Choose data that can answer the question
Primary data are collected specifically for the current study through surveys, interviews, observation, experiments or measurements. Secondary data already exist, often as administrative records, national surveys, organisational databases or open repositories. Neither option is automatically stronger.
The defensible choice is the source that best represents the constructs, population, period and level of detail required by the research question.
Compare the trade-offs
Primary data provide greater control over instruments, sampling and timing, but collection can be expensive, slow and vulnerable to low response. Secondary data can offer wider coverage, long time series and lower cost, but the researcher inherits the original definitions, sampling decisions, missing values and quality limitations.
- Use primary data when the required variables or experiences are not already captured.
- Use secondary data when credible sources already cover the needed population, period and indicators.
- Combine both when new contextual evidence is needed to explain patterns found in an existing dataset.
Audit the source before analysis
Document who collected the data, why it was collected, how units were sampled, how variables were defined and whether revisions occurred. Check licences, confidentiality rules, geographic and temporal coverage, missingness and comparability.
A large dataset does not correct weak measurement. State limitations openly and avoid conclusions that exceed the design or variables available.