Reporting on student retention – From retrospective to predictive

South Africa has a low-participation, high-attrition higher education system, according to the National Development Plan, a government blueprint for economic growth. Student retention – and, in particular, corresponding drop-out and graduation rates – have become critical issues, as in most countries around the world.

My article on “The emergence of reporting on student retention in South Africa: From retrospective views to predicting the future” (pp 111-115), published in Student Affairs and Services in Higher Education: Global foundations, issues, and best practices, looks at ways data can be used in South Africa to predict which students are more likely to drop out.

Many students believe that rising tuition fees is a major reason for students dropping out of university, especially those who have limited financial support. South Africa faced numerous protests in 2016 related to proposed increases in fees that resulted in a renewed focus on student retention as part of the continuing debate about the cost of attending university.

Some background

The National Department of Education reported in 2001 that the undergraduate graduation rate was 15% – one of the lowest in the world. The department also reported a 30% student drop-out rate in the first year of study at higher education institutions in 2005, a figure that it estimated was costing the Treasury about US$350 million (ZAR4.5 billion in June 2017) in grants and subsidies to universities without a commensurate return on investment.

The high drop-out figures obliged the department to encourage cohort studies. This could be done once the National Higher Education Management Information System or HEMIS, which collects university data, began operating in 2000, making it possible to trace student performance for an initial five years.

Affordable higher education, and how to accommodate more students and improve their success rates as well as institutional efficiency, are the subject of increasing focus and concern in South African higher education.

A Consortium of Higher Education Associations in the United States received a Ford Foundation grant to visit South Africa in 2002 to identify and establish the principal needs for higher education and student affairs.

The most pressing need identified was research and data collection; student retention was just beginning to be recognised as an issue of concern. Universities needed to conduct more student-related research and devise unified ways of reporting retrospectively on retention.

The HERD-SA initiative

A grant proposal, Higher Education Retention Data for South Africa – HERD-SA – was developed after consortium visits and was approved by the Kellogg Foundation in 2005.

The HERD-SA project’s purpose was to standardise the measurement of retention rates in terms of South African definitions, to develop appropriate methodology for reviewing and presenting retention data, and to compile a profile of student characteristics from the data collected to establish what, if any, relationships the data may have to student retention.

A group of 10 universities took part in the project, which was completed in December 2007. That year a workshop was held between the vice-chancellors’ group Universities South Africa (then called Higher Education South Africa), and HERD-SA to discuss ways forward.

All universities were invited to the workshop, whose purpose was to: a) review the HERD-SA project and the results it compiled; b) discuss the value of collaborative research in the area of student retention, and the role of HESA; c) consider best practice in reporting student retention data; and d) discuss the continuation of the HERD-SA project and the critical issues that must be addressed.

The participants agreed on the need to take HERD-SA’s work forward. This was justified by the needs to: a) continue the work of gaining acceptance of standardised concepts and definitions relating to student retention; b) continue developing and disseminating effective ways of analysing and presenting data on student progress, including factors affecting retention; and c) explore ways of facilitating the use of research on student advancement to help improve learning and retention.

A flurry of research into retention

After the HERD-SA project, various reports were published relating to retention, throughput and graduation trends. Further analysis of higher education performance has also shown low student throughput rates, and a report by the Council on Higher Education evaluated the feasibility of restructuring three- and four-year undergraduate degree and diploma courses.

The new Department of Higher Education and Training analysed studies of first-time entering undergraduate cohorts at institutions from 2000 to 2014. It found general improvement in throughput rates of the 2006 and 2009 cohorts compared with 2000. The department now requires universities to report on progress in student retention in their annual performance plans.

Locally developed software – the Higher Education Data Analyser, or PowerHEDA – has been used by 17 of South Africa’s 26 public universities in supplying retrospective and forward-looking management information and dashboard reports on student retention.

It is clear that monitoring and supporting first-year students remains critical. Predicting student retention is, therefore, an increasing concern for administrators and government owing, in part, to the costs associated with students not completing their studies.

The application of predictive analytics in higher education has started in the last 10 years. Predictive analytics is also referred to as academic analytics and involves analyses of data using statistical modelling techniques. Learning analytics, on the other hand, refers to statistics at an individual student level.

Using analytics to improve retention

Great attention is paid to understanding the enrolment behaviour of students and many projects focusing on students’ experiences in the first year of study have been implemented by some universities in an attempt to improve student retention.

Given high drop-out rates in the first year of study, the key benefit of predictive analytics is that it can assist universities to introduce targeted intervention strategies in the first year of study in order to reduce the number of students leaving before second year.

Retrospective views on student retention are a critical first step in identifying trends. But being able to predict the future success of students will enable universities to introduce intervention strategies that can be expected to improve success wherever students struggle with courses.

Higher education researchers continue to play an important role in providing sophisticated analyses. The changes in business intelligence, big data and learning analytics affect the work of institutional researchers and new skill sets will be required to ensure sufficient support is provided to policy-makers.

A Lourens is extraordinary professor at North-West University in South Africa. E-mail: Amanda@idsc.co.za.