The Traditional Approach: The Kirkpatrick Model - Training and Development

The Kirkpatrick model (1998) of training evaluation has dominated training evaluation discussion since it was first published forty years ago (Kirkpatrick 1959a,1959b, 1960a, 1960b). It suggests that training should be evaluated at four"levels": 1—participant reactions, 2—learning, 3—on-the-job behaviors, and 4—results from behavior change. The American Society for Training and Development (ASTD) has embraced this framework in its learning outcomes report (Bassi and Ahlstrand 2000). As part of its benchmarking service, participants provided data on standardized measures of Level 1 and Level 3 outcomes.

Despite its popularity with practitioners, in recent years the four-level taxonomy has come under increasingly intense criticism (Alliger, Tannenbaum, Bennett,Traver, and Shotland 1997; Alliger and Janak, 1989; Holton 1996; Swanson andHolton 1999). The chief criticisms are:

  • Not supported by research. Research has consistently shown that the levels within the taxonomy are not related, or only correlated at a low level.
  • Emphasis on reaction measures. Research has shown that reaction measure shave a nearly zero correlation with learning or performance outcome measures.
  • Failure to update the model. The model has remained the same for the last forty years with little effort to update or revise it.
  • Not used. As will be discussed in the next section, the model is not widely used. Despite forty years of urging people to use it, most do not find it a useful approach.
  • Can lead to incorrect decisions. The model leaves out so many important variables that four-level data alone is insufficient to make correct and informed decisions about training program effectiveness.

All rights reserved © 2020 Wisdom IT Services India Pvt. Ltd Protection Status

Training and Development Topics