Tom Stafford

University of Sheffield, UK

Factors influencing optimal skill learning: data from a simple online game

We have used data from a simple online game (n>1.2 million players), tracking players from their first attempts up to their first hundred games to test and extend cognitive theories of learning. Our results, on the whole, validate the role in skill learning of known factors such as practice, spacing and exploration of the strategy space (Stafford & Dewar, 2014), and allow extensions in a real-world domain. This large, observational, dataset presents particular challenges and opportunities for theorist of learning. I will discuss the analytic strategies we used to try to identify factors most likely to be causal. I will also argue that very large data behooves us to move from thinking about simple effects to what I call ‘parametric comparison’ (Stafford & Haasnoot, under review) – analyses that reveal changes in the size and direction of influence a factor has on learning, not just the demonstration that it has an effect. Two planned extensions of this work are to investigate the same intrapersonal factors in learning in different domains and to investigate how the factors known to optimise learning interact with motivation.

Download Tom Stafford’s PowerPoint presentation as a pdf file

Visit Tom Stafford’s webpage