A meta-analysis study examines whether differences in learning efficacy between static and dynamic visuals exist.
DR Juan Cristóbal Castro-Alonso, an educational psychology researcher with focus on multimedia at Universidad de Chile, has led a meta-analysis to establish whether gender imbalance exists in research on learning through visualisations and how it might be affecting study outcomes. Previous studies have demonstrated that because they have ignored gender distribution among its participants, multimedia learning research comparing effectiveness of static and dynamic visualisations in STEM disciplines (science, technology, engineering or mathematics) have revealed mixed results. This is mainly due to the researchers’ failure to report gender compositions, and therefore making it impossible to draw conclusions on the moderating effects of these differences on learning. Castro-Alonso’s motivation behind this study was to gain a better understanding on whether or not the use of multimedia makes learning science more efficient, and if it does, then under what circumstances.
In previous studies, Castro-Alonso et al have discovered that different types of instructional materials might be better at supporting the learning of different types of tasks. Arguably, these differences are because of the functioning of specific neural systems that are behind the acquisition of particular skills and are also mediated by the amount of mental resources required to process the specific multimedia materials. Take for instance videos or animations (which are both dynamic materials), in which the learner must often process more information due to the transient nature of dynamic presentations. However, it has been found that dynamic materials appear to be better at supporting manipulative tasks (for instance, tasks that require physical action; such as replicating a Lego model). At the same time however, static images seem more effective at supporting learning for non-manipulative tasks (for example, tasks that are more cognitive than physical; such as solving abstract symbol problems).
Cognitive load theory can be used to explain the difference between dynamic materials and static images. The theory suggests that a learner has a limited amount of cognitive resources (due to their limited working memory), and therefore any loss of this capacity to tasks that do not directly support learning or require more processing, is as a result detrimental to the overall effectiveness of learning.
Mirror neurons
With a more evolutionary approach towards cognitive load theory, some researchers have suggested that dynamic images might be able to support learning tasks that require movement. This is because manipulative tasks evolved as a primary skill in humans, and as a result are likely processed by a separate system (such as by mirror neurons).
Mirror neurons are a specialised circuit, which allows learning of manipulative skills from transient images without any extra effort. Opposite to this, learning a non-manipulative skill from a moving image requires effort to remember the information before the learner even attempts the task at hand. As a result, by the time the student actually attempts the task itself, processing learning materials has already diminished their cognitive resources. When a situation like this occurs, the use of dynamic multimedia may hinder rather than support the students’ performance.
However, Castro-Alonso suggests that the data on learning from visuals is complex and sometimes difficult to interpret. Firstly, it remains unclear whether differences between learning efficacy from static versus dynamic images in fact exist. For instance, studies that report these differences often neglect to control for many potentially moderating variables that relate to the actual materials (these factors include the appeal, media, size, and interaction). Secondly, participant-related variable (i.e., gender, spatial abilities, and/or attained level of education) remain an area that has been largely unexplored by researchers.
The study
Castro-Alonso et al have recently published a meta-analysis (a study compiled of other studies) that examined whether differences in learning efficacy between static and dynamic visuals, exist when studies are pooled together. Alongside this, the study also set out to identify any factors that have an effect on any existing differences. In this study, researchers searched for and included studies that reported results of randomised experiments where two groups of students were compared: a group learning from static images and a group leaning from dynamic visualisations. The studies that were chosen to be included reported measurable outcomes that could be used in statistical analysis and where they included tasks that could be categorised as either STEM-related (science, technology, engineering or mathematics) or manipulative-procedural (e.g. paperfolding, Lego models). If the gender ratio for a participant sample was not reported, the study was excluded from this meta-analysis.
After combining and analysing results from 46 studies that included over 5,000 participants, Castro-Alonso et al found that there was indeed some difference in the efficacy of learning in the groups exposed to the static versus dynamic images. The results showed that students that worked with dynamic visuals, performed slightly better. In addition to this, gender appeared to impact the benefits of visualisations. For example, in samples that included more females, static visualisations performed better, whereas samples with fewer females and more males seemed to favour dynamic visualisations.
Lets’ talk about gender bias
In research concerning the use of visuals in the classroom, the gender of study participants is often neglected as a possible moderator, so much so that relevant ratios are often left out of study reports. In the meta-analysis, 48 of all eligible studies had to be excluded because they did not report gender ratios. This is a problem because by not controlling for gender in experimental designs and in data analysis, as well as not reporting gender ratios, this has produced a body of research that is unclear and difficult to interpret. Moreover, these findings are often used as basis for policy recommendations, even though they lack a strong evidence base.
Building on the results of their meta-analysis, Castro-Alonso is leading a collaboration that aims to better understand the role of gender and visuospatial processing in learning STEM from multimedia materials. Specifically, he and colleagues will examine how cognitive load is moderated by students’ characteristics in both immediate and delayed tests on the rotation and translation of organic chemistry molecular representations (this topic is often challenging for undergraduates).
The study is designed to involve a randomised sample of 720 undergraduates in Chile (50% female) and is combined with outreach seminars to increase teachers’ and students’ awareness about cognitive load and gender differences in STEM education. Once the study has been completed, the results will be shared with the global scientific community through publications in Web of Science journals. Castro-Alonso hopes that greater insight into the role of gender and visuospatial processing in learning science and other STEM topics through multimedia will result in the development of study materials that are well suited for all students. He also hopes that engagement with the public will increase awareness of the evidence behind teaching and learning recommendations and will result in policies and investments that are supported by scientific methods. A highly recommended resource to gain such insights is Dr Castro-Alonso’s book on ‘Visuospatial processing for education in health and natural sciences’ (published by Springer Ltd), where leading world experts discuss a number of critical topics.
Juan Cristóbal Castro-Alonso
Associate Researcher
Institute of Education, Center for Advanced Research in Education
Universidad de Chile
+56 2 2978 1238
jccastro@ciae.uchile.cl
