IB Psychology HLTopic 5 — Research DesignPaper 3 & IAPractical skill~11 min read
Sampling Techniques and Who Takes Part
Sampling is the least glamorous decision in research design and one of the most consequential. Whoever ends up in your study decides what your conclusion is really about — and in almost every school project, the sample is chosen for convenience and the conclusion is written as if it were chosen for science.
📚 What you need to know
A target population is the whole group you want to say something about; a sample is who you actually study.
Opportunity sampling: whoever is available and willing at the time.
Self-selecting (volunteer) sampling: people respond to an advert and put themselves forward.
Random sampling: every member of the population has an equal chance of being chosen.
Stratified sampling: the population is split into categories and sampled proportionally.
Snowball sampling: existing participants recruit others, used for hard-to-reach groups.
Each method has one signature bias, and naming it is where the marks are.
Target population first
Before choosing a method, decide who the study is about. Sometimes the population is very specific — if you are researching the experience of single teenage parents, your sample has to come from that exact group. Sometimes it is broad: for a study of short-term memory duration, most adults between roughly 18 and 60 would do, and no distinct target population is needed.
That decision drives everything else. A narrow population makes random sampling almost impossible, which is why the harder-to-reach the group, the more likely a researcher is to use snowball sampling.
The middle panel is what most school studies actually do. There is nothing wrong with that, as long as the conclusion is written honestly.
The five methods, compared
Method
How it works
Main strength
Signature weakness
Opportunity
Take whoever is available and willing at the time
Quick, easy and cheap; participants have agreed, so the study runs smoothly
Unconscious researcher bias in who gets approached; cannot generalise
Self-selecting
Advertise and let people volunteer
Willing, enthusiastic participants who are less likely to disrupt the study
Volunteer bias — volunteers share personality traits such as being outgoing
Random
Every member of the population has an equal chance of selection
Eliminates researcher bias in selection; usually representative
Time-consuming and often impractical; can still produce an unbalanced sample by chance
Stratified
Divide by key characteristics and sample each category proportionally
A small-scale reproduction of the population, so easy to generalise from
Slow; and you cannot always classify every person into a subgroup confidently
Snowball
Participants recruit other similar participants
Reaches hidden populations; the referral builds trust
Very narrow scope; the researcher has little control over who joins
Opportunity sampling gets a worse reputation than it deserves. It is the honest choice for a school IA. What loses marks is not using it — it is using it and then writing a conclusion about “teenagers” when you asked eleven people in your own year group.
Where the bias comes from
Each method fails in a predictable way, which is why naming the method lets you name the bias automatically.
Opportunity sampling fails through the researcher. People approach those they feel comfortable with, select those they think will be interested, and quietly avoid social groups they are wary of. The sample ends up shaped by the researcher’s own comfort.
Self-selecting sampling fails through the participant. Volunteers are, almost by definition, more sociable, more curious and more compliant than average, and that compliance shades into acquiescence bias and demand characteristics once the study starts.
Random sampling fails through chance and practicality. You rarely have a complete list of the population, not everyone selected will agree to take part, and randomness can still hand you an all-male sample that misrepresents the group entirely.
Stratified sampling in one example: if 18% of your target population are males aged 30 to 40, then 18% of your sample must be too. The proportions in the sample mirror the proportions in the population — that is the whole idea.
🧩 Choosing a sampling method for your IA
Write down your target population in one sentence before anything else.
Ask if you can list everyone. If not, random sampling is off the table immediately.
Ask if subgroups matter for your topic. If age or gender should affect the result, stratify.
Be honest about time. Most school projects end at opportunity or self-selecting sampling.
Name the method in your write-up and name its bias in the same paragraph.
Match your conclusion to your sample. Do not claim more people than you studied.
Worked examples
WORKED EXAMPLE
Identify the method and its weakness
A researcher studying attitudes to recycling stands outside a supermarket on a Tuesday morning and asks passing shoppers to complete a short questionnaire. Identify the sampling method and explain one limitation.
Step 1: Name the methodOpportunity sampling — whoever happens to be available and willing.
Step 2: Identify who is systematically missing
Tuesday morning shoppers skew towards people not in full-time employment, so working adults and students are underrepresented.
Step 3: Add the researcher-side biasUnconscious bias in who gets approached — researchers tend to approach people they feel comfortable with.
Step 4: State the consequence
The sample does not represent the target population, so external validity is low and the findings cannot be generalised.
Opportunity sampling; unrepresentative sample lowers external validitythe time and place detail is usually the clue the question wants you to use
WORKED EXAMPLE
Choose a method and justify it
A researcher wants to interview people who have left a high-control religious group. Very few are publicly identifiable and most would be uncomfortable being approached directly. Suggest a suitable sampling method and evaluate it.
Step 1: Name the methodSnowball sampling — find one or two participants, then ask them to recruit others.
Step 2: Justify it
The population is hard to reach and would be unlikely to respond to a public advert. Being referred by a peer in a similar situation builds trust.
Step 3: Give the strength
Without it, this group’s experiences would go unrepresented in psychological research entirely.
Step 4: Give the limitation
The researcher has little control over who joins, and participants tend to recommend people with similar experiences, so the scope is narrow and credibility may be questioned.
Snowball sampling; access and trust, at the cost of scope and control“evaluate it” needs both sides even when the choice is clearly right
💡 Exam tip
Learn one signature weakness per method. Volunteer bias for self-selecting, researcher bias for opportunity, impracticality for random.
Never confuse random sampling with random allocation. Sampling is who gets in; allocation is which condition they do.
Link every sampling criticism to external validity or generalisability by name.
Say who is missing from the sample, not just that it was unrepresentative.
Snowball sampling is the answer whenever a question mentions a hidden, stigmatised or hard-to-reach group.
Stratified sampling questions often involve percentages. Be ready to say the sample must mirror the population proportions.
⚠ Common mix-up
Thinking random sampling guarantees representativeness. It removes researcher bias, but chance can still produce a lopsided sample.
Calling opportunity sampling random. It is the opposite — availability is not chance.
Mixing up self-selecting and opportunity. Volunteers come to you; opportunity means you go to them.
Assuming stratified sampling is always best. It is only worth the effort if the subgroups actually matter for the topic.
Forgetting that snowball samples are narrow by design. The referral chain is both the strength and the weakness.
Writing “the sample was biased” with no method named. The method is what tells the examiner you know why.
Up next: Reading and Drawing Graphs — the start of the HL data analysis unit, where the arguments about design turn into actual numbers on a page.
Want this explained one-to-one?
Book a free session with an experienced IB Psychology tutor and get your trickiest topics made simple.
