Experiments

What an experiment is

An experiment is a controlled test. The American Psychological Association's Dictionary of Psychology defines an experiment as a series of observations conducted under controlled conditions to study a relationship, with the purpose of drawing causal inferences about that relationship. In plainer language, the Noba Project's Research Designs module puts it this way: in experiments, researchers actively make changes in one variable and watch for changes in another variable. That is the gold-standard idea. Psychology reaches its strongest conclusions — this caused that — through experiments, because the researcher, not nature, decides what happens to whom and when.

An original example holds the shape together. Suppose a researcher wants to know whether a short nap helps people stay alert. She cannot simply ask people how alert they feel after napping, because people who nap may differ from people who do not in a hundred ways. Instead she runs an experiment: some participants nap for ten minutes, others do not, and everyone then works on the same timed puzzle. The nap is the one thing that differs between the two conditions; everything else — the room, the hour, the puzzle, the instructions — is the same. If the nappers solve more puzzles, the difference has one clear suspect.

The variables: independent, dependent, and control

Every experiment runs on three kinds of variables. The independent variable is the one the researcher deliberately changes — in the nap study, whether a participant napped. The dependent variable is the one the researcher measures afterward to see whether anything happened — the number of puzzles solved. The names encode the logic: the outcome depends on the manipulation. The APA dictionary defines the independent variable as the variable that is specifically manipulated in order to assess its effect, and the dependent variable as the outcome observed to change after variation of the independent variable.

The third kind is quieter but just as important. A control variable is something that could affect the measured outcome but is not what the study is about; the researcher holds it constant so it cannot be the cause of any difference. The same room, the same time of day, the same puzzle difficulty — each is a control variable. Hold them steady and they are removed from the running as explanations. The nap study's clean result depends on the control variables staying truly constant.

Experimental group, control group, and random assignment

Participants in an experiment are divided into groups. The experimental group is exposed to the manipulation of the independent variable — the nap. The control group is the comparison: its members receive no intervention, or an established one, so it shows what happens without the manipulation. In the nap study, the control group simply waits quietly for ten minutes and then does the same puzzle. The experimental group's score means little on its own; it means something only against the control group's score.

Who lands in which group cannot be left to preference, because people who volunteer for a nap and people who refuse it may differ in alertness before the experiment even starts. So the researcher decides by chance: random assignment. The APA dictionary defines it as assigning participants to conditions entirely at random, so that each participant has an equal likelihood of being assigned to any particular condition. A coin flip, a die roll, or a random-number list will do. The honest reason this matters: random assignment makes the two groups approximately comparable in every respect except the treatment — the nappers and the non-nappers end up with similar ages, sleep histories, and moods on average. When the groups start out comparable and only one thing differs, a difference at the end has one credible cause.

The logic, the limits, and the honest framing

The core argument is short: change one thing, hold everything else steady, compare. If the groups differed only in the manipulation and were comparable before it, then the manipulation is the best explanation for any difference in the outcome. That is the experiment's claim to the title of gold standard.

The honest note is that the claim comes with two limits. First, experiments can be artificial. As OpenStax's Psychology textbook observes, experiments are often conducted in very artificial settings, which calls into question how well the findings apply to real-world settings. A quiet lab room with a word list is not a dorm hallway with a phone buzzing. Second, some experiments are impossible — practically or ethically. A researcher cannot randomly assign children to different parenting styles and follow them for twenty years, however much she would like an answer. Where the manipulation cannot be done, experiments cannot go, and psychology must lean on other methods with weaker claims.

That is the reality check: experiments trade realism for control. You gain the ability to say “this caused that”; you pay for it in settings that may not match ordinary life. Reading an experimental result well means asking both questions at once — did they control enough, and does this artificial setting still tell us about the real world?