Learning From Multiple Representations: Prior Knowledge Moderates the Beneficial Effects of Signals and Abstract Graphics.

Andrea Vogt,Ingmar Baetge,Tina Seufert,Melina Klepsch

doi:10.3389/fpsyg.2020.601125

Abstract

Multimedia learning research addresses the question of how to design instructional material effectively. Signaling and adding graphics are typical instructional means that might support constructing a mental model, particularly when learning abstract content from multiple representations. Although signals can help to select relevant aspects of the learning content, additional graphics could help to visualize mentally the subject matter. Learners’ prior knowledge is an important factor for the effectiveness of both types of support: signals and added graphics. Therefore, we conducted an experimental study situated in a university course of computer science with N = 124 participants. In our 2 × 2 factorial design, we investigated the effects of signals and illustrating graphics on learning outcomes and their potential interplay. Based on our regression analysis, we revealed prior knowledge as a significant moderator. Although learners with low levels of prior knowledge can profit from all types of help but still gain rather weak learning outcomes, learners with medium levels of prior knowledge profit from the synergy of both helps. With higher levels of prior knowledge, signals were particularly hampering. To improve the understanding of these supportive or hampering effects, a more fine-grained analysis of these processes and motivational effects is necessary.

Highlights

In natural science and STEM education, learners are often confronted with highly abstract subject matters such as physical principles, mathematical systems, or computer programs
To ensure that no systematic group differences existed between the experimental conditions, a MANOVA was conducted with study time, prior knowledge, and verbal and spatial ability as dependent variables
When performing a Shapiro–Wilk test, multivariate normal distribution can be assumed for the variables learning outcome and prior knowledge for each experimental subgroup (p > 0.050), and the variances can be classified as homogenous based on the Bartlett’s test (p > 0.240)

Summary

Introduction

In natural science and STEM education, learners are often confronted with highly abstract subject matters such as physical principles, mathematical systems, or computer programs. The abstract content is often presented in abstract formats, such as mathematical or chemical formulas or computer code. Such unfamiliar, intangible formats are especially challenging for novice learners (van Meter et al, 2020). From an instructional design perspective, the question is how information on abstract subjects can be best conveyed. Teachers or text-book designers use additional representations such as explanative or exemplifying texts, diagrams, or pictures to overcome these difficulties and make the abstract content easier to understand. Besides the well-intentioned use of additional, that is, multiple representations to enrich the abstract aspects, learners have to link the easier, accessible

Objectives

Results

Discussion

Conclusion