Contents
Academy · Research Data Analytics & AI · Experimental research design
Experimental Research Design: Definition, Types and Examples
An experimental design manipulates one or more variables and measures the consequences. Random allocation and a control condition enable such a design to provide evidence of causation, and the remaining elements of the design exist to preserve these two features.
Definition of experimental design
In an experimental design, the researcher manipulates one or more independent variables and measures the effect on a dependent variable, while holding other influences constant or distributing them evenly between conditions. It is the only family of designs that supports a causal claim, and it does so by virtue of three features rather than the statistics applied afterwards.
- Manipulation — the independent variable is set by the researcher, not observed.
- Control — a comparison condition against which the effect is measured, and control of extraneous variables that would otherwise confound it.
- Randomisation — allocation by chance, which distributes both known and unknown differences evenly across conditions.
Components of an experiment
Five elements must be specified before the study opens, and imprecision in any one of them is the source of most reviewer queries.
- Independent variable — what is manipulated, and at which levels.
- Dependent variable — what is measured, with the instrument and the time point stated.
- Control condition — placebo, standard care, or an untreated group.
- Extraneous and confounding variables — what is held constant, what is measured and adjusted for, and what is left to randomisation.
- Allocation — the randomisation method, the sequence generation, and whether allocation was concealed.
Principal types of experimental design
Between-subjects (independent groups). Each participant experiences one condition. The design is straightforward to analyse and free of carryover effects, but it requires more participants and relies on randomisation to make the groups comparable.
Within-subjects (repeated measures). Each participant experiences every condition and therefore serves as their own control. The design is considerably more efficient but is susceptible to order and carryover effects, which counterbalancing or a washout period is intended to address.
Matched pairs. Participants are paired on a relevant variable, and one member of each pair is randomised to each condition. The design combines some of the efficiency of within-subjects designs with the freedom from carryover of between-subjects designs.
Factorial. Two or more independent variables are manipulated together, which is the only means of detecting an interaction, that is, an effect of one variable that depends on the level of another.
Randomised controlled trial. The clinical application of the between-subjects design, with allocation concealment, blinding where possible, a pre-registered protocol and a pre-specified primary outcome.
Crossover trial. A within-subjects design in clinical research, suitable only for stable conditions and treatments whose effect does not persist into the next period.
Selection among experimental designs
The trade-off in one line
Within-subjects designs gain statistical power at the risk of carryover. Between-subjects designs avoid carryover at the cost of additional participants. Factorial designs answer more questions from the same sample but require sufficient participants in every cell to detect the interaction, for which they are usually under-powered.
Whichever design is chosen, the choice and its justification belong in the protocol, because the design determines the analysis, the sample size and which reporting checklist governs the write-up.
Reporting of experimental studies
A randomised trial is reported under CONSORT 2025, with prospective registration and a participant flow diagram 1. Where the design departs from a parallel-group trial (cluster, crossover, non-inferiority, or pilot designs), an extension applies, and the journal’s instructions for authors will specify it.
Laboratory experiments on animals are reported under ARRIVE 2.0 2. Non-randomised experimental work in humans usually follows TREND (Transparent Reporting of Evaluations with Nonrandomized Designs) or STROBE (Strengthening the Reporting of Observational Studies in Epidemiology), depending on the journal. Naming the standard in the methods and completing the checklist before submission is the least costly quality step available.
Common sources of error in experiments
- Allocation that is not concealed — randomisation is only as effective as the concealment of the next assignment.
- Outcomes changed after the fact — the single most common reason a trial is questioned after peer review has otherwise gone well.
- Under-powering — especially in factorial designs, where the interaction needs considerably more participants than the main effects.
- Attrition handled silently — participants lost to follow-up must be accounted for in the flow diagram and in the analysis population.
- Ecological validity traded away — tight control can produce an effect that holds in the laboratory but not elsewhere, a limitation that should be addressed in the discussion.
Key points for authors and researchers
- Manipulation, control and randomisation, rather than the statistical analysis applied afterwards, are what justify a causal claim.
- Within-subjects designs increase statistical power at the risk of carryover effects; between-subjects designs avoid carryover at the cost of requiring more participants.
- Allocation concealment is as important as randomisation itself.
- The primary outcome should be fixed before the study opens; changing it afterwards is the most common reason a trial is questioned.
