Sampling Methods and Stratification
Why Sampling?
A population is everyone or everything the investigation concerns. But collecting data from every member of a population is often impractical — too expensive, too time-consuming, or simply impossible. Instead, a researcher selects a sample: a manageable subset from which conclusions about the whole population can be drawn.
The quality of those conclusions depends critically on how the sample is selected. A badly chosen sample — one that over-represents some groups and under-represents others — produces biased results. No amount of clever analysis can fix a biased sample. This is why choosing the right sampling method is one of the most important decisions in the planning stage of any investigation.
Before sampling can begin, you need a sample frame: a list of population members available for sampling. For a school survey, the sample frame might be the roll of all registered students. For a national survey, a sample frame might be the electoral register or postcode address file. The sample frame may not include every member of the population — some individuals are always missed — and this itself can be a source of bias.
Methods That Rely on Judgement or Opportunity
Judgement sampling (also called purposive sampling) is where the researcher deliberately chooses participants they believe are typical. It is fast and avoids obvious outliers, but the researcher’s judgement introduces subjectivity and potential bias. If the researcher’s beliefs about the population are wrong, the sample will be wrong in a systematic way.
Opportunity sampling (also called convenience sampling) means selecting whoever is available. Surveying people who happen to be walking past at a particular time, or asking friends and classmates, are examples. It is the cheapest and fastest method, but it is highly susceptible to bias: the people who are available may systematically differ from the wider population in ways that affect the results.
A sample obtained by opportunity or judgement is not automatically worthless. If you are investigating an exploratory question — trying to identify variables worth studying in more depth — a convenient sample gives a quick first look. In early-stage research, speed can outweigh the need for perfect representativeness.
However, any conclusions drawn from such samples must be clearly caveated: they may not apply to the wider population. State this limitation when evaluating an investigation that used judgement or opportunity sampling.
Random Sampling
In a simple random sample, every member of the population has an equal chance of being included, and selections are made using a genuinely random process. This is the gold standard for avoiding systematic bias: no group is systematically favoured or excluded.
Random selection can be achieved using various techniques, including: assigning numbers to all members of the sample frame and using a random number generator, shuffling a pack of cards, or rolling dice. The specification notes these methods but is explicit: you will not be asked to derive a sample using these techniques in the exam — understanding the principle is what matters.
Systematic Sampling
Systematic sampling selects every ‘k’th member of the sample frame after a random start. For example, to select 20 from a list of 200 people, you calculate k = 200 ÷ 20 = 10, choose a random starting point between 1 and 10, then select every 10th person after that.
Systematic sampling is easier to implement than simple random sampling for large populations, and it spreads the sample evenly across the list. However, it can introduce a hidden pattern if the sample frame itself has a regular structure. If a list alternates male/female and you sample every other entry, you might end up with an all-male or all-female sample.
Systematic sampling steps: (1) calculate the sampling interval k = population size ÷ sample size; (2) choose a random start between 1 and k; (3) select every k-th member after that starting point.
Stratification: Not a Sampling Method
This is one of the most commonly misunderstood ideas in the specification. Stratification is a process that happens before sampling — it is not itself a method of sampling. The specification is explicit: stratification is not a method of sampling but a method which may be used before sampling takes place.
Here is how stratification works: first, you divide the population into meaningful subgroups called strata (singular: stratum) based on a characteristic relevant to the investigation. Then, you select participants from each stratum separately, using a random method, in proportion to the size of each stratum in the population.
Example: a school has 300 students: 120 in Year 10 and 180 in Year 11. You want a sample of 50. If you stratify by year group: Year 10 contributes 120/300 × 50 = 20 students; Year 11 contributes 180/300 × 50 = 30 students. Within each stratum, you then select randomly. This ensures the sample reflects the year-group balance of the school.
Stratification divides the population into subgroups BEFORE sampling. The actual sampling method used within each stratum is then random sampling. Calling stratification a “sampling method” is a common error that costs marks.
Quota Sampling
Quota sampling looks similar to stratification on the surface: you set a target for each subgroup. But there is a crucial difference. In stratified random sampling, participants within each stratum are chosen randomly. In quota sampling, the researcher fills each quota using whoever is available — there is no random selection within the quota. A market researcher might be told to interview 30 women aged 18–25 and 30 women aged 26–40, but they then choose those 60 participants by opportunity — approaching people in a shopping centre until the quotas are met.
Quota sampling is fast and proportional to the population structure, which is why commercial market research firms often use it. However, because selection within each quota is not random, it is susceptible to the same biases as opportunity sampling. The people a researcher approaches in a shopping centre on a weekday morning systematically differ from those who are at work.
Every sampling method trades off between cost/speed and bias risk. Understanding these trade-offs is what the specification is really testing:
- Opportunity: Fast, cheap, highly biased. Useful only for pilot/exploratory work.
- Judgement: Fast, avoids obvious outliers, still subjective. Researcher’s assumptions must be sound.
- Random: Unbiased in expectation, but depends on a sample frame that covers the relevant population and may still miss subgroups by chance in small samples.
- Systematic: Practical for large populations; watch for hidden patterns in the list.
- Stratification then random: Best for populations with important subgroups — guarantees proportional representation.