler:

When certainty is impossible, learn to think in chances.

Possible and probable are not the same thing

Suppose somebody tells you:

"It's possible."

Fine.

But possible covers an enormous range.

It could mean:

"This happens roughly half the time."

Or:

"Technically, this has happened once."

Those statements should produce very different decisions.

Yet ordinary conversation often collapses them into the same phrase:

It could happen.

Yes.

A meteorite could land on your car tomorrow.

That does not mean you should spend tonight shopping for meteorite insurance.

Possibility asks:

Can this happen?

Probability asks:

How often should I expect something like this to happen?

That second question usually matters much more.

Probability does not predict one future

This is one of the first ideas that makes probability easier to understand.

Suppose an event has a 70% chance of happening.

That does not mean:

• it will happen;
• it is "basically guaranteed";
• it should happen this particular time;
• a failure proves the estimate was wrong.

It means that under the assumptions behind the estimate, this outcome is considered more likely than its alternative.

Probability describes uncertainty.

It does not eliminate it.

Think about a fair coin.

Heads has a 50% probability.

If you flip it once and get tails, nothing has gone wrong with mathematics.

You observed one outcome from an uncertain process.

Real decisions work the same way, except the probabilities are usually much harder to know.

Rare does not mean impossible

People sometimes make the opposite mistake.

They hear:

"Only a 2% chance."

and translate that into:

"Can't happen."

But 2% is not zero.

If enough opportunities occur, rare events eventually appear.

That is why good probability thinking requires holding two ideas at once:

Unlikely events happen.

and:

Their existence does not make them likely.

This sounds simple.

But emotionally, it can be difficult.

The moment an unlikely event happens to somebody we know—or appears vividly in the news—it becomes much easier to imagine.

And imagined events can start feeling probable.

That connects directly with risk perception.

How dangerous something feels and how likely it is to happen are related questions.

They are not identical ones.

Start with the base rate

Imagine somebody tells you:

"I met a new entrepreneur. They are incredibly confident, energetic, persuasive, and obsessed with their company. Do you think the business will become enormously successful?"

Your mind immediately has something to work with.

You can picture the person.

They sound like a successful founder.

Maybe they remind you of stories about famous entrepreneurs.

So you start predicting from the description.

But there is another piece of information worth asking about first:

How often do businesses of this kind achieve the outcome we're talking about?

That background frequency is a base rate.

It tells you what generally happens in the relevant reference class before you become hypnotized by the details of one specific case.

The same idea appears everywhere:

Before asking whether this particular project will finish on time:

How often do similar projects finish on time?

Before asking whether this candidate will succeed:

How often do comparable hires succeed under similar conditions?

Before becoming excited about an investment:

How do similar investments usually perform?

Before assuming an unusual event is common:

How common is it in the first place?

Base rates are not the whole answer.

Specific information matters.

But the base rate gives the specific information somewhere to start.

Why stories can beat statistics

A number is abstract.

A person is not.

Consider:

"Only a small percentage of projects like this fail."

Then somebody tells you:

"My friend tried exactly this and lost everything."

Which one do you feel?

Probably the friend.

Stories have detail.

Faces.

Emotion.

Cause and effect.

You can imagine yourself inside them.

Base rates are usually less charismatic.

They sit quietly in the corner being correct in percentages.

Research by Daniel Kahneman, Amos Tversky, Maya Bar-Hillel, and others explored conditions under which people can underweight prior probabilities or base rates when vivid case-specific information is available.

The important nuance is that humans do not simply "ignore statistics."

Whether people use base-rate information depends on things such as how relevant the information appears and how the problem is represented.

Still, the practical warning is useful:

A vivid detail can feel more informative than it actually is.

This is one of the reasons understanding how the brain makes decisions is useful.

Our intuitions are extremely good at turning details into stories.

Probability sometimes requires us to step outside the story.

A simple base-rate example

Let's make this concrete without turning the article into homework.

