KIRKPATRICK'S LEVELS OF TRAINING CRITERIA: THIRTY YEARS LATER

TLDR

Kirkpatrick’s 1959–1960 model of training evaluation, which proposes four ascending levels—reactions, learning, behavior, and results—has remained popular but rests on three problematic assumptions: that the levels are ordered, causally linked, and positively intercorrelated. This article investigates the validity of those assumptions, the prevalence of each level in published studies, the correlations among levels, and the implications for researchers and training managers.

Abstract

Kirkpatrick's model (1959a, 1959b, 1960a, 1960b) of training evaluation criteria has had widespread and enduring popularity. This model proposed four “levels” of training evaluation criteria: reactions, learning, behavior, and results. Three problematic assumptions of the model may be identified: (1) The levels are arranged in ascending order of information provided. (2) The levels are causally linked. (3) The levels are positively intercorrelated. This article examines the validity of these assumptions, the frequency of each level in published evaluation studies, correlations from the literature in regard to Assumptions 2 and 3, and implications for the researcher and training manager.