USING LEARNING CURVES FOR FORMATIVE ASSESSMENT OF DELIBERATE PRACTICE
Martin V. Pusic,NYU School of Medicine,New York,NY

Rationale: In recent years there has been an explosion of interest in the systematic use of the deliberate practice instructional strategy as advocated by Anders Ericsson and others. This is generally done within a competency framework in which the learner is encouraged to participate in effortful repetitions, with feedback, of a given activity until they have improved to an agreed upon level of competence. The learning curve is often used to represent the form and quality of learning with successive repetitions; however, the question of how to graph and use a reliable and valid learning curve is rarely addressed in health education practice.

Objectives: By the end of the session, participants will be able to: list at least three learning situations where a learning curve would be applicable; list four threats to the validity of a learning curve; organize data in a Microsoft Excel spreadsheet so as to produce first a scatter plot approximation to a learning curve and subsequently a fitted line; and describe one setting in their own practice where learning curves could be of utility.

Methods and Content: The primary goal of this session is to allow pediatric educators to consider the use of learning curves in their education mission. Education researchers could also benefit from the framework that we will present. We will present two common scenarios where a learning curve representation could be useful: practice of a procedural skill (e.g. iv insertion) with a binary outcome and practice of a cognitive skill with a continuous outcome (e.g ECG interpretation). Using examples from our own research and the literature, we will present a validity framework for the use of learning curves. Participants will help each other identify potential uses of learning curves for formative feedback to learners. Subsequently, using a mini-journal club format, participants will assess the validity of a study in which learning curves are the main outcome. Using example datasets, participants will use Microsoft Excel" to properly structure data so as to produce first a scatter diagram and then a fitted learning curve with an initial analysis as to slope and intercept. Finally, using these examples, we will present an overview of more sophisticated analyses.