5. Forecasting Methods in Higher Education: An Overview

Forecasting is the first, crucial stage of planning in any organization, and in higher education (HE) in particular. Student enrollment projections are particularly important, since they affect institutions’ income, the number of faculty needed, facility requirements, budgets, etc. There are overviews of forecasting and classifications in general and for particular methods and applications. However, to the best of our knowledge, the last overview of forecasting in HE was published in 1997. Since then, two major approaches sipped from business to HE and became dominant in HE forecasting: data mining and questionnaires for marketing. The purpose of this chapter is to provide an updated overview of forecasting methods used in HE and their main areas of application. We cover a large array of forecasting methods and areas of HE application, we classify them, and point at examples from the literature, rather than providing an exhaustive annotated review, since there are too many publications in the literature on forecasting in HE. Counting the number of articles published in the Web of Science during the last 20 years, we find that, out of six main forecasting methods identified and classified, four methods are used most often in HE: regression, simulation, data mining (including its sub-methods), and questionnaires. Furthermore, four areas of application for forecasting are used most often in HE: enrollment, marketing, teaching, and performance. The two relatively new forecasting methods used in HE, during the last 20 years, are data mining and questionnaires. These two, relatively new forecasting methods, educational data mining and questionnaires (for marketing), are classified in this chapter as active forecasting methods in HE, as they provide the administrator with control over the forecast by pointing (directly or indirectly) at actions which can achieve a better-targeted forecast. While the old methods, time series, and ratio methods, are classified as passive methods with no control. Though regression and simulation forecasting methods are often active, they can sometimes be passive.