Sampling and Research Designs
Sampling and Research Designs
Before a psychologist can run a study, they must decide who will take part and how those participants will experience the conditions of the study. Both decisions affect whether the results can be trusted and applied beyond the study itself. A target population is the whole group the researcher is interested in — for example, all GCSE students in England. Studying an entire target population is almost never possible, so researchers select a smaller sample to represent it.
A sample must be representative of the target population it comes from. If it differs systematically from the population — for example, if it only includes one age group — conclusions cannot be safely generalised to the wider population.
Methods of Sampling
The specification identifies four sampling methods. Each strikes a different balance between practicality and how representative the resulting sample is likely to be.
Random sampling gives every member of the target population an equal chance of selection, usually by listing the population and using a lottery or random number generator. Strength: it removes researcher bias in who is chosen. Weakness: it requires a complete list of the population, which is often impossible to obtain, and selected people may refuse to take part.
Stratified sampling divides the population into subgroups (strata), such as by age or gender, then randomly selects participants from each stratum in proportion to its size. Strength: the sample mirrors the structure of the target population, so it is highly representative. Weakness: it is time-consuming and requires detailed knowledge of the population’s make-up.
Volunteer sampling involves participants putting themselves forward, often by responding to a poster or online advert. Strength: it is quick to set up and participants have given genuine consent to take part. Weakness: volunteers tend to share unusual traits, such as being unusually keen or curious, which can bias the sample.
Opportunity sampling selects whoever is conveniently available at the time, such as a researcher approaching students in a school corridor. Strength: it is fast, cheap and easy to carry out. Weakness: it is biased towards people who happen to be in that place at that time, so it may not represent the target population.
Four sampling methods: random (equal chance, unbiased but needs a full list), stratified (proportional subgroups, representative but slow), volunteer (self-selected, easy but biased towards keen people) and opportunity (whoever is available, quick but biased).
Research Designs: Organising Participants Across Conditions
Once a sample has been chosen, the researcher must decide how participants will experience the conditions of the study. This is called the research design. The specification identifies three designs, each solving the problem of comparing conditions in a different way.
In an independent measures design, different participants are used in each condition. Strength: participants only take part once, so there are no order effects such as practice or fatigue. Weakness: differences between the groups (individual differences) might explain the results rather than the independent variable, and more participants are needed overall.
In a repeated measures design, the same participants take part in every condition. Strength: individual differences cannot explain any difference between conditions, since the same people are compared with themselves. Weakness: order effects can occur — participants may improve through practice or grow tired or bored across conditions. Researchers can reduce this problem by counterbalancing, which changes the order conditions are presented in for different participants.
In a matched pairs design, different participants are used, but they are matched on characteristics relevant to the study (such as age or prior ability) before one member of each pair is placed in each condition. Strength: it reduces the impact of individual differences without the order effects of repeated measures. Weakness: matching takes time and it is very difficult to match participants on every relevant characteristic, so some individual differences may remain.
Independent measures avoids order effects but risks individual differences between groups. Repeated measures removes individual differences but risks order effects. Matched pairs is a compromise — it reduces individual differences without order effects, but matching is time-consuming and never perfect.
Sampling method and research design are chosen together, based on the aims of the study and the practicalities the researcher faces. Both choices affect the generalisability of the results: a poorly chosen sample or design can undermine even the most carefully measured study.