How Is Hypothesis Testing Used in Criminology?

First up — what on earth is hypothesis testing?

Alright folks — bit of a crossover topic today. Maths meets crime.
Yep, we’re talking about hypothesis testing in criminology.

Now, that might sound like two worlds that shouldn’t collide — but they do, all the time.
Statisticians and criminologists actually hang out in the same data sets.
Well, metaphorically anyway.

Let’s see how this works without getting lost in the jargon, yeah?

🔙 Previous topic:

“Revisit the basics of hypothesis testing before exploring its use in criminology.”

Okay, start from scratch.
Hypothesis testing is just… asking, “Is this thing real, or just random noise?”

That’s it. Seriously. You’ve probably done it a dozen times in A-Level stats — only this time the numbers are about people and behaviour.

Criminologists use it to test ideas like:

“Does a new youth program lower reoffending?”
“Has CCTV really reduced thefts?”
“Do longer sentences deter crime?”

We call that first assumption — the “nothing’s really changed” idea — the null hypothesis, H₀.
Then there’s the alternative hypothesis, H₁ — that something has changed.

It’s like the court system, right? We assume innocence (no change) until the evidence is strong enough to say otherwise.
That’s a line I use in class a lot — it helps people remember the logic.

Okay, but how do criminologists actually use it?

Let’s do a real one.

Suppose a city tries a new rehab scheme to stop young offenders reoffending.
The question is: does it actually work?

So the hypotheses would be:

H₀: The program makes no difference to reoffending.
H₁: The program reduces reoffending.

Then they collect some data — maybe one group gets the program, another doesn’t.
And after a few months, they check who reoffended.

If the difference is big enough that it’s unlikely to just be chance (like less than a 5% probability — that’s your p < 0.05), then they say,

“There’s evidence this program really helps.”

And notice the phrasing — “evidence suggests.”
Never “proves.” The exam boards (AQA, OCR, Edexcel — all of them) are very picky about that.

Right, so what are the steps?

Let’s keep it human. It always goes something like this:

Step 1: Write your two hypotheses.
For example, “Community policing has no effect on burglary rates” (H₀) versus “It lowers burglary rates” (H₁).

Step 2: Choose your significance level — often 5%.
That’s like saying, “We’ll only believe this if the odds of being wrong are really small.”

Step 3: Gather your data.
Could be crime reports, surveys, or stats from the police.

Step 4: Calculate your test statistic — maybe a z-score, t-test, or chi-squared, depending on the setup.

Step 5: Compare that to your critical value — the threshold.
If it falls into the critical region, you reject H₀.

Step 6: Say what that means — in context.
That’s the bit that gets marks in OCR and Edexcel. You need to translate it. Like:

“There’s evidence community policing reduces burglary rates.”

Plain, simple, contextual.

Let’s look at some real-style examples

📍 Stop and Search

Imagine you’re checking whether stop-and-search affects ethnic groups equally.

You’d set it up as:

H₀: Ethnicity and stop rates are unrelated.
H₁: They are related.

Then use a chi-squared test — you’ve seen this one before — to check if what you observed differs too much from what you’d expect by chance.

If it’s significant, you reject H₀.
In other words, there’s evidence of bias in stop rates.

That’s a proper statistical foundation for a social issue.

📍 Prison Education Programs

Let’s say prisons start offering education courses. The idea? Reduce reoffending.

Two groups — those who took the course and those who didn’t.
Run a t-test to compare their average reoffending rates.

If that difference is statistically significant, great — we’ve got evidence education helps.

And, by the way, AQA loves questions like that: same maths, different story.

📍 CCTV and Crime

Now, this one’s classic.
“Does CCTV reduce crime?”

Compare crime rates before and after cameras go up.
If the drop’s significant, you say, “There’s evidence CCTV is linked to lower crime.”

But — and this is where OCR likes to catch people — correlation isn’t causation.
Maybe the cameras helped, maybe something else changed.

Always write that.

“Correlation does not necessarily imply causation.”
That line alone could grab you a reasoning mark.

Why this matters so much

Here’s the big takeaway.

Hypothesis testing keeps criminology honest.

It stops people from saying, “This policy works!” just because they want it to.
Instead, it forces them to check if the data actually backs it up.

Without it, criminology would just be opinion dressed up as fact.

And that’s why you — yes, you, sitting here doing A-Level stats — are learning something genuinely powerful. You’re learning the language of evidence.

Tiny classroom story

A couple of years ago, one of my Year 13s, Sam, ran a mini research project.
He looked at whether adding police patrols near train stations reduced antisocial behaviour.

He gathered data from two weeks before and two after the change.
Did a small hypothesis test — found no significant difference.

He looked gutted. Thought it meant he’d “failed.”

But no — I told him, “That’s still real. It means your evidence doesn’t support the change — and that’s just as valuable.”

Because in proper research, no evidence of change is still evidence.
That’s what separates science from guesswork.

Common mistakes (seen them all, promise)

Let’s fix the usual traps:

🚫 Saying “proved” instead of “evidence suggests.”
✅ Always write carefully: “There’s evidence that…”

🚫 Writing H₀ and H₁ vaguely.
✅ Spell them out in context: what’s being compared and why.

🚫 Ignoring the story.
✅ Always link your stats to real behaviour: reoffending, policing, etc.

🚫 Forgetting to mention significance level.
✅ Always include it: 5% is the go-to.

And, sneaky one —
🚫 Saying “not significant” means “no effect.”
No. It just means “not enough evidence yet.”

That’s one of those subtle but important points OCR loves.

So… why should we care?

Because it’s about fairness.
Numbers don’t lie (well, unless you collect them badly).

Hypothesis testing gives structure to debates that could otherwise spiral into bias and emotion.
It’s what turns “I think” into “the evidence shows.”

In criminology, that matters a lot — people’s lives and policies depend on these decisions.

🧭 Next topic:

“Discover how the normal distribution underpins hypothesis testing.”

👩‍🏫Final reflection

I always say this:
Maths is only as powerful as what you use it for.

And in criminology, you’re using it to test fairness, justice, truth — all through numbers.

Once you see that connection, hypothesis testing stops feeling like a formula and starts feeling like a superpower.

🎯 Bring Maths to Life

Start your revision for A-Level Maths and Statistics today with our Year 13 Maths revision course, where we show you how to use numbers to understand real-world patterns — from test scores to crime rates.

We’ll help you see how tools like hypothesis testing turn data into understanding — one confident step at a time.

How Is Hypothesis Testing Used in Criminology?