Data types, sources and sampling

Two different questions about data

Psychologists collect data to investigate behaviour and mental processes. To understand that data, ask two separate questions: What form does it take? and Where did it come from?

The first question distinguishes quantitative from qualitative data. The second distinguishes primary from secondary data. These classifications work together, rather than being four competing categories.

Quantitative data: measurements and counts

Quantitative data expresses information numerically. A memory-test score, the time taken to complete a task and the number of times a behaviour occurs are all quantitative data.

Suppose a psychologist investigates recall in silence and with music playing. Recording how many words each participant remembers produces quantitative data. The scores allow the researcher to compare performance across the two conditions and summarise the results using statistics, tables or graphs.

Experiments often produce numerical measurements. Observations can produce counts of behaviour, and questionnaires can produce numbers through closed questions or rating scales.

Numerical data is useful for identifying patterns and making comparisons. A clearly defined scoring rule, used consistently for every participant, can also help produce reliable, or consistent, results. However, a number may reveal little about a participant's experience. A low recall score tells us how many words someone remembered, but does not by itself explain why they struggled.

Qualitative data: descriptions and meanings

Qualitative data expresses information in non-numerical form, such as words or images. It can describe people's thoughts, feelings and experiences in their own terms.

In the same memory investigation, a participant might explain, 'I kept listening to the lyrics instead of concentrating on the words.' This account is qualitative data. It gives detail about the participant's experience that a recall score alone would miss.

Interviews, diary accounts and open questionnaire questions can produce qualitative data. Researchers can look for recurring themes and patterns in these accounts.

Qualitative data can provide rich detail and help researchers understand what an experience means to an individual. However, it can take considerable time to analyse. Interpretation also matters: two researchers might understand the same account differently, reducing consistency. An unstructured interview is difficult to repeat exactly because its questions and follow-up discussion can vary.

Neither form is always best. Quantitative data suits a question about how much performance differs; qualitative data suits a question about how participants experience the task. Collecting both can give a broader understanding.

Primary and secondary data: who collected it?

Primary data is new information collected first-hand for the researcher's own investigation. A psychologist who runs a recall experiment and records participants' scores is collecting primary data. Interviewing those participants about their experience also produces primary data.

Collecting primary data allows the researcher to design the procedure around their own aim and control how information is gathered. For example, they can use the same instructions and scoring rules for everyone. However, recruitment and data collection require time and resources, and the researcher's expectations may influence what they notice or record.

Secondary data is pre-existing information reused for the current investigation. A psychologist who analyses recall results collected by another research team is using secondary data. Existing interview transcripts or diary material can also be used as secondary data.

Secondary data can save time and allow researchers to examine information from several investigations. However, the original researchers may have had different aims, used unsuitable measures or studied a different population. Older data may no longer reflect current circumstances. The researcher therefore needs to check how it was collected and whether it fits the new question.

The same dataset can be primary for the researchers who collected it and secondary for someone who later reuses it. Its numerical or descriptive form does not change simply because a different researcher uses it.

Why psychologists use samples

The target population is the whole group the researcher wants to investigate. A sample is the smaller group selected from it who actually take part. Studying everyone may be impractical because of cost, time or difficulty finding participants.

The target population depends on the research question. It might be office workers in Scotland or people aged 60–85 who have dementia. A sample should reflect the characteristics of that population that matter for the investigation. This is what makes it representative.

Researchers often want to generalise: apply findings from their sample to the wider target population. How confidently they can do this depends partly on how well the sample represents that population.

Judging what a sample can tell us

Consider a study of 25 office workers from Glasgow. If the researcher wants to draw conclusions about office workers throughout Scotland, workers from other areas are missing. Their working conditions and experiences may differ, so the Glasgow findings cannot confidently be assumed to apply to all Scottish office workers.

A sample is biased when relevant groups or characteristics are over-represented or under-represented. Including workers from different parts of Scotland in proportions that reflect the target population could improve representativeness. This is the principle behind stratified sampling.

Sample size also matters. In a small sample, each participant's result has more influence on the overall pattern, so an unusual result can make a substantial difference. A larger sample can make findings less dependent on a few individuals. However, recruiting many more workers from the same narrow group would not necessarily solve the problem of bias.

Small samples can still be useful, particularly for detailed interviews or case studies. The appropriate conclusion may be a detailed account of those participants' experiences rather than a claim about everyone.

When evaluating scientific findings, connect the conclusion to the evidence: identify the intended population, examine who actually participated, consider the sample's size and representativeness, and judge how far the claim can reasonably extend.

Classify data on two dimensions

• Form: quantitative = numerical measurements or counts; qualitative = non-numerical descriptions, words or images.
• Source: primary = collected first-hand for the current investigation; secondary = pre-existing data reused.
• Either source can provide either form of data.

Match data to the research aim

• Quantitative: easy to summarise and compare; may miss detail and personal meaning.
• Qualitative: rich accounts of experience; analysis can be slow and interpretation may vary.
• Primary: control over collection and relevance to the aim; requires time and resources.
• Secondary: saves collection time; check original methods, relevance and age.

Evaluate sampling

• Identify the target population and the sample actually studied.
• Check whether relevant groups are missing or over-represented.
• Larger samples reduce the influence of individual results, but do not automatically remove bias.
• Generalise only as far as the sample reasonably supports.