Designing studies to avoid confounds

The ability to design studies that are free from confounded variables is an acquired skill that separates the true psychological scientist from a layperson. The latter might be quite capable of generating interesting questions that could be addressed by psychological research; turning these questions into a study that cleanly tests the hypotheses, however, can be quite a challenge. It is this topic that the present chapter addresses. Consider the fundamental goal of psychological research: to discover the causes and consequences of behavior. The only way to make such discoveries is to be able to examine data from a study that is free from alternative explanations. Such alternative explanations most often arise when an experiment contains confounded variables. Confounded variables involve the “simultaneous variation of a second variable with an independent variable of interest so that any effect on the dependent variable cannot be attributed with certainty to the independent variable” (Elmes, Kantowitz, & Roediger, 2003, p. 436). A well-designed study is one in which the researcher has carefully considered potential alternative explanations and designed the study so that these alternative explanations are no longer viable. Psychological research can be categorized into two broad classes: experimental studies and correlational studies. In the former case, the researcher manipulates the variable of interest (the independent variable) and observes its effects on the dependent variable. The latter involves examining variation that occurs naturally (e.g., the variation between emotional awareness and happiness) and attempts to draw conclusions regarding this relationship.