Definition · Plain-language
Observational study
An observational study measures variables and outcomes as they occur naturally, without the researcher intervening, manipulating conditions or assigning participants to groups.
The step most authors miss
Doing CRediT right? Don’t stop at the statement.
A CRediT statement credits you inside one paper. The recognition CRediT was built for happens when those roles are tied to you, persistently. Sign in with your ORCID — free — and claim your CRediT contributions on casrai.org, the home of the standard. They become a verified, portable part of your identity, not a line that disappears into one PDF.
Free: claim your contributions, then export a journal-ready CRediT statement, schema.org structured data, JATS XML, CSV or BibTeX — and preview your public profile. A membership publishes that profile publicly and verifies the journals you serve.
Watching without intervening
The defining feature of an observational study is the absence of intervention: the researcher records what happens without manipulating the independent variable or assigning participants to conditions. People or events are studied in their existing circumstances — patients who already smoke or do not, communities that already differ — and the researcher measures exposures and outcomes as they stand. This contrasts sharply with an experiment, where conditions are deliberately created and controlled. Observational designs are essential whenever manipulation would be unethical (you cannot assign people to smoke) or impossible, and they keep behaviour in its natural context.
Types of observational study
Observational research spans several forms. In quantitative epidemiology and social science, cohort, case-control and cross-sectional studies are all observational designs that measure exposures and outcomes without intervention. In qualitative and behavioural research, observation describes a data-collection stance: naturalistic observation watches behaviour in its everyday setting without interference; participant observation has the researcher join and take part in the group being studied; and structured observation records predefined behaviours against a coding scheme. Observation may be overt, with participants aware they are watched, or covert, raising distinct ethical considerations around consent and privacy.
Association, not proof of cause
Because observational studies do not assign exposures at random, the groups being compared often differ in more ways than the one of interest, so confounding variables provide ready alternative explanations for any association found. This is why an observational finding — that a behaviour and an outcome occur together — shows correlation but cannot on its own prove causation. Well-designed observational studies reduce this limitation through matching, statistical adjustment, and careful measurement, and they remain the only feasible route to many important questions. A further concern is the observer effect, where being watched changes the behaviour under study.
Key facts
At a glance
- Definition: measures variables as they occur naturally, without intervention
- Key absence: no manipulation and no random assignment
- Used when: experiments are impractical or unethical
- Types: naturalistic, participant and structured observation; cohort, case-control, cross-sectional
- Strength: studies behaviour and exposures in their real context
- Limitation: confounding means it shows association, not proven causation
Common misconceptions
What people often get wrong
Often heard: An observational study can prove that one thing causes another.
Actually: It can establish association but not causation on its own. Because exposures are not randomly assigned, confounding variables may explain the link, so causal claims need experiments or strong additional evidence.
Often heard: Observational study just means qualitatively watching people.
Actually: It is broader. The term covers any non-interventional design, including quantitative cohort, case-control and cross-sectional studies, as well as naturalistic and participant observation in qualitative research.
Often heard: Observing participants has no effect on what they do.
Actually: It can. The observer effect means people may alter their behaviour when they know they are being watched, which is one reason researchers sometimes use covert observation or unobtrusive measures.
Going deeper
