EXPERIMENTAL DESIGNS: TRUE AND QUASI DESIGNS

Experimental designs are the research tool for making credible cause-and-effect claims—specifically, establishing that an independent variable (X) causes an outcome (Y) only after researchers control for other factors that could distort the relationship. In everyday speech people use “causes” loosely, but in research the word “cause” demands evidence strong enough to rule out confounding influences. That’s why experimental designs are treated as the most direct quantitative approach for causality, even though they can feel intimidating to students.

The lecture frames causality as a product of manipulation: researchers deliberately change X (the independent variable) and observe whether Y (the dependent variable) changes. The key is not just giving a treatment, but isolating its effect by holding other variables constant—meaning extraneous variables must be identified and controlled so they cannot threaten the inference that X leads to Y. A central warning is that correlations can be misleading: two variables may move together statistically without any causal link, as in the common claim that smoking causes lung cancer when non-smokers still develop lung cancer due to other unseen factors. Experimental logic is designed to prevent that kind of spurious relationship from masquerading as causality.

To make those causal claims, experimental studies rely on several operational requirements. First, participants must be organized into two sufficiently large, homogeneous groups with similar characteristics: a control group that receives no treatment (or a placebo-like alternative) and an experimental group that receives the intervention. Second, researchers must define inclusion and exclusion criteria clearly so the groups are comparable. Third, the treatment effect is assessed by comparing pre-test and post-test outcomes across the two groups; the experimental group’s change is interpreted relative to the control group’s lack of change.

The biggest dividing line between “true” and “quasi” experimental designs is randomization. True experimental designs use random selection and random assignment so participants have an equal chance of being placed in either group, which helps ensure equivalence and reduces bias. Quasi-experimental designs lack full randomization because researchers often work with already formed groups—such as schools, classes, or communities—where it’s impractical to rearrange participants. In those settings, researchers still measure change using pre-tests and post-tests, but the design depends on matching and careful control rather than random assignment.

The lecture also introduces standard experimental notation (O1, O2, O3, O4 and X), where odd positions represent pre-tests and even positions represent post-tests, and X denotes the intervention. It distinguishes three broad categories: pre-experimental (typically one group with pre-test and post-test and no control group), quasi-experimental (two groups without random assignment), and true experimental (two groups with random assignment). Across all of them, the goal remains the same: strengthen internal and external validity by controlling threats so researchers can defend the claim that X causes Y. The session closes by previewing upcoming discussion of threats to internal and external validity and how to mitigate them.