GENERIC EARLY WARNING SIGNALS FOR CRITICAL TRANSITIONS: An assessment of the signals’ utility as a predictive management tool through an application on the undergraduate distance student’s program withdrawal problem

Abstract

Complex systems range from business entities, the human brain, to the climate. These systems have tipping points at which a small perturbation can trigger a critical transition leading to an emergence at an alternate stable state. Although there are differences in the nature of different complex systems, their behaviors exhibit universal characteristics as they near a tipping point. Among such characteristics are the common generic early warning signals that precede critical transitions. The signals include: critical slowing down in which the rate of recovery from perturbations decreases over time; an increase in the variance and skewness of the state variable; an increase in the autocorrelation of the state variable; flickering between different states; and an increase in spatial correlations. The presence of such signals has significant management implications, as the identification of signals prior to the tipping points could allow management to identify intervention points. A review of literature did not identify any applications for the signals in managing undergraduate program student withdrawal at distance universities, hence the research gap. This research assessed the signals through an intensive case study of undergraduate program student withdrawal at a Canadian Distance University by comparing the incidences of the signals among inactive students to the incidences of the signals among graduates. Findings showed support for the signal on the rise in flickering, represented in the increase in the student’s non-pass rates prior to withdrawing from a program; moderate support for the signal of critical slowing down, reflected in the longer time a student spends in a course; and moderate support for the signals on increase in autocorrelation, skewness and variance in the grade variable. The research also extended knowledge by investigating whether the emergence of a program withdrawal status is self-similar at the program level and the course level. Findings moderately supported self-similarity as a potential signal. In conclusion, the research into the signals and self-similarity suggests that the signals could be potentially utilized as a predictive management tool. These findings represent the beginnings of future research into the creation of program withdrawal Complexity based models and the possible identification of intervention points.