College Dropout Identification
William C. Blanchfield
Abstract
William C. Blanchfield
Abstract
The process of selecting and evaluating high school students for the scarce openings in our colleges goes on continuously. The process may be informal interviews and personal judgment being the major ingredients; or it may be a highly structured mathematical process, most often using regression-correlation techniques. Neither method has been distinguished by its success. The shortcomings of the subjective approach are obvious: too often the interview is hurried and the important indicators of possible success in college are not uncovered. The structured research has most often been an attempt to predict future quality point averages using prior data. The reasons for the lack of previous success in the structured research are twofold: 1.) Regression-correlation is only suitable when the prior data is adequately separated; and 2.) The premise upon which the technique is based is not suitable: that is, that it is possible to rank students in a significant manner to predict future success or failure. Rather than ranking students on a scale, the emphasis should be on identifying potentially successful or dropout students. The technique ideally suited for this is Multiple Discriminant Analysis (MDA).
OpenAlex reports 57 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
The process of selecting and evaluating high school students for the scarce openings in our colleges goes on continuously. The process may be informal interviews and personal judgment being the major ingredients; or it may be a highly structured mathematical process, most often using regression-correlation techniques. Neither method has been distinguished by its success. The shortcomings of the subjective approach are obvious: too often the interview is hurried and the important indicators of possible success in college are not uncovered. The structured research has most often been an attempt to predict future quality point averages using prior data. The reasons for the lack of previous success in the structured research are twofold: 1.) Regression-correlation is only suitable when the prior data is adequately separated; and 2.) The premise upon which the technique is based is not suitable: that is, that it is possible to rank students in a significant manner to predict future success or failure. Rather than ranking students on a scale, the emphasis should be on identifying potentially successful or dropout students. The technique ideally suited for this is Multiple Discriminant Analysis (MDA).
Key concepts: Dropout (neural networks), Premise, Ranking (information retrieval), Identification (biology), Process (computing), Quality (philosophy), Rank (graph theory), Psychology