If allowed to remain in a data set, outliers can create skewness and in other ways create problems for the researcher. Accordingly, the researchers who conducted the studies that appear in Excerpts 2.27 and 2.28 deserve credit for taking extra time to look for outliers before conducting any additional data analyses.
I should point out, however, that outliers potentially can be of legitimate interest in and of themselves. Instead of quickly tossing aside any outliers, researchers would be well advised to investigate any "weird cases" within their data sets. Even if the identified outliers have come about because of poorly understood directions, erratic measuring devices, low motivation, or effort to disrupt the study, researchers in these situations might well ask the simple question,"Why did this occur?" More importantly, outliers that exist for other reasons have the potential, if considered thoughtfully, to provide insights into the genetic, psychological, or environmental factors that stand behind extremely high or low scores.
(From Chapter 2 of the 6th edition, pp. 41-42)
Copyright © 2012
Schuyler W. Huck