Traditional, web-based, and hybrid instruction: A comparison of training methods.

Abstract

Previous efforts to summarize differences between traditional and web‐based training program learning outcomes have been limited by ignoring pre‐training differences due to non‐random assignment to conditions. In the present meta‐analysis, pre‐training and post‐training differences between media (traditional, web‐facilitated, hybrid, and fully online instruction) were summarized across a variety of moderators. Self‐selection into condition (e.g. college students choosing to take a traditional or online course according to their preferences) lead to substantial outcome differences before training begins; trainees choosing online courses know more (d = .19) about the material on which they are going to be trained than those choosing traditional courses. This makes interpretation of post‐training outcome differences alone unwise. To address this, three approaches were taken: 1) pre‐training and post‐training between‐subject d’s (traditional vs. web‐based instructions) were compared, 2) traditional and web‐based within‐subject d’s (post‐training vs. pre‐training scores) were compared, and 3) between‐subject post‐training d’s adjusted by pre‐training differences for those studies in which self‐selection was present were examined. These analyses led to several conclusions, including 1) within‐subject outcome gains are large in comparison to between‐subject differences in media, 2) differences do exist across criterion type (knowledge, observable skill, problem solving skill, attitudes, and perception of gain), 3) the effectiveness of web‐based courses in comparison to traditional courses appears to be changing over time, perhaps as the design of online courses becomes more sophisticated, 4) the use of learner control in online courses appears to produce worse outcomes for learners, 5) online courses produce better outcomes than traditional courses when those courses take place over a longer time period, and 6) learners self‐selecting into online courses seem to be older and have greater computer/Internet experience than those self‐selecting into traditional courses. Areas of e‐learning research needing more primary empirical research are identified, and suggestions for training design practitioners are discussed.