ERIC Number: ED233282
Record Type: RIE
Publication Date: 1983-Apr
Reference Count: 0
Curtis, Jonathan; And Others
Secondary school students who drop out of school are put at great social and economic disadvantage. If potential dropouts can be identified early, prevention may be possible. To construct a prediction model which, through readily available school information, will aid in the identification of students likely to drop out, schools in the Austin, Texas, school district provided data on students sampled according to birthdate (N=5,039). The longitudinal study encompassed four school years (1977-1981), and students were classified into four groups: non-leavers, transfers, dropouts, and other/unknown. The variables studied were gradepoint average (GPA), grade placement, sex, ethnicity, and number of serious discipline problems. Statistical analyses showed that students who had low GPA's, were behind in grade for their age, had been involved in serious discipline incidents, were female, and were non-black had a higher than average probability of dropping out. Most of these variables had face validity for predicting dropping out. Two findings were somewhat misleading: in fact, girls as a group had a lower drop out rate than boys; however, an individual girl with certain characteristics might be more likely than a similar boy to drop out. Although scores of Blacks on variables were similar to scores of Hispanics, Blacks were less likely to drop out, apparently due to influences outside of the variables scored. Altogether, about 78% of dropouts and non-leavers were correctly classified. (WAS)
Publication Type: Reports - Research; Speeches/Meeting Papers
Education Level: N/A
Authoring Institution: Austin Independent School District, TX. Office of Research and Evaluation.
Identifiers: Texas (Austin)
Note: Paper presented at the Annual Meeting of the American Educational Research Association (67th, Montreal, Canada, April 11-14, 1983). Marginal legibility for figures 10 and 11.