An experiment in psychology is a study designed to test whether one thing actually causes another. It’s the only research method that can establish cause and effect, which is why it sits at the center of psychological science. What separates an experiment from other types of studies comes down to three features: the researcher manipulates something, measures the outcome, and controls everything else that might interfere.
The Three Requirements of a True Experiment
For a study to count as a true experiment, it needs all three of these elements working together.
First, the researcher manipulates an independent variable, which is the factor they believe will influence behavior. If you’re testing whether sleep deprivation affects memory, the amount of sleep participants get is the independent variable. The researcher decides who sleeps eight hours and who sleeps four. That deliberate manipulation is what makes a study experimental rather than observational.
Second, the researcher measures a dependent variable, which is the outcome they care about. In the sleep example, that might be the number of words participants recall from a list the next morning. The dependent variable is what changes (or doesn’t) as a result of the manipulation.
Third, the researcher controls extraneous variables, meaning all the other factors that could muddy the results. If participants in the sleep-deprived group also happened to drink more coffee, you couldn’t tell whether memory differences came from sleep loss or caffeine. Good experiments hold these outside influences constant so the only meaningful difference between groups is the independent variable itself.
Why Random Assignment Matters
The most powerful tool for controlling extraneous variables is random assignment, where each participant has an equal chance of landing in any group. This sounds simple, but it solves a surprisingly deep problem. People differ in countless ways: motivation, intelligence, mood, health, prior experience. If participants choose their own group, or if the researcher assigns them based on some characteristic, those pre-existing differences could explain the results instead of the manipulation.
Random assignment doesn’t make groups identical. What it does is spread individual differences evenly across groups so that no single factor systematically favors one condition over another. This is why randomized experiments are considered the gold standard for causal claims. Without randomization, you’re left guessing whether the outcome was caused by your manipulation or by something about the people in each group.
Experimental vs. Control Groups
Most experiments split participants into at least two groups. The experimental group receives the treatment or manipulation, while the control group does not. The control group acts as a baseline, showing what happens without the intervention. If participants in the experimental group perform differently from controls, and everything else was held constant, the researcher has evidence that the independent variable caused the difference.
Some experiments use a placebo control, where the control group receives a fake version of the treatment. This helps account for the possibility that simply believing you’re being treated can change behavior or outcomes.
Lab Experiments vs. Field Experiments
Psychology experiments happen in two broad settings, each with trade-offs.
Lab experiments take place in controlled environments where the researcher can minimize outside interference. They’re excellent for precise measurements, like tracking brain activity or reaction times down to the millisecond. The downside is artificiality. A lab is not the real world, and people who know they’re being watched often behave differently. This is sometimes called the Hawthorne effect. Lab studies also tend to use small, narrow samples, which can make it harder to generalize results to the broader population.
Field experiments take place in natural, real-world settings like classrooms, workplaces, or public spaces. They sacrifice some control for realism. Because participants are often unaware they’re in a study, their behavior is more authentic. The challenge is that researchers can’t lock down every variable the way they can in a lab, so unexpected factors can creep in.
Quasi-Experiments and Their Limits
Sometimes true random assignment isn’t possible. You can’t randomly assign people to experience childhood trauma or to have a specific medical condition. In these cases, researchers use quasi-experiments, which look like true experiments but lack randomization. A researcher might compare students in two different schools, or people who chose to meditate versus those who didn’t.
Quasi-experiments are useful when ethical or practical constraints rule out random assignment, but they carry a significant weakness. Because participants aren’t randomly sorted into groups, any observed difference might stem from an unmeasured confounding variable rather than the factor being studied. A quasi-experiment can suggest a causal relationship, but it can’t confirm one with the same confidence as a true experiment.
How Results Are Evaluated
After collecting data, researchers use statistical tests to determine whether the difference between groups is meaningful or could have happened by chance. The conventional threshold in psychology is a p-value below 0.05, meaning there’s less than a 5% probability that the results occurred randomly if no real effect exists. This cutoff traces back to the statistician R.A. Fisher and has been a standard convention for decades, though it’s not an absolute rule.
A p-value below 0.05 doesn’t prove the effect is large or important. It simply means the pattern in the data is unlikely to be pure noise. Researchers also look at confidence intervals and effect sizes to understand how strong and reliable a finding is. At the 0.05 threshold, roughly 1 in 20 tests will produce a false positive even when no real effect exists, which is why replication matters so much.
Controlling for Bias
Even well-designed experiments can be distorted by bias. If participants know which group they’re in, their expectations can shape their behavior. If researchers know which participants received the real treatment, they might unconsciously interpret results more favorably for that group.
The double-blind procedure addresses both problems at once. Neither the participants nor the researchers interacting with them know who is in the experimental group and who is in the control group. This prevents participants from adjusting their behavior based on expectations and prevents researchers from subtly treating the two groups differently. Double-blinding minimizes observer bias and confirmation bias, making results more trustworthy.
Internal and External Validity
Two concepts help evaluate how much you can trust an experiment’s conclusions. Internal validity refers to whether the study was designed and carried out in a way that supports a clean causal interpretation. Threats to internal validity include selection bias (groups that differ in important ways before the study starts), attrition bias (one group losing more participants than the other), and detection bias (measuring outcomes inconsistently across groups).
External validity refers to whether the findings apply beyond the specific study. An experiment conducted entirely on college students in a single university lab may produce internally valid results that don’t generalize to older adults, different cultures, or real-world conditions. Studies with narrow participant pools, short time frames, or strict exclusion criteria tend to have weaker external validity. The ideal experiment scores high on both, but in practice, tightening control to boost internal validity often comes at the cost of real-world applicability.
Ethics in Psychological Experiments
Every psychological experiment involving human participants must follow ethical standards set by the American Psychological Association. The most fundamental requirement is informed consent. Before participating, individuals must be told the purpose of the research, expected duration, procedures involved, any foreseeable risks or discomfort, their right to withdraw at any time without penalty, the limits of confidentiality, and who to contact with questions.
For experiments involving treatments or interventions, participants must also be told the experimental nature of the treatment, what alternatives are available, and how group assignment will work. These rules exist because psychology’s history includes studies that caused real harm to participants, and modern standards are designed to prevent that from happening again.
Pre-Registration and Transparency
In recent years, psychology has grappled with the finding that many published results couldn’t be replicated by other researchers. One major response has been pre-registration: formally recording a study’s hypotheses, methods, sample size, and planned analyses in a public repository before collecting any data. This prevents researchers from adjusting their methods or cherry-picking results after seeing the data, a practice that inflates false positive rates.
Pre-registered studies typically specify group sizes (with statistical power calculations), inclusion and exclusion criteria, experimental tasks, how participants will be assigned to conditions, and the exact statistical tests that will be used. The more details locked in beforehand, the greater the study’s transparency and the more credible its findings. Pre-registration doesn’t guarantee good research, but it makes it much harder to present exploratory findings as if they were predicted all along.