Meta-Analysis to Support an Integrated Theory of Multimedia Learning: An ECR Synthesis Proposal    

The research reported here was supported by the US National Science Foundation, through award #1661231 to the University of Illinois, Urbana-Champaign. The opinions expressed are those of the authors and do not represent views of the National Science Foundation.


Overview of this Synthesis Proposal.

Despite the ubiquity and importance of multimedia (MM) presentations for learning STEM content in formal and informal settings, research consistently documents low levels of learning. The literature is diverse and scattered, but is growing rapidly; consequently, inadequacies of existing theories are increasingly being recognized. We propose a meta-analytic review of the literature on learning from text+diagrams, animation, simulations, hyper linked documents, and other MM. The review will be conducted in the context of a new meta-theory of multimedia learning. Coding and analyses will be the main activity in Year 1 and conference and web dissemination in Year 2.

Intellectual Merit.

Students are confronted with text + diagram representations of STEM content in textbooks, the Internet, scientific simulations, and other multimedia on a daily basis. Numerous researchers have pointed out the incompleteness of current theories of multimedia learning (e.g., de Jong, 2010); our team has developed a new integrative theory of multimedia learning, and we are seeking funding to complete a meta-analytic review to support the theory. We will then disseminate the theory and meta-analytic literature review as a book-length monograph (Springer) and conference presentations and workshops.


By multimedia we mean physical or electronic display that uses any combination of presented verbal (printed or spoken words) and visual (images) materials; by learning we mean gaining knowledge, including forming an integrated mental model of the phenomenon that merges information from the verbally-presented and visually-presented materials, and transfer to uninstructed material and/or domain.


Our new, integrative theory merges and encompasses findings from four existing theories: the Cognitive Theory of MM