Experimental Design — I can describe key features of experiments (IV/DV, control, random assignment) and why they support causal claims.

Historical Context & Motivation

For most of human history, people explained behavior through intuition, philosophy, or tradition. If someone asked, "Does punishment reduce bad behavior?" the answer typically came from personal experience or religious authority—not from systematic testing. The problem is that personal experience is full of biases. You notice the times punishment seemed to work and forget the times it didn't. Psychology needed a more reliable method, one borrowed from the natural sciences: the experiment.

The idea of conducting controlled experiments on human thought and behavior took centuries to develop. Early physiologists studied reflexes and reaction times in laboratories, but it wasn't until psychologists formalized rules—manipulating one factor, holding others constant, and randomly assigning participants—that the discipline earned its place among the sciences. These rules are what make the difference between a convincing finding and mere speculation.

The central question that experimental design addresses is deceptively simple: How can we tell whether one thing actually causes another? Observations and surveys can show that two things go together (correlation), but only a well-designed experiment can tell us that changing one thing produces a change in another. The rest of this lesson unpacks exactly how experiments accomplish that.

Core Principles & Key Definitions

An experiment is defined by a specific set of features that together allow a researcher to draw causal conclusions—claims that one factor directly produces a change in another. Without every piece in place, the study may still be valuable, but it cannot make the leap from "these two things are related" to "this thing caused that outcome." Let's walk through the four essential building blocks.

A confounding variable (also called a confound) is any factor other than the IV that could explain changes in the DV. Confounds are the enemies of clean experiments. If a researcher lets students choose whether to study with or without music, maybe the students who choose silence are simply more motivated—and motivation, not the absence of music, explains their higher scores. Random assignment is the primary weapon against confounds because it distributes those hidden differences evenly across groups.

Visual Explanation — Anatomy of an Experiment

The diagram below maps out the complete flow of a true experiment, from selecting participants to drawing conclusions. Follow the arrows to see how random assignment creates equivalent groups, how the IV is applied to only the experimental group, and how the DV is measured in both groups for comparison.

Notice that the only planned difference between the two paths is the IV manipulation. Everything else—the testing environment, the instructions, the measurement tools—stays the same. When researchers hold all other conditions constant, any difference in the DV between groups can be attributed to the IV. This is the logic that turns an experiment into a tool for establishing cause and effect.

How Experimental Design Supports Causal Claims

Psychologists often say, "Correlation does not imply causation." But what exactly makes an experiment different from a correlational study? The answer comes down to three requirements for establishing causation, which only a true experiment satisfies.

Three Requirements for Causation

1. Covariation: The IV and DV must change together. When the IV is present, the DV shifts; when it is absent, the DV does not. The experiment tests this by comparing the experimental group to the control group.
2. Temporal precedence: The cause must come before the effect. In an experiment, the researcher manipulates the IV first and measures the DV afterward, locking in the correct time order.
3. Elimination of alternative explanations: Confounding variables must be ruled out. Random assignment and controlled conditions accomplish this by making the groups equivalent on every factor except the IV.

Correlational studies can show covariation (e.g., students who sleep more tend to earn higher grades), but they cannot guarantee temporal precedence or rule out confounds. Maybe students who sleep more are also less stressed, and it is the lower stress—not the sleep itself—that improves grades. An experiment would randomly assign some students to get extra sleep and others to maintain their normal schedule, then compare their grades. Because groups were randomly assigned, pre-existing differences like stress levels are distributed evenly.

The Role of Operational Definitions

For an experiment to be replicable, both the IV and DV need operational definitions—precise descriptions of how each variable is set up or measured. Saying "I manipulated stress" is vague. Saying "Participants in the experimental group completed a timed math task while being told their results would be shared publicly" is an operational definition of the IV. Saying "Stress was measured by salivary cortisol levels in nanomoles per liter" is an operational definition of the DV. Operational definitions allow other researchers to repeat the study and verify the findings.

Controls, Confounds, and Experimental Validity

Beyond random assignment, researchers use additional strategies to eliminate confounds and increase the experiment's internal validity—the confidence that the IV truly caused the observed change in the DV. The diagram below illustrates the most common types of experimental controls and the threats they address.

A placebo control is especially important in drug studies: the control group receives a sugar pill or sham treatment so that any improvement caused by mere expectation (the placebo effect) is visible in both groups. A double-blind procedure goes a step further by ensuring neither the participants nor the experimenters know who is in the treatment group, preventing subtle cues from influencing the results. These techniques raise a study's internal validity—the degree of confidence that the IV alone produced the effect.

Worked Example — Designing an Experiment

Imagine a psychology class wants to test the following hypothesis: "Students who chew gum during a memory test will recall more words than students who do not chew gum." Let's walk through the process of designing a proper experiment.

Strengths and Limitations of Experiments

Experiments are the gold standard for establishing causation, but they are not perfect. Every research method involves trade-offs, and understanding those trade-offs helps you evaluate studies critically rather than accepting every claim at face value.

Connecting to Advanced Research Concepts

The simple two-group experiment you've learned about is the foundation, but real psychological research often uses more complex designs. As you advance in psychology—especially in AP Psychology or college-level courses—you'll encounter variations that build on today's core ideas.

The key insight to carry forward is that all of these advanced methods are extensions of the same core logic: manipulate something, control everything else, and measure the outcome. Whether the design has one IV or five, the goal remains the same—isolating causal relationships by systematically ruling out alternative explanations. Mastering the basics now gives you a framework for understanding any experiment you encounter in the future.

Practice Problems

Lesson Summary