Factors Influencing Knowledge and Skill Decay after Training: A Meta-Analysis

Training is an essential endeavor in many organizations. Based on a 2009 report by the American Society for Training and Development, an estimated $134.1 billion was spent on formal training among U.S. organizations (Paradise and Patel, 2009). However, it is commonly thought that only 10 percent of these dollars invested in training results in “enduring behavioral change” (Wexley and Latham, 2002, p. 261). These low returns on investment are costly to organizations. From an organization’s perspective, maximizing the amount of posttraining knowledge and skills retained is directly linked to achieving a high return on investment. Consequently, the present chapter presents a metaanalysis of existing organizationally relevant training research on knowledge and skill decay. Results based on 111 independent effects retrieved from 35 reports suggested an overall moderate decay effect (d = –0.38). The amount of decay was minimal for periods of nonuse less than 1 day (d = –0.08). For longer periods of nonuse, decay was moderate to large (ds ranged from = –0.24 to –0.84), but there was no straightforward monotonic relationship between the amount of decay and length of nonuse. That is, beyond 1 day, a greater length of nonuse did not always translate into greater decay. Decay was related to several methodological factors. Specifically, decay was greater when the operationalization of acquisition was criterion-versus duration-based, forcognitive versus skill-based criteria, and when the instructional environment was less structured. Furthermore, posttraining decay-prevention interventions showed strong decay-mitigating effects with large gains in performance after formal training. Decay was also related to the combination of cognitive and physical (i.e., psychomotor) task demands as well as task complexity. Tasks with moderate cognitive demands and minimal physical demands were associated with the greatest decay, whereas tasks with minimal cognitive demands but strong physical demands were associated with the least decay. Complex tasks were associated with modest levels of decay that were unlikely to be moderated by additional factors, but for simpler tasks decay effects were less robust and dependent upon other factors. Regression analysis indicated that decay was primarily related to the length of nonuse, amount of cognitive task demands, and the closed-/open-looped distinction such that longer periods of nonuse, greater cognitive demands, and closed-looped tasks were associated with greater decay. These results are discussed with respect to Arthur, Bennett, Stanush, and McNelly’s (1998) previous meta-analysis of decay effects. Practical and theoretical implications are also discussed.