Understanding dropout of adult learners in E-learning
MetadataShow full item record
Although many studies related to e-learning have been conducted in the field of adult education and human resource and organization development, relatively little attention has been given to why adult learners actually drop out. The purpose of this study was to determine which specific set of variables can best predict the dropout of adult learners from e-learning courses in the workplace. Based on Keller’s (1987) ARCS model, a self-completion forced choice survey instrument scale was developed to obtain information about learners’ motivation to participate in e-learning in the workplace. The sample used for this study was a non-random convenience sample of employees in a South Korea company. Two hundred fifty-nine usable surveys were returned, yielding a final response rate of 12.26 percent. A logistic regression model was proposed to accomplish the purpose of the study. The primary results were: (1) The overall assessment of the proposed logistic regression model consisting of individual background variables (Number of e-learning courses completed, Age, Gender, Educational level, Marital status, Number of learning hours for the course, Mandatory/voluntary attendance, and Hours worked per week) and motivational variables (Attention, Relevance, Confidence, and Feedback) revealed that the model had a moderate association between the predictor variables and Dropout (Nagelkerke’s R-Square, .456). (2) The Gender, Number of e-learning courses completed, and Attention predictor variables had a substantive relationship to the dropout of adult learners from an e-learning course (Byx* = .40, 36, and .22, respectively). (3) The logistic regression model consisting of the Number of e-learning courses completed, Gender, Learning hours for the course per week, Hours worked per week, and Attention variables was chosen due to its efficient predictability of dropout of adult learners. This model correctly classified 48.6% of the completers and 97.9% of the dropouts, for an overall accuracy rate of 84.5% for the model.