A comparison of applications-based and construct-based training methods for dss generator software
Lorne Olfman · Indiana University eBooks · 1987
The growth of computer use by those who are not professional programmers and analysts makes it imperative that effective software training methods be applied to support these users (referred to as end-users). This research study postulates that more personally relevant software training will lead to better understanding, perceived usefulness, and actual use of the target software. However, individual differences of trainees, such as learning style, were expected to moderate the outcomes. Two methods of training--construct-based training (a typical current approach) and applications-based training (an approach that emphasizes personal relevance)--were compared in a field experiment. Seventy employees from two organizations volunteered for a one-day training session in Lotus 1-2-3, a popular end-user software package. Six training sessions were held using each method. Trainees completed pre-training computer experience and attitude questionnaires and the Kolb Learning Style Inventory. Training outcomes were measured at the end of the session, and trainees were interviewed about their actual use of the software eight weeks after training. Overall, differences between training methods were not statistically significant. It is likely that this was due to the large amount of hands-on experience and orientation to problem-solving in both training methods. The direction of differences and qualitative findings indicated that applications-based training was a more effective method, especially for trainees who had no previous experience with Lotus 1-2-3. Applications-based trainees also used Lotus 1-2-3 more after training than construct-based trainees. This difference was related to a number of factors including better understanding of how to apply the software to a specific problem (as measured during training) and better understanding of how to apply the software on the job. Individual difference variables provided mixed results in terms of moderating outcomes.