A true experimental design is a research setup where participants are randomly assigned to groups, at least one group receives a treatment or intervention, and the results are compared against a control group that does not. This combination of features makes it the strongest method for establishing cause and effect. If you can say “X caused Y” rather than “X is associated with Y,” a true experiment is almost certainly behind that claim.
The Three Core Features
True experiments are built on three requirements that work together. Remove any one of them, and you no longer have a true experimental design.
- Manipulation of a variable. The researcher deliberately changes something, called the independent variable, to see what effect it has. This could be a new teaching method, a therapy technique, a drug, or access to a service. The key is that the researcher controls who gets exposed and who doesn’t.
- A control group. At least two groups exist: one that receives the intervention (the treatment group) and one that does not (the control group). Comparing outcomes between these groups is what allows researchers to isolate the effect of the intervention from everything else going on.
- Random assignment. Every participant has an equal chance of ending up in either group. This is the feature that separates true experiments from all other designs, and it deserves its own section.
Why Random Assignment Matters So Much
Random assignment is the single ingredient that gives true experiments their power. It means that no pattern exists between who ends up in the treatment group and any characteristic of that person. Age, motivation, health status, income, personality: all of these get distributed roughly evenly across both groups purely by chance.
This solves a problem that plagues nearly every other type of study. Without randomization, the groups might differ in ways the researcher can’t see or measure. Imagine testing a new reading program by letting students volunteer for it. The kids who sign up are probably more motivated to begin with, so any improvement you observe could be the motivation, not the program. Random assignment eliminates that ambiguity. It controls for both known and unknown variables that could otherwise confuse the results.
Random assignment is not the same as random selection. Random selection refers to how participants are chosen from a larger population, which affects whether your findings generalize broadly. Random assignment refers to how those participants are split into groups, which affects whether you can claim cause and effect. A study can have one without the other.
Common True Experimental Designs
Pretest-Posttest Control Group Design
This is the most widely recognized format. Both groups are measured before the experiment begins (the pretest), then the treatment group receives the intervention while the control group does not, and both groups are measured again afterward (the posttest). The pretest establishes a baseline, so researchers can compare how much each group changed. If the treatment group improved significantly more than the control group, and both groups started at a similar level, that improvement can be attributed to the intervention.
The pretest also serves as a safety check on randomization. Even with random assignment, small differences between groups can occur by chance. A pretest lets researchers verify that the groups were roughly equivalent before the intervention started.
Posttest-Only Control Group Design
Sometimes a pretest is impractical or could influence the results. If you’re testing whether a video changes people’s attitudes, asking them about those attitudes right before showing the video might prime them to think differently. In a posttest-only design, participants are randomly assigned to groups, the treatment is administered, and then both groups are measured. There is no baseline measurement.
This design works because random assignment, when done properly, makes the groups equivalent before the study begins. Any difference in posttest scores between the two groups can be attributed to the intervention rather than preexisting differences. The tradeoff is that you can’t verify baseline equivalence the way a pretest allows, so this approach works best with larger sample sizes where randomization is more reliable.
Solomon Four-Group Design
This design addresses a subtle problem: what if taking the pretest itself changes how participants respond to the treatment? It uses four groups instead of two. One treatment group gets a pretest, another does not. One control group gets a pretest, another does not. All four groups take the posttest. By comparing outcomes across the four combinations, researchers can detect whether the pretest interacted with the treatment. If pretested participants respond differently to the intervention than non-pretested participants, that interaction shows up in the data. It’s the most thorough design but requires twice the participants, so it’s used less frequently.
What True Experiments Protect Against
The value of a true experiment lies in how many alternative explanations it rules out. Research methodologists have identified seven classic threats to a study’s internal validity, meaning its ability to correctly attribute an outcome to the intervention rather than something else. True experiments address the most important of these.
Selection bias is the most obvious threat. If people in the treatment group differ systematically from those in the control group before the study starts, any observed effect could be due to those preexisting differences. Random assignment directly eliminates this.
History refers to outside events that happen during the study. If a public health campaign launches in the middle of your experiment, both groups are equally exposed to it, so it can’t explain differences between them. Maturation covers natural changes over time: people get older, heal on their own, or simply have better days. Again, both groups experience these changes equally. Testing effects, where taking a pretest improves performance on the posttest just through practice, are handled the same way: both groups take the pretest, so any practice benefit applies equally.
These protections are why true experiments sit at the top of the evidence hierarchy for causal claims. When a study concludes that an intervention actually caused an outcome, it’s because the design eliminated every other plausible explanation.
How True Experiments Differ From Quasi-Experiments
The dividing line is straightforward: quasi-experiments lack random assignment. Everything else can look identical. A quasi-experiment might have a treatment group, a control group, a pretest, and a posttest. But if participants weren’t randomly assigned to those groups, the design is quasi-experimental.
This matters because without randomization, preexisting differences between the groups could be mistaken for effects of the intervention. As one review in Psychiatry Research put it, pre-existing differences between treatment and control groups are “especially likely to occur” in designs that lack randomization. Quasi-experiments use statistical techniques to try to account for these differences, but they can never fully eliminate the possibility that some unmeasured variable is driving the results.
Quasi-experiments aren’t inferior by choice. They exist because true experiments aren’t always possible. You can’t randomly assign people to smoke or not smoke, to experience poverty or not, to have a particular diagnosis. In these situations, quasi-experimental designs offer the best available evidence. But when random assignment is feasible, a true experiment provides stronger grounds for causal conclusions.
Limitations of True Experiments
Despite their strengths, true experiments come with real constraints. The most significant is that tightly controlled conditions can create findings that don’t translate to the real world. Laboratory studies of how drugs affect thinking and reaction time, for example, test relaxed, rested, healthy volunteers in a quiet room. That’s a poor stand-in for stressed patients navigating daily life. The internal validity is high (you know the drug caused the measured effect), but the ecological validity, whether those results hold outside the lab, can be low.
Ethical boundaries also limit where true experiments can go. Randomly assigning people to harmful conditions, withholding a known effective treatment, or denying access to essential services are all off the table. This is why much of what we know about the health effects of smoking, pollution, and childhood adversity comes from observational studies rather than experiments.
Practical challenges add further friction. True experiments can be expensive, time-consuming, and difficult to maintain. Participants drop out, contamination between groups occurs when control group members gain access to the treatment, and blinding (keeping participants unaware of their group assignment) isn’t always possible. Each of these issues can chip away at the clean causal story the design is built to produce.