- Factorial designs require substantially more participants to detect an interaction than to detect main effects.
Frequently asked questions
Randomisation. A quasi-experimental design manipulates an independent variable and keeps a comparison group, but allocation is not random, so the groups may differ at the outset and the causal claim is correspondingly weaker.
For a causal claim, yes; a comparison condition is required to represent what would have occurred without the intervention. Single-arm designs have a place in early-phase work, and their results are reported as estimates with confidence intervals rather than as comparisons.
It depends on the effect size considered meaningful, the variability of the outcome, the desired power, and the design. It is a calculation that belongs in the protocol, and a factorial design requires considerably more participants than a two-arm design.
Whenever the first condition could alter a participant’s response to the second, for example through learning, fatigue, or a treatment whose effect persists. If a washout period cannot be justified, a between-subjects design should be used.
No. If nothing is manipulated it is an observational study, however sophisticated the analysis. Methods such as target trial emulation strengthen causal inference from observational data but do not make it an experiment.
References
- CONSORT 2025 statement and checklist, via the EQUATOR Network. read 2 October 2026. www.equator-network.org
- ARRIVE 2.0 guidelines for reporting animal research. read 2 October 2026. arriveguidelines.org
- ICH E9(R1) addendum on estimands and sensitivity analysis in clinical trials. read 2 October 2026. www.ich.org/page/efficacy-guidelines
- Campbell DT, Stanley JC. Experimental and quasi-experimental designs for research. Rand McNally, 1963. read 2 October 2026. search.worldcat.org/search?q=Campbell+Stanley+experimental+and+quasi-experimental+designs+for+research
Each source carries the date it was last read. Where a standard moves, this article is revised rather than left up with a note.
How to cite this article
Pick your referencing style and copy it. Your university or target journal may vary the punctuation — check its guide if it has one.
Pubrica Academy. (2026, October 2). Experimental Research Design: Definition, Types and Examples. Pubrica. https://pubrica.com/academy/experimental-design/experimental-research-design-definition-types-examples/
Pubrica Academy. “Experimental Research Design: Definition, Types and Examples.” Pubrica, 2 Oct. 2026, pubrica.com/academy/experimental-design/experimental-research-design-definition-types-examples/.
Pubrica Academy (2026) Experimental Research Design: Definition, Types and Examples. Pubrica. Available at: https://pubrica.com/academy/experimental-design/experimental-research-design-definition-types-examples/ (Accessed: 15 October 2026).
Pubrica Academy. “Experimental Research Design: Definition, Types and Examples.” Pubrica. October 2, 2026. https://pubrica.com/academy/experimental-design/experimental-research-design-definition-types-examples/
Pubrica Academy, ‘Experimental Research Design: Definition, Types and Examples’ (Pubrica, 2 October 2026) <https://pubrica.com/academy/experimental-design/experimental-research-design-definition-types-examples/> accessed 15 October 2026.
Pubrica Academy. Experimental Research Design: Definition, Types and Examples. [Internet]. Pubrica; 2026 Oct 2 [cited 2026 Oct 15]. Available from: https://pubrica.com/academy/experimental-design/experimental-research-design-definition-types-examples/
Pubrica Academy, “Experimental Research Design: Definition, Types and Examples,” Pubrica, Oct. 2, 2026. [Online]. Available: https://pubrica.com/academy/experimental-design/experimental-research-design-definition-types-examples/
@misc{experimental_research_design_definitio_2026,
author = {{Pubrica Academy}},
title = {Experimental Research Design: Definition, Types and Examples},
howpublished = {Pubrica},
year = {2026},
month = {oct},
url = {https://pubrica.com/academy/experimental-design/experimental-research-design-definition-types-examples/},
note = {Accessed: <span data-cite-acc="iso">2026-10-15</span>}
}This article is written to be learned from. Cite it if you are referring to what it says; if you are using the evidence behind it, cite the sources listed above instead.
Pubrica AcademyWritten and maintained by Pubrica’s subject matter experts. Articles are revised when the underlying standards change, and each article states the date on which it was last reviewed.
How Pubrica can help
Design is the one stage of a study that cannot be revised retrospectively. Statistical analysis plan specifies the analysis before the data exist, Pubrica’s biostatistics team calculates the required sample size rather than estimating it, and reporting checklist review assesses the completed manuscript against the reporting standard that the design requires.