In: Psychology
Dr. Paddock is a counseling psychologist who is interested in decreasing adjustment issues in first-year college students. She is curious if having students create collages of their first few weeks of school and then mailing them home will help students feel they have integrated their new life with their old and, as a result, will help them feel less homesick. She samples a group of 100 incoming college freshmen at her university and measures how homesick they are during the first week of school. During Week 4 of school, she has them make the collage and send it home. During Week 7 of school, she measures their homesickness again. She notices a significant reduction in the amount of homesickness from the pretest to the posttest and concludes that her treatment is effective. Imagine in Dr. Paddock’s study that only 90 of the original participants completed the measure of homesickness during Week 7 (10 participants had left the university and were unavailable). What kind of threat to internal validity does this pose? How does this affect her conclusion that her treatment for homesickness worked?
Internal validity means how well or accurate a study is, and how well the independent variable and dependent variable are connected to each other. In this study, in the beginning, there are 100 students, but later on during the post test 10 students drop out. So, even though there has been a reduction measured in homesickness, the sample which was used in the pretest is not the same as that of the posttest. This makes the internal validity of this study weak, as the confounding variables are the missing students, and the post test sample is not the same as the pre-test sample.
So, it cannot be said that homesickness has actually been reduced, as the other 10 students are missing. The reduction in homesickness happened in the post test of the student after the treatment, but it cannot be said that it happened because of the treatment, the alternate explanation or the confounding variable here is that 10 students dropped out which may have effected the results.