Based on research on nonresponse bias summarized by Rosenthal and Rosnow in 1975, the researchers hypothesized that responses from the 43% of college students who voluntarily returned surveys of students' attitudes and behavior would be more congruent with college goals and expected lifestyle than responses one year later from a randomly selected group of 7% of the students. When percentages of students from the volunteer and random groups were compared on each of 82 items of a survey of students' attitudes and behaviors, most items showed differences of 3% or less. Of the 23 items which varied between 5 and 9%, only six were statistically significant. When compared with the college's stated goals and lifestyle expectations, only three of these items were congruent, which did not support the original hypothesis. With such small and random differences, the authors concluded that either method was equally valid in sampling students' self-reported attitudes and behavior.
Stevick, Richard A.; Barnes, James H.; and Orke, Sandra G., "Comparing survey results of volunteer and randomly-selected subjects" (1990). Educator Scholarship. 30.