Field Experiment — Randomized Experiment in Real-World Settings
Field Experiment · Also known as: field trial, natural field experiment, randomized field experiment, field RCT
A field experiment applies the logic of a randomized controlled trial in a naturally occurring, real-world environment rather than an artificial laboratory. Participants are randomly assigned to treatment and control conditions while going about everyday activities, allowing researchers to estimate causal effects with high internal validity while preserving a level of ecological realism that laboratory settings cannot offer. The design is especially prominent in economics, public health, political science, and development research.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
+8 more
When to use it
Use a field experiment when the goal is to establish a causal effect of a clearly defined intervention in a real-world context, and when random assignment to conditions is ethically and logistically feasible. It is the preferred design when high ecological validity is required — when knowing that an effect occurs in the lab is not sufficient, and the question is whether it holds in everyday behavior. Do not use a field experiment when randomization is ethically prohibited (e.g., withholding a proven lifesaving treatment), when the intervention cannot be standardized across sites, when the sample is too small to achieve adequate statistical power, or when the research question is exploratory and causal inference is not yet the goal — in those cases, observational or qualitative designs are more appropriate.
Strengths & limitations
- Strong internal validity: random assignment rules out confounding and supports causal conclusions.
- High ecological validity: results reflect real-world behavior, not artificial laboratory compliance.
- Can be scaled to large, representative samples drawn from real populations.
- Widely accepted as a credible evidence standard by policymakers, funders, and systematic reviewers.
- Flexible design options (stratification, clustering, factorial arms) allow complex questions to be addressed.
- Logistically demanding and often expensive: delivering treatments and collecting follow-up data in the field requires substantial coordination.
- External validity still depends on the specific population and context studied — results from one setting may not transfer to another.
- Randomization is not always ethically or politically feasible; gatekeepers may refuse to allow random denial of a service.
- Hawthorne effects (participants changing behavior because they know they are being observed) can inflate treatment estimates.
- Attrition and non-compliance weaken internal validity if not anticipated and handled in the analysis plan.
Frequently asked
What distinguishes a field experiment from a natural experiment?
In a field experiment, the researcher actively assigns treatment — randomization is under the researcher's control. In a natural experiment, an external event or policy change creates an as-if-random assignment that the researcher exploits after the fact but did not engineer. Field experiments offer stronger control over the assignment mechanism; natural experiments rely on the credibility of the as-if-random assumption.
What distinguishes a field experiment from a laboratory experiment?
Laboratory experiments bring participants into an artificial setting where the researcher controls the environment tightly. Field experiments are conducted in participants' everyday environments without removing them from their normal context. Laboratory designs offer tighter experimental control; field designs offer greater ecological validity. Harrison and List (2004) further distinguish artefactual, framed, and natural field experiments by the degree to which the experimental nature of the study is disclosed to participants.
How large does my sample need to be?
Sample size depends on the expected effect size, the variance of the outcome, the desired power (typically 80% or 90%), and the significance level. Clustered designs require larger total samples than individually randomized ones. Use a power calculator before recruitment — underpowered field experiments are among the most common and costly design mistakes in this literature.
Is it ethical to withhold a potentially beneficial treatment from the control group?
Ethical acceptability depends on context. If there is genuine uncertainty about whether a treatment is effective (equipoise), randomization is defensible. If the treatment is already known to be effective, withholding it from a control group raises ethical concerns. Waitlist control designs — where the control group receives the intervention after the measurement period — are a common ethical solution.
Can I combine field experiment data with observational data?
Yes. Combining experimental and observational data can improve external validity or generalizability if the experimental sample is unrepresentative. Methods such as reweighting or transporting treatment effects require assumptions about the overlap between the experimental and target populations, and should be pre-specified.
Sources
- Harrison, G. W., & List, J. A. (2004). Field experiments. Journal of Economic Literature, 42(4), 1009–1055. DOI: 10.1257/0022051043004577 ↗
- Gerber, A. S., & Green, D. P. (2012). Field Experiments: Design, Analysis, and Interpretation. W. W. Norton. ISBN: 978-0393979954
How to cite this page
ScholarGate. (2026, June 3). Field Experiment. ScholarGate. https://scholargate.app/en/experimental-design/field-experiment
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Cluster Randomized Controlled TrialExperimental design↔ compare
- Factorial ExperimentExperimental design↔ compare
- Laboratory ExperimentExperimental design↔ compare
- Natural ExperimentExperimental design↔ compare
- Randomized Controlled TrialExperimental design↔ compare