Sampling and Bias | A Level Statistics Success

Alright everyone, let’s dive into a topic that sounds simple but quietly decides whether your whole statistics question works or not — sampling and bias.

Now, this one gets underestimated all the time. People think, “It’s just choosing people to ask, right?”
Well, kind of. But do it badly, and all your clever probability and confidence intervals go straight out the window.

Sampling is the foundation of statistics. If your sample’s dodgy, your results will be too — no matter how perfect your maths is.

🔙 Previous topic:

“Check how regression results inform your sampling strategy.”

What “Sampling” Really Means

So, what is a sample?

In statistics, a sample is a small part of a population that we actually collect data from.
It’s meant to represent the whole group without us needing to ask or measure everyone.

For example, if you’re studying students’ study habits, you don’t need to interview all 800 in your school — you just pick a smaller, representative group.

AQA sometimes starts with:

“Explain why a sample is used instead of a census.”

And you can confidently say:

“Because it’s quicker, cheaper, and less time-consuming.”

Just remember — it’s also less accurate than a full census. There’s always a trade-off.

Types of Sampling You Need to Know

Each exam board — AQA, Edexcel, OCR — loves testing this bit. They want to see you know the names, the differences, and (crucially) the pros and cons.

Let’s go through them teacher-style, plain and simple:

Type of Sampling	How It Works	Pros	Cons / Common Issues
Simple Random	Every individual has an equal chance. Usually by random numbers or names out of a hat.	Unbiased, easy to understand.	Time-consuming, needs a full list of population.
Systematic	Pick every nth person (like every 10th student).	Easy and quick.	Could be biased if there’s a pattern in the list.
Stratified	Divide the population into groups (e.g. by age or gender) and sample each in proportion.	More representative, fair for subgroups.	Needs good population info.
Quota	Interview until each group quota is filled (often in surveys).	No sampling frame needed, quick.	Can be interviewer-biased.
Cluster	Select entire groups randomly, like whole classes.	Convenient, practical.	Less accurate, may not represent everyone.

If you can name these and give one advantage and disadvantage each, that’s at least half the marks in any sampling question.

OCR often adds:

“Explain how stratified sampling reduces bias.”

You’d say:

“Because it includes all subgroups in proportion, avoiding over- or under-representation.”

Simple, clear, and mark-scheme approved.

Bias — The Sneaky Problem

Now then — bias.
This is the word that examiners love and students dread.

Bias just means the data isn’t fair — it’s tilted one way or another, often without you realising.

Here are some classic ways it sneaks in:

Selection bias: Only certain people get chosen (like surveying your own friends — they’re too similar).
Response bias: People lie, exaggerate, or answer how they think you want.
Non-response bias: Some people don’t reply at all, and their views might be different.
Measurement bias: The way you collect data influences results (like a badly worded question or faulty equipment).

AQA once asked:

“Explain why a voluntary response sample might be biased.”

The perfect answer?

“Because people who choose to respond may have stronger opinions than those who don’t.”

That’s it. Short, clear, and worth full marks.

How to Reduce Bias

The good news — you can control most bias if you plan carefully.

Here’s the checklist I tell my classes to memorise:

✅ Use random selection wherever possible.
✅ Keep questions neutral — no emotional wording.
✅ Make the sample size big enough to represent the population.
✅ Avoid asking only one type of person (e.g. same age group, same location).
✅ Record answers exactly, not how you interpret them.

Edexcel often phrases it as:

“Explain how bias could be reduced in this survey.”

They’re really asking: “How can we make it more representative?” So focus your answer on fairness and randomness.

Sample Size — Bigger Isn’t Always Better

Right, now this one’s interesting.

A larger sample does make your results more reliable — but only up to a point. After that, you’re just wasting time.

If your sample is well-chosen, even 50 people can represent 5,000. But if it’s biased, even 500 won’t fix it.

So don’t fall into the trap of writing, “Bigger sample = better.”
Add:

“Provided the sample is random and representative.”

OCR has rewarded students for that exact phrase.

Real Classroom Example

A few years back, I gave my Year 13s a task: find out whether students prefer studying in the morning or at night.

One group went around asking everyone in their first lesson at 8:45 a.m.
Guess what? Their results said everyone loved mornings.

Another group asked only those hanging around after school. Their results said the complete opposite.

Both groups did the maths right — but their samples were biased from the start.
That’s the point. Even perfect calculations can’t fix a poor sample.

Random vs Convenience Sampling

Quick one before we move on.

Some students confuse random with convenient.
Asking whoever’s nearby isn’t random — it’s just easy.

If you only question people walking past your classroom, that’s a convenience sample, not a fair one.

AQA occasionally phrases this as:

“Explain why this sample may not be representative.”

Answer:

“Because the participants come from one location and may not reflect the wider population.”

Clean, short, and you’re done.

Practical Tips for Exams

Here’s how to handle any sampling question calmly:

Step 1: Identify the population (who or what are we studying?).
Step 2: Spot what type of sample is being used.
Step 3: Decide if it’s fair or biased — and why.
Step 4: Suggest one improvement.

If you write those steps clearly, you’ll often earn every mark, even if your exact wording isn’t perfect.

And if you get a graph question tied to sampling — like comparing two surveys — focus on patterns and fairness, not the actual numbers.

Common Mistakes to Avoid

Here’s my little “exam red flag” list:

🚫 Using “random” when you mean “arbitrary.”
🚫 Forgetting to mention why bias happens.
🚫 Thinking sample = population (they’re not the same).
🚫 Writing “bigger sample” without explaining representativeness.

And the sneakiest one?
🚫 Forgetting to say what the sample represents.
If you don’t write “sample of students in the school” or “sample of people aged 18–25,” you lose clarity marks.

🧭 Next topic:

“Next, see how to present data effectively after collecting reliable samples.”

Final Teacher Reflection

I once told my students: “Statistics is only as honest as your sample.”

And it’s true. The whole field depends on fairness.
You can’t make good predictions or test sensible hypotheses if your data’s biased from the start.

So when you’re revising this topic, think less about the formulas and more about fairness. Ask yourself, “Would this give everyone an equal voice?”

That’s the mindset that separates an average answer from an A* one.

Make Statistics Feel Simple

Start your revision for A-Level Maths today with our 3-day A Level Maths course, where we teach statistics, mechanics, and pure maths step by step — in plain, friendly language.

We’ll help you master topics like Sampling and Bias so you can spot errors, explain patterns, and approach data questions with real confidence.

Sampling and Bias | A Level Statistics Success