Primary and Secondary Data Sources
Where Does Data Come From?
Every statistical investigation needs data — but that data has to come from somewhere. The two fundamental categories are primary data and secondary data. Understanding the difference between them — and knowing the advantages and limitations of each — is essential when planning an investigation and when evaluating someone else’s research.
Primary Data
Primary data is collected first-hand, specifically for your investigation. The main methods include:
- Experimental: The researcher controls conditions to test a cause-and-effect relationship. A laboratory experiment removes external influences (high level of control). A field experiment takes place in a natural setting, which improves realism but reduces control. A natural experiment observes what happens when conditions change for non-research reasons — for example, observing the effect of a new bus route on school attendance rates.
- Simulation: Uses models (often computer-based) to generate data when real-world collection is impossible or unethical — for example, simulating traffic patterns or the spread of an infectious disease.
- Questionnaire: A set of written questions completed by participants. Cheap and scalable, but dependent on honest responses.
- Observation: Watching and recording behaviour directly. Can capture what people actually do rather than what they say they do, but observer presence can influence behaviour.
Primary data has a major advantage: it is collected specifically for your purpose, so it should measure exactly what you need. The disadvantage is that it takes time and resources to collect.
Secondary Data
Secondary data already exists. Sources include published statistical databases, government censuses, academic studies, media reports, and reference publications. In the UK, organisations such as the Office for National Statistics (ONS), NHS Digital, the Met Office, and the Department for Education all publish large, high-quality datasets that researchers can use without collecting any data themselves.
The advantages of secondary data are significant: it is fast to access, often covers large populations, and may span decades of historical records that no individual researcher could replicate. The census in England and Wales, for example, is conducted every ten years and records information about people and households — a dataset of that scale is simply not achievable through primary collection for most researchers.
However, secondary data has limitations. It was collected for a different purpose, so it may not measure exactly what you need. Variable definitions may differ — what counts as “unemployed” changed in UK statistics several times between the 1970s and 1990s, making long-run comparisons tricky. You also need to acknowledge the source of any secondary data you use, as the specification requires.
Secondary data: advantages — fast to access, large sample sizes, historical depth. Limitations — may not measure exactly what you need, variable definitions may have changed, source must be acknowledged and its quality assessed.
Census vs Sample
A census collects data from every member of the population. The census in England and Wales, conducted every ten years, aims to record information about every household. A census gives complete data with no sampling error, but it is expensive, time-consuming, and in some cases impossible (not every person can be reached).
In contrast, sampling involves collecting data from a carefully chosen subset. Most practical investigations use samples. The challenge is making the sample representative enough that conclusions about the population are valid.
Reliability and Validity
These two words are used precisely in statistics and are often confused. Reliability is about consistency: does the measurement give the same result every time you use it? Validity is about accuracy of purpose: does the measurement capture the thing you actually want to know about?
A useful way to see the difference: imagine a bathroom scale that always reads 3 kg too heavy. It is reliable — it gives a consistent reading every time — but it is not valid for measuring your true weight. Conversely, asking students “How healthy do you feel?” on a scale of 1–10 might produce variable answers from the same student on different days (poor reliability) and may not accurately capture their clinical health status (poor validity).
Students frequently write “the data is reliable and valid” as if the two words mean the same thing, or they use one when they mean the other. In the exam, using these terms imprecisely will cost marks.
If a question asks about reliability, focus on: would repeated measurements give the same result? If it asks about validity, focus on: does the measurement actually capture what you want to measure? A questionnaire asking about income to proxy for social class is valid if income correlates well with social class — but only reliable if the same person answers the same question consistently across administrations.
Bias and Sensitivity
Bias is a systematic error — it pushes results in one direction rather than producing random noise. Bias can arise from: the way a sample is selected (sampling bias), the way questions are worded (response bias), or the sensitivity of the topic being researched.
Sensitive topics — income, health, political views, personal relationships — are particularly prone to response bias. People may give socially desirable answers rather than honest ones, or they may refuse to answer at all (non-response bias). When the people who refuse to answer share a characteristic relevant to the investigation, this creates a systematic distortion in the results.
Level of Control and Minimising Data Distortion (Higher)
In a controlled experiment, the researcher manipulates one variable (the explanatory variable) and keeps all others constant, so that any change in the response variable can be attributed to the manipulation. The level of control refers to how completely the investigator can hold other variables constant.
High control (laboratory experiments) makes causation easier to claim but may reduce ecological validity — the results may not apply to real-world settings. Low control (natural experiments, observational studies) reflects real-world conditions better but makes it harder to isolate cause and effect because other variables are free to vary.
Extraneous variables — variables that influence the response variable but are not the focus of the investigation — must be identified and managed. Techniques include controlling them directly (holding them constant across all participants), including a control group (a group that receives no treatment, against which the experimental group is compared), and randomisation (randomly assigning participants to groups so that extraneous variables are spread evenly).
Reliability = consistency across repeated measurements. Validity = the measurement captures what it is supposed to measure. A data source can be one without the other.
The specification explicitly states that secondary data sources should be acknowledged. In practice, this means recording where the data came from, when it was collected or published, and who collected it. This matters for two reasons.
First, it allows the reader to check the original source and assess its quality. A statistic from a government body such as the ONS is more likely to be reliable than one from a company’s own press release. Second, without acknowledgement, using someone else’s data without credit is a form of academic misconduct. In professional statistical work and in your investigation write-up, always note the source.