Abstract:

Universities today have a different landscape than they did a generation ago. Globalization towards a growing technology and knowledge-centric economy has heightened the demand for highly skilled and knowledgeable workers. To take advantage of this economy, education reforms are required for an educated and skilled workforce. These reforms would include widened access of education to non-traditional communities, such as the adult learners. These learners are often working adults enrolled in part-time programs, and some are returning to education after a break of few years (Melkun, 2012). The influx of adults to higher education presents key challenges to the institutions where they matriculate.

There is a need for an early alert system to identify adult learners who struggle with the transition from work to university. Beck and Davidson (2001) defined an early alert system as “mechanism for identifying students most likely to fail academically or to encounter serious problems assimilating into the college environment” (p. 709). Predictive modelling is used to identify learners at-risk so as to provide timely intervention. Predictive models have no impact on academic success without effective intervention strategies aimed at supporting the at-risk learner. This paper describes a pilot project of an early-alert system in a Singapore university for adult learners in the Business School.

The early alert system is based on a predictive model to identify learners who may be at-risk. The sample population that is used to build the predictive model are adult learners who completed 25 credit-units and are completing the next 25 credit-units of a 130 credit-units part-time bachelor degree program in the business school. The 34 input variables used to construct the predictive model are from 623 adult learners who enrolled from July 2015, January 2016, July 2016, January 2017, July 2017 and January 2018 semesters. The input variables can be broadly categorised into Demographics & Work Background variables (e.g. age and gender), Prior Education variables (e.g. e.g. polytechnic GPA, ‘O’ level Math and English grades) and University variables (e.g. Discipline enrolled and Years into Study).

The model is then validated based on 533 adult learners enrolled in the July 2018 semester and finally deployed on 558 adult learners enrolled in the January 2019 semester. The results are displayed in a dashboard that allows users to identify the adult learners that are predicted to be at-risk for intervention.

Various levels of intervention strategies will be used. Some adult learners will be receiving an email from their Head of Programme while others will be monitored. Results from the various strategies will be explained and the implications of the results will be discussed.

References:
[1] Beck, H. & Davidson, W. (2001) Establishing an Early Warning System: Predicting Low Grades in College Students from Survey of Academic Orientations Scores. Research in Higher Education. 42(6). 709-723.
[2] Melkun, C. H. (2012). Nontraditional Students Online: Composition, Collaboration, and Community. The Journal of Continuing Higher Education, 60(1), 33-39.

Keywords:

Adult learner, academically at-risk, predictive modelling, intervention.