Imagine a factory produces electronic components.

Historically:

1 out of every 100 components is defective.

The factory uses a scanner.

The scanner:

• correctly flags 90% of defective components;
• but also incorrectly flags 10% of good components.

A component gets flagged.

What is the chance that it is actually defective?

Your first reaction may be:

"The scanner catches 90%, so probably around 90%."

But let's stop using percentages for a moment.

Imagine 1,000 components

Out of 1,000 components:

10 are actually defective.

If the scanner catches 90% of them:

9 defective components get flagged.

Now look at the 990 good components.

If 10% are falsely flagged:

99 good components also get flagged.

So the scanner flags:

108 components in total.

Of those:

9 are actually defective.

That is roughly:

8 out of every 100 flagged components.

The result feels surprising because the original defect is rare.

The scanner's accuracy matters.

But so does the base rate.

Why "out of 1,000" can be easier than percentages

Researchers have spent decades studying ways to make probability reasoning easier.

One useful approach is to convert conditional percentages into what are often called natural frequencies.

Instead of saying:

"The condition occurs in 1%, the test sensitivity is 90%, and the false-positive rate is 10%..."

you say:

"Imagine 1,000 cases..."

Then count what happens.

Research by Gerd Gigerenzer and Ulrich Hoffrage found that frequency formats can substantially improve performance on certain Bayesian reasoning problems.

Later research, including a meta-analysis by Michelle McDowell and Perke Jacobs, found a broader facilitation effect while also showing that details of representation and task design matter.

The practical lesson does not require you to learn Bayes' theorem.

When percentages become confusing:

Turn them into people, objects, or cases.

Instead of:

"3%"

try:

"3 out of 100."

Instead of several interacting percentages:

"Imagine 1,000 cases. How many start in each group? How many remain after the next piece of information?"

Sometimes the fog disappears immediately.

The denominator matters

Imagine these two statements:

"10 people experienced the problem."

and:

"10 out of 12 people experienced the problem."

Very different.

Now:

"10 out of 2,000 people experienced the problem."

Also very different.

The numerator gets attention because it tells the story.

The denominator tells you the scale.

Whenever somebody gives you a number without the population behind it, ask:

Out of how many?

Ten failures sounds alarming.

Ten failures among twelve attempts probably should.

Ten failures among ten million attempts tells a different story.

Always look for the denominator.

Personal experience is useful—and dangerous

In sales, you hear a lot of stories.

One customer says:

"I bought this before and it was perfect."

Another says:

"My brother had one and it was a nightmare."

Those experiences matter.

They are real observations.

But they are also tiny samples.

One experience can tell you:

This outcome is possible.

It usually cannot tell you:

This outcome is common.

I saw the same pattern repeatedly when customers were making large purchases.

One vivid experience—good or bad—could suddenly dominate everything else they knew.

And I understand why.

If your friend had a terrible experience, a spreadsheet about average reliability feels emotionally weak beside that story.

But I learned to ask a different question:

Is this story evidence about the overall pattern, or evidence that this particular outcome can happen?

Sometimes it is both.

Sometimes it is mostly the second.

That distinction is useful far beyond sales.

Do not learn probability from one outcome

Suppose you make a decision with an estimated 80% chance of success.

It fails.

Was the decision foolish?

Not necessarily.

An 80% probability contains a 20% probability of failure.

Now suppose you take a reckless chance with a 10% probability of success.

It works.

Were you brilliant?

Again, not necessarily.

This is where probability connects directly with outcome bias.

One result does not automatically tell you whether the original decision process was sound.

If you judge probability only after seeing the outcome, luck starts rewriting the lesson.

Think in repeated worlds

Here is a useful mental trick.

When evaluating an uncertain decision, imagine you could repeat the same decision many times under similar conditions.

Would you like the strategy?

Suppose somebody offers you a game:

• 80% chance to gain $10;
• 20% chance to lose $1.

One play could still lose.

