Extraneous Variance in Research
Good Study Design Controls for Nuisance Variables
One of the ways researchers can control for extraneous variance is by ruling out alternative explanations through good research design.
When a researcher establishes a relationship between two variables in a study, he must take care before asserting that the effect is a causal one. It’s possible that some other extraneous or nuisance variable is causing the outcome. Consider the following three methods researchers can use to control for nuisance variables in a study.
History Threat
Researchers must show that some other event isn’t responsible for the study outcome. Research scholars call this the history threat. For example, while running a series of anti-smoking public service announcements in a city, a movie star shares that he’s just received a diagnosis of lung cancer from 40 years of smoking. Did the reduction in cigarette purchases come from the public service announcement or the reaction to the celebrity’s cancer? To control for this, the researchers shouldn’t do a single group study. The researchers should choose a similar city as a control group to rule out this threat to internal validity.
Experimental Mortality
Experimental mortality can bias the results of a study. When subjects withdraw from the study before completion, the results of the study may show artificially inflated results or may minimize an effect that would have been present without the mortality. For example, in an adult learning study, individuals who feel the control educational program isn’t effective might get discouraged and drop out. The researchers would then be guilty of artificially inflating the posttest average of the experimental group. If mortality is a problem, researchers can compare the dropout group to the non-dropout group on pretest measures. If there are no major differences, it may be reasonable to assume that the mortality occurred across the entire sample and isn’t biasing results greatly. If the differences are large, the researchers should address this in the discussion section of their study.
Subject Awareness
Alterations can occur when researchers test subjects and they become anxious or wise to the nature of the test in ways that affect their responses. This is more common in pre-post test designs. Take the example of a study researchers are conducting to examine the efficacy of peer counseling on the attitudes of problem drinking in college students. The researchers administer a pretest to determine the attitudes of the participants regarding drinking. Taking this pretest can make the students aware of the problem in a way they wouldn’t have been without the test. Researchers may alleviate this problem in part by using broad language in the test that conceals the variable under investigation.
Learn more about ways to ensure reliability and validity in research.
Sources:
Reinard, J. (1998). Introduction to Communication Research, 2nd Ed. McGraw-Hill: Boston.
Verdugo, E.D. (1998). Practical Problems in Research Methods. Pyrczak Publishing: Los Angeles.
