7.2 Designs, Sampling, and Variable Control

Match the Claim to the Design

A research design is a plan for answering a question. On the EPPP, design questions usually test whether the conclusion fits the way data were gathered. If a researcher manipulates an independent variable, controls plausible alternatives, and uses appropriate assignment, a causal inference may be stronger. If a researcher only measures variables as they naturally occur, the study may describe association or prediction without proving causation.

The most common distinction is between random selection and random assignment. Random selection concerns who is sampled from a population and therefore affects generalization. Random assignment concerns how participants are placed into conditions and therefore affects internal validity. A study can have one without the other. For example, a campus experiment may randomly assign volunteers to conditions but still have limited generalizability beyond that volunteer sample.

Experimental designs include between-groups designs, within-subjects designs, factorial designs, and randomized controlled trials. Factorial designs examine more than one independent variable and can test main effects and interactions. A within-subjects design can reduce individual-difference noise, but it may introduce order effects, fatigue, or practice effects. Counterbalancing helps manage those order problems.

Quasi-experimental designs are common in applied settings because true random assignment may be impossible or unethical. A school, clinic, hospital, or community agency may compare intact groups or use a waitlist. These designs can be valuable, but selection differences must be considered. The exam may reward an answer that says the intervention is associated with improvement, while rejecting an answer that claims the intervention caused improvement without enough control.

Correlational designs measure relationships among variables. Correlation can support prediction, but it does not establish direction or rule out a third variable. A strong positive correlation means that higher scores on one variable tend to go with higher scores on another. It does not explain why. Regression adds prediction and can control statistically for measured variables, but it does not magically create experimental control.

Single-case designs are important in clinical and applied psychology. In an ABAB reversal design, repeated baseline and intervention phases help show whether behavior changes when the intervention is introduced and withdrawn. Multiple-baseline designs can be useful when withdrawal is impossible or unethical. The key is replication across behaviors, settings, or participants, with stable baselines and clear phase changes.

When answering design items, look for the strongest justified wording. Do not inflate a design. Do not dismiss a useful design because it is not perfect. The best answer usually states what the design can show, what it cannot show, and which validity issue is most relevant.