But across repeated comparable opportunities, the structure is attractive.

Now reverse it:

• 20% chance to gain $1;
• 80% chance to lose $10.

You could win once.

You may even tell everyone how clever you were.

The structure is still terrible.

For many everyday decisions, you cannot literally repeat the situation.

But asking:

"Would I want to make decisions like this repeatedly?"

helps separate process from luck.

Probability is not consequence

This distinction matters enormously.

Suppose there is:

a 20% chance of losing $5

versus:

a 1% chance of losing your home.

Which should you care about more?

You cannot answer by probability alone.

Likelihood is one dimension.

Consequence is another.

This is why our risk perception article separates:

"How likely is it?"

from:

"How bad would it be?"

Do not combine them prematurely into one emotional word such as:

risky.

A low-probability catastrophic event may deserve serious preparation.

A high-probability minor inconvenience may deserve almost none.

Good decisions consider both.

Ask: compared with what?

A probability without comparison can be misleading.

Imagine someone says:

"This option has a 10% chance of failure."

That sounds useful.

But what are the alternatives?

Option B may have:

25%

Doing nothing may have:

40%

Suddenly 10% looks quite different.

Or perhaps another option has:

1%

Now 10% looks worse.

This is why the question:

"Is it risky?"

is often weaker than:

"Risky compared with what?"

Decisions usually involve competing probabilities.

Not probability versus certainty.

Do not demand certainty where none exists

People sometimes continue researching because they believe one more piece of information will make uncertainty disappear.

Eventually they want to reach:

"Now I KNOW."

Sometimes that is possible.

Often it isn't.

You can research:

• a new career;
• a major purchase;
• a business project;
• an investment;
• a hiring decision;

and still finish with uncertainty.

At some point the question changes from:

"Can I know exactly what will happen?"

to:

"Do I know enough to choose responsibly?"

Probability thinking is partly the ability to make a decision before uncertainty reaches zero.

Because uncertainty rarely reaches zero.

The planning fallacy is partly a probability problem

Consider a project.

You imagine the path:

Step 1 works.

Then Step 2.

Then Step 3.

Then launch.

Each individual assumption seems reasonable.

But several uncertain assumptions multiplied together can produce a much less certain overall plan.

This is closely connected with the planning fallacy.

The inside view asks:

"Can I imagine this plan working?"

Usually yes.

The stronger question is:

"How often do plans like this work this smoothly?"

That brings you back to reference classes and base rates.

Possible?

Absolutely.

Typical?

That is a different question.

Be suspicious of precision without evidence

There is something comforting about precise numbers.

Compare:

"I think this has a decent chance."

with:

"There is a 73% chance."

The second sounds smarter.

But where did 73 come from?

Sometimes precise probabilities come from excellent data and models.

Great.

Sometimes they come from somebody's confidence wearing a tie.

Do not confuse:

precision

with:

accuracy.

A useful question is:

"What evidence supports this probability?"

Was it based on:

• historical data?
• a model?
• expert judgment?
• a tiny sample?
• one anecdote?
• a guess?

Numbers deserve the same skepticism as stories.

Sample size matters

Suppose a restaurant receives:

5 stars from 3 reviews.

Another receives:

4.8 stars from 3,000 reviews.

Which rating tells you more?

The first may genuinely be excellent.

But the estimate is much less stable.

Small samples swing dramatically.

One unhappy customer changes the average.

One lucky result changes the story.

This principle appears everywhere.

If somebody tells you:

"Everyone I know thinks this."

ask:

"How many people is everyone?"

If somebody says:

"We've tried it twice and it worked both times."

good.

You have two observations.

Not a law of nature.

The CantDecide Chance Check

Here is the practical framework.

When you face an uncertain decision, run through these seven questions.

1. What exactly is the event?

Avoid vague statements such as:

"This might go badly."

Define "badly."

For example:

"The project may exceed the budget by more than 20%."

"The candidate may leave within six months."

"The product may fail before three years."

