ACT Science: Know When Data Shows Correlation, Not Causation

Correlation: Two variables move in the same direction (both increase or both decrease together). Example: Ice cream sales and drowning deaths both increase in summer. They are correlated. Causation: One variable directly causes the other to change. Example: Temperature causes ice cream sales to increase (causation), but ice cream sales do not cause drownings (even though they correlate). ACT Science tests whether you recognize that correlation alone is not proof of causation. Just because two things move together doesn't mean one caused the other.

Real example from ACT: A study shows that students who eat breakfast have higher test scores than students who skip breakfast. This is a correlation. But the study is not an experiment (breakfast was not randomly assigned). Other factors might explain the difference: wealthier families can afford breakfast and better schools, or more motivated students eat breakfast. Without an experiment that controls other variables, you cannot conclude that eating breakfast causes higher scores.

Question 1: "Did the researcher manipulate one variable and measure the effect on another?" If no, it's likely only a correlation. If yes, it might be causation (if other variables were controlled). Question 2: "Are there other plausible explanations for the observed pattern?" If yes, causation is not proven. If no (and the study was an experiment), causation is likely. Question 3: "Is this an observation or an experiment?" Observations show correlation. Experiments with controls can show causation. Use these three questions and you'll distinguish correlation from causation every time.

Example: A graph shows that countries with higher coffee consumption have lower rates of depression. Correlation: yes, both variables move together. Causation: unclear. Why? Other explanations exist: wealthier countries can afford coffee and better mental health care. The study was observational, not experimental. You cannot conclude coffee causes lower depression rates.

Scenario 1: "A researcher heats water and measures how quickly sugar dissolves. At higher temperatures, sugar dissolves faster." Likely causation (experiment with temperature manipulated, sugar-dissolving-rate measured). Scenario 2: "A survey finds that people who exercise regularly report better sleep quality." Likely correlation only (observational study; other factors like stress could explain both). Scenario 3: "Vitamin D levels and bone density show a positive correlation across populations." Correlation only (observational); vitamin D might help, or sunlight exposure might be the real cause. Scenario 4: "In a controlled lab study, increasing oxygen levels in a closed container increases bacterial growth rate." Likely causation (experiment with oxygen manipulated, growth measured, other variables controlled). Apply the three-question test to each: Scenario 1 = causation, 2 = correlation, 3 = correlation, 4 = causation.

Drill: For each, write "correlation only" or "causation likely." (1) Students who study more get higher grades. (2) A study increases study time by 1 hour for half the students and measures grade improvement. (3) Tall basketball players shoot better than short ones. Check: (1) correlation, (2) causation, (3) correlation.

ACT loves testing whether you fall into the trap of assuming correlation proves causation. Many science questions present observational data and ask what can be concluded. The correct answer often is "correlation is shown, but causation cannot be concluded." Students who confuse these concepts pick wrong answers confidently. Master the three-question test and you'll answer these questions correctly while peers argue about what "probably" causes what.

Implement the three-question method on every ACT Science section you encounter. Before accepting a causation claim, ask the three questions. By test day, distinguishing correlation from causation will be a reflex, and you'll catch trap answers that overstate what the data proves.