Research Methods

Threats to Internal Validity

(Go back to the Research Methods page)


Directions: Identify the possible threats to internal validity associated with the research scenarios described below. For each threat you identify, say why you think it might affect the results of the study.

1. Companies that market test preparation courses for the SAT, ACT, GRE, etc. maintain the effectiveness of their services in leading to higher test scores by students. Suppose researchers recorded the scores of students who enrolled in the test preparation courses, then recorded the scores on the tests after the students had completed the course. The researchers might very well find that students’ scores went up in retesting.

2. Researchers who want to find out if participants’ attitudes change over the course of a randomized experiment frequently administer an attitude survey at the beginning of the study. Then the investigators apply appropriate experimental treatments to different groups. Then at the end of the study, the participants again complete an attitude survey. It would not be surprising if the researchers reported that attitudes changed over time.

3. Researchers are studying the levels of depression in students and how to improve their emotional state. After random assignment of participants to conditions, the investigators administer experimental treatments or a placebo. In order to avoid alerting the students to the nature of the study, the researchers initially measure depression by embedding questions from the Beck Depression Inventory (BDI) about depression in another survey. At the end of the study, they give the BDI in its original form.

4. Researchers who want to compare the behaviors of people experiencing bulimia with non-bulimics could identify appropriate people in each category and follow them for a year to see how the behaviors evolve over time. Suppose researchers compared bulimics being treated in clinics with clinic patients being treated for problems not related to eating disorders who were matched for age and gender. The researchers could gain insight into the pattern of behaviors, and any differences, over time for the two groups.

(Go back to the Research Methods page)


Updated January 8, 2004