Probability needs a defined event.

Otherwise you are measuring fog.

2. What is the base rate?

Ask:

What usually happens in situations like this?

Look for the relevant reference class.

Not the broadest possible group.

Not the group that conveniently supports the answer you want.

A genuinely comparable group.

3. What specific information should change the base rate?

Base rates are a starting point.

They are not a prison.

Maybe this case really is different.

Ask:

"What evidence makes it different?"

Experience?

Better technology?

Different market?

Strong performance history?

Different circumstances?

Then ask:

How much should that information actually change my estimate?

Do not jump automatically from:

"Usually unlikely"

to:

"This time definitely different."

4. Can I turn percentages into frequencies?

If the numbers become confusing, imagine:

100 cases

or:

1,000 cases.

Ask:

"How many start in each group?"

"How many would produce this result?"

Seeing the actual groups can make the logic much easier.

5. What is the consequence?

Keep likelihood and severity separate.

Ask:

"If this happens, how bad is it?"

and separately:

"How likely is it?"

Then combine those judgments when making the decision.

6. What am I comparing this with?

Do not ask:

"Does Option A have risk?"

Ask:

"How does Option A compare with B, waiting, or doing nothing?"

The alternative also has probabilities.

7. What probability would actually change my decision?

This is the question people often skip.

Suppose an outcome has:

10% probability.

Would you choose differently at 5%?

What about 20%?

40%?

If your decision is identical across all reasonable estimates, spending another six hours debating whether the true number is 14% or 17% may not help.

The purpose of probability is to improve the decision.

Not to win an argument with uncertainty.

A 60-second version

For ordinary decisions, use this shorter version:

What could happen?

Define the event.

How often does something like this normally happen?

Find the base rate.

What makes this case different?

Update the estimate.

Out of 100, what would this look like?

Turn probability into frequency.

How bad is the outcome?

Separate likelihood from consequence.

Compared with what?

Evaluate alternatives.

Six questions.

No equations required.

For important medical, financial, legal, engineering, or safety decisions, use relevant real-world data and qualified domain expertise rather than relying on a general framework alone.

Probability should make you calmer—not falsely confident

Good probability thinking has an interesting effect.

It can reduce panic without creating complacency.

Instead of:

"THIS COULD HAPPEN."

you can ask:

"How likely?"

Instead of:

"That almost never happens, so I can ignore it."

you can ask:

"How severe would it be if it did?"

Instead of:

"It worked last time."

you can ask:

"Was last time representative?"

Instead of:

"I'm sure."

you can say:

"I'm reasonably confident, and here's why."

That is not weakness.

That is calibrated confidence.

You are allowed to say "I don't know"

Sometimes there is not enough information for a useful probability estimate.

That matters.

People often feel pressure to turn uncertainty into a number:

40%.

70%.

95%.

But sometimes the responsible answer is:

"I genuinely don't know."

You can still make a decision.

You may choose:

• to gather more information;
• to reduce exposure;
• to create a backup;
• to run a small experiment;
• to delay an irreversible commitment;
• or to accept the uncertainty.

Unknown is a legitimate category.

Fake precision is not an improvement.

Final thought

Probability sounds like mathematics.

But in everyday life, it is really a way of thinking.

It teaches you to stop asking only:

"Can this happen?"

and start asking:

"How often should I expect something like this to happen?"

It reminds you that:

possible is not probable;

unlikely is not impossible;

likely is not guaranteed;

one story is not a base rate;

one outcome is not a pattern;

and a precise number is not automatically a good estimate.

Most important decisions happen before we know how the story ends.

So certainty cannot be the standard.

The better standard is:

Did I understand the uncertainty well enough to make a responsible choice?

You will still sometimes be wrong.

That is unavoidable.

But being wrong after thinking clearly about the probabilities is very different from being surprised because you treated every possibility as equally likely.

The future will always contain uncertainty.

You do not need to eliminate it.

You need to learn how to think in chances.