Experiment
An experiment is a planned investigation that changes one factor on purpose and measures the effect to test a hypothesis.
Definition
An experiment is a test you set up to find out whether your hypothesis is right: you change one thing on purpose and watch what happens to something else. More precisely, the investigator manipulates an independent variable, holds other variables constant, and measures a dependent variable, repeating trials so the results can be attributed to the change rather than to chance. A well-designed experiment is a controlled intervention aimed at a causal claim, with a control condition, a procedure others can replicate, an estimate of uncertainty, and an analysis plan chosen before the data come in. Watching the moon each night is an observational study; an experiment requires that you manipulate something.
Example
To see whether ramp height changes how far a cart rolls, a student sets the ramp at three heights, releases the cart from the top each time, and measures the distance it travels on the floor. A group tests how water temperature affects the time for $5\ \text{g}$ of salt to dissolve in $100\ \text{mL}$, stirring at the same rate for three trials each at $10^\circ\text{C}$, $30^\circ\text{C}$, and $50^\circ\text{C}$, and gets mean times of $48\ \text{s}$, $26\ \text{s}$, and $14\ \text{s}$. Another group compares bare wood, felt, and sandpaper at a fixed $20^\circ$ angle, five releases per surface from the same $0.50\ \text{m}$ mark, finding mean speeds of $2.4$, $2.1$, and $1.7\ \text{m/s}$ with ranges under $0.1\ \text{m/s}$.
Key Insight
Change only one thing at a time; if you change the ramp height and the cart together, you cannot tell which caused the difference. An experiment answers a narrow question well and cannot answer a question it was not designed for. Reading beyond what the design supports is the most common error in lab reports.