Attrition Among College Students

Abstract

The ability to complete four years of college within four years after matriculation was examined in a longitudinal study of 36,405 college students entering 246 colleges and universities as freshmen in 1961. It was found that students who do not complete four years of college come from lower socioeconomic backgrounds, have lower grades in high school, and have a lower level of initial educational aspiration than do students who complete four years of college. An analysis of the effects of 36 college characteristics on student persistence in college was performed. After controlling statistically for differential student input to the various college environments, 21 significant college effects were observed. It was suggested that students are more likely to complete four years if they attend colleges where student peer relationships are characterized by friendliness, cooperativeness, and independence, where the students frequently participate in college activities, where there is a high level of personal involvement with and concern for the individual student, and where the administrative policies concerning student aggression are relatively permissive. Attrition Among College Students In the fall of 1961 a study of all entering freshmen students at a national sample of 248 colleges and universities was conducted at the National Merit Scholarship Corporation (Astin, 1965). The 127,212 students who participated in the study provided information on their socioeconomic backgrounds, high school activities and achievements, and educational and vocational aspirations. 1 In the summer of 1965 the Office of Research of the American Council on Education 2 followed up randomly selected samples of students from each of the 246 colleges and universities included in the 1961 survey. 3 Questionnaires were mailed to 60,078 of the original 127,212 students -or approximately 250 students per institution. 4 In this follow-up, data were obtained concerning the student's current educational and vocational plans and his experiences during the college years. In addition, a number of measures of the college environment were available from a follow-up questionnaire completed by 30,570 of the 60,078 students in the summer of 1962 (Astin, 1967). The data obtained from this four-year longitudinal study have made it possible for us to examine how a variety of student characteristics are related to dropping cut of college, and to estimate the impact of various features of the college environment on student persistence in college. This report presents the results of an analysis of the personal and environmental factors associated with dropping out of college. For the purposes of our study, the student's criterion status or was defined as follows: a nondropout was any student who reported in 1965 that he had completed four or more academic years of college work, whether or not he had received a bachelor's level or equivalent degree. Thus, students who'had changed colleges or had dropped out but re-enrolled and still managed to complete at least four years of college were defined as nondropouts. All other students, that is, students who left the institution they entered in the fall of 1961 and had not completed four years of college by the summer of 1965 were considered to be dropouts, even if they were enrolled in college at the time of the follow-up survey. This dichotomously scored criterion measure served as the dependent variable in the study. (Since some of the students classified as dropouts in this study will eventually graduate, and some of those classified as nondropouts will never receive a terminal degree, this study may be regarded as an analysis of the factors associated with completing four academic years of college work within the four years following matriculation.) Method The first step in the analysis of data obtained from this study was to relate the student's dichotomous criterion score (dropout or nondropout) to the precollege input data that he originally provided in the fall of 1961. An criterion score based on the student's input data was thus computed, and a residual criterion performance score obtained by removing the effect of this expected score (that is, subtracting it) from his actual criterion score. The final step in the analysis was to relate the residual criterion score (now statistically independent of input) to the various environmental characterisitcs of the institutions attended by the students. The statistical model used for computing the expected criterion score was that of linear multiple step-wise regression utilizing the com-