Randomized Experiments
What is a randomized experiment?
Imagine you want to know whether a teaching method you’ve learned about in GS750 really helps students in your class. You notice that students who took the class with these new implementations seem to get higher scores. But wait — maybe those students were already more motivated, or had more free time. How can you tell if it was really the teaching method that caused the improvement?
To do this, researchers try to compare groups that are as similar as possible — except for the thing they’re testing (like the new teaching method).
One simple (although not always possible) way to do that is a randomized experiment, where individuals are randomly selected into treatment and control groups. Because the assignment is random, the groups are fair, and any difference in outcomes can be attributed to the treatment itself.
For example, you might randomly assign some students to get the new method (the “treatment”) and others to continue with the old method (the “control”).
Observational Studies
While randomized experiments are the gold standard for identifying cause-and-effect relationships, they are not always possible. Sometimes it’s unethical, impractical, or too expensive to assign treatments randomly. In these cases, researchers rely on observational data, where they simply collect information about what naturally happens and look for patterns.
An observational study is a type of research where we observe and collect data on people, groups, or things without actively assigning treatments or interventions. Because the groups may differ in ways other than the factor we’re interested in, it can be harder to tell what actually causes an outcome compared to a randomized experiment.
Test yourself
Drag each scenario to the appropriate box: “Randomized Experiment” or “Observational Study”.
Scenarios:
A pharmaceutical company randomly assigns patients to receive either a new drug or a placebo to test its effect on blood pressure.
A researcher looks people’s natural coffee-drinking habits and tracks their sleep quality over a month.
A public health study tracks the diets of different communities and measures heart disease rates without assigning specific diets.
A school randomly assigns some students to a new math tutoring program and compares their test scores to students in the regular program.