SLS - Sustainable Learning of Statistics

Background

Higher education systems all over the world increasingly emphasise the importance of high-quality learning that can be used in real life and work contexts. When discussing the issue of quality of learning, two intertwined challenges arise. The first is that people frequently forget what they have learned and their learning appears to evaporate. The second challenge is that people learn but cannot apply their knowledge except in the context where they learned it. They have trouble using it in new situations. Successfully overcoming these two challenges is the mark of high-quality learning—that is, learning that enables students to deal with novel situations in powerful ways now and in the future. We call such learning ‘sustainable learning’, which refers to the general meaning of sustainability: the propensity of something to continue and grow over time.

Our concern in this project was the learning of statistics: the science of learning from data and the science that is measuring, controlling and communicating uncertainty. In today’s data-rich world, statistics has become one of the most central topics of study. It is found in most disciplines, and almost all post-graduates programmes provide courses handling quantitative data. 

We studied how post-graduates’ students learning about statistical aspects of the world around them relates to their sustainable learning of statistics.

Purpose and aims

The purpose of this project was to investigate how sustainable learning in the domain of statistics can be enhanced, particularly statistics taught in postgraduate research methods courses. We used the variation theory (VT) of learning (Marton, 2015) to pursue the following aims:

1. To provide a theoretical and practical case for how sustainable learning in the domain of Statistics can be enhanced; and
2. To test the conjectures of the VT of learning and contribute to its theoretical development.

We investigated the assumption that teaching statistics - when the necessary conditions for powerful learning are met - can increase the potential for sustainable learning. Such necessary conditions are those that allow learners to experience necessary patterns of variation and invariance in critical aspects of the object of learning, which is the competence of being able to use statistical skills in new contexts now and in the future.

Significance

The significance of this project can be summarised in five points:

1. VT has so far been mainly used to study the relationship between learning and teaching from the point of view of teaching; in this project, the same relationship was studied but from the point of view of learning.
2. This project dealt with a complex object of learning aimed at empowering students to use statistical knowledge and reasoning in new contexts and settings, now and in the future.  
3. Our aim was not just to improve the learning of core statistical concepts but to see how students use such learning in the context of statistical problem solving and novel situations. Therefore, this project is significant for statistics education research.
4. Most of the previous studies inspired by VT were in the contexts of primary or secondary school level. Investigating VT conjectures in a new discipline and new level (higher education) should widen the sphere of VT.
5. Facing and overcoming the challenge of bridging the two different contexts (VT and variation in statistics) contributes to our understanding of students’ learning both within the theory and within statistics education.

Results and outcomes

The project showed the importance of not only learning individual statistical concepts, but also differentiating and coordinating interconnected statistical ideas in relation to quantitative research problems. 

Variability, as a statistical concept, emerged as foundational to reasoning under uncertainty. At the same time, foregrounding variability alone was not sufficient. Structured patterns of variation and invariance supported the discernment of critical aspects, while opportunities for fusion supported their integration and coordination.

The longitudinal assessments showed that several coordinated elements of students’ statistical reasoning remained evident five months after the course. Although individual trajectories varied, the overall pattern did not indicate systematic fading of learning.

The project produced a theoretically structured and empirically grounded design for a complete master’s-level quantitative research methods course, developed and refined through two instructional iterations. It also produced an assessment instrument consisting of three parallel and comparable test versions designed to assess statistical reasoning and sustainable learning over time. The tests were administered before teaching, immediately after teaching, and five months later, and required students to reason about novel statistical tasks.

More information