FAYETTEVILLE, Ark. - Companies in the United States spend billions of dollars each year to train employees on computer skills. But success varies because so little is understood about the characteristics of any specific method and the reasons or circumstances under which a method may be effective. A University of Arkansas researcher has characterized the most successful method and found an additional component that can significantly increase information retention.

"A training method that works well for spreadsheets may fail entirely as a method to train database skills. Or it may work well with one group, but not with another," explained Davis. "Ineffective training reduces productivity and employee morale, but making training more effective requires an understanding of the underlying processes. This is the first study to analyze and explain the observational learning processes that link modeling-based training to training outcomes."

Davis, professor and department chair in information systems in the Walton College of Business, conducted his research with Mun Yi, assistant professor of information systems at the University of South Carolina. Their results appear in the summer issue of Information Systems Research.

The researchers used behavior modeling, a common method in computer training that uses videotape or a facilitator to introduce basic concepts and demonstrate a specific skill to a training group.. Trainees are then given the opportunity to try that skill on their own computers. The trainees acquire the skill through observational learning - they watch it being demonstrated and then try to repeat what they have seen.

The study involved training a group of freshman business majors to use Microsoft Excel spreadsheet software. The 95 participants had little or no previous experience with any kind of spreadsheet software. In addition to extra credit in a course, each participant received $10. Confidential feedback was provided to encourage participants to master the skills.

"We identified the four dimensions of observational learning - attention, retention, production and motivation," Davis explained. "We were testing a new model designed to show the influence of modeling-based training on outcomes by looking at its effect on underlying observational learning processes."

Study participants were divided into two groups. Both groups received standard modeling training, but one group also received a retention intervention. The training was conducted on two consecutive days with two three-hour training sessions each day. In each session two workshops were conducted simultaneously in separate labs, each implementing a different training condition with a different facilitator.

After completing a pre-training questionnaire, trainees in the first group watched two video segments and were then given 15 minutes to practice. They then watched the other three video segments and had another 15 minutes for practice. Participants were then given a multiple-choice test of knowledge and a hands-on task performance test.

Participants in the second group followed exactly the same schedule with the addition of a retention intervention. Following the pre-training test, these trainees were given blank papers labeled with section headings for each activity. At the end of each video segment the tape was paused and the trainees were instructed to summarize the presentation by writing down key points under each heading. Following the second practice session, the participants were given 5 minutes and instructed to review their summaries and mentally picture themselves completing the computer operations. They were asked to repeat this mental rehearsal as many times as possible until they were told to stop.

After this exercise, the participants in the second group were also given the multiple-choice knowledge test and hands-on task performance test. As a follow-up, both groups were administered these tests again 10 days after the initial study.

As expected, the retention intervention significantly influenced the training outcomes of computer self-confidence, knowledge and task performance. The retention intervention significantly improved these outcomes by influencing the retention component of the observational learning processes, but it had no effect on attention, production or motivation components.

"This new model provides a more complete account of the mechanisms by which modeling-based methods affect training outcomes," Davis said. "It not only enables future research to systematically evaluate the effectiveness of a wide range of modeling-base training, but it can be used by trainers to refine ongoing training programs."