A study can observe a difference between groups without assigning the exposure that separates them. Another study can deliberately assign an intervention and measure what follows. Distinguishing these designs helps identify the alternative explanations a result must address, but the label alone does not establish that a causal claim is justified.
Ask who determined the exposure of interest
NLM’s publication-type definitions describe an observational study as one in which investigators do not assign participants to the specific interventions being studied. Participants may still receive interventions through other circumstances, such as ordinary care.NLM — Publication characteristics with scope notes (opens in a new tab)
The corresponding design question applies more broadly: did the research plan assign the factor whose effect is being evaluated, or did the investigators measure an existing difference? Collecting a sample is not itself proof that the exposure of interest was assigned.
| Design question | What to look for |
|---|---|
| Was exposure assigned? | The study plan and allocation description |
| Was assignment random? | A stated randomisation method |
| Was exposure already present? | How the observed groups arose |
| When was the outcome measured? | Timing relative to the exposure and selection |
Keep assignment and measurement separate. A study can use sophisticated laboratory measurements while its central exposure comparison remains observational.
Work through an observed association
In an original hypothetical study, samples with higher endogenous peptide levels also have higher values of a biological marker. That is an observed association between two measurements.
Several explanations remain possible: the peptide could influence the marker, the underlying biological state could influence both, or the marker-related process could influence peptide levels. The association alone does not choose among them.
A measured adjustment for one background variable can address a specific alternative explanation under its assumptions. It does not prove that every relevant confounder was measured correctly or that the statistical model fully represents the relationships.
Read the assigned comparison carefully
Now imagine a separate laboratory experiment assigning a peptide condition or comparator to suitable experimental units. That intervention changes the design question, but the credibility of its effect estimate still depends on allocation, controls, measurement and analysis.
If all peptide-condition samples are processed on one day and all comparators on another, the observed difference can also reflect the day. Assigning an intervention does not remove every systematic difference introduced by the design.
Randomised allocation can reduce selection-related differences, while masking can reduce influences during handling or assessment. They address distinct problems; the word experiment does not tell you whether either was used.
An assigned intervention also establishes evidence only for the system and comparison studied. A controlled laboratory result should not be promoted into a clinical effectiveness claim without the appropriate additional evidence.
State the inference the design supports
For the hypothetical observational study, an accurate summary says that the two measurements were associated in the sampled population and describes the main limitations. It should not replace associated with caused unless the broader causal analysis supports that conclusion.
For the experiment, describe the estimated difference between assigned conditions and the features that make the comparison interpretable. Do not omit a design limitation merely because assignment occurred.
Observational evidence can help describe natural variation, identify hypotheses and examine questions that cannot be addressed by assignment. Its value depends on the question and methods, rather than occupying a universally useless category.
When combining evidence, preserve these roles. Agreement between observational and interventional findings can be informative, but their separate assumptions and populations remain part of the interpretation.
Sources and further detail
- NLM — Publication characteristics with scope notes (opens in a new tab)
Observational Study definition read. Clinical terminology is used to establish the assignment distinction, not to classify a real trial. All association and laboratory scenarios are original hypothetical examples.
Sources checked 19 September 2026. Worked examples are illustrative unless a supplied report is explicitly identified. This article has not undergone independent scientific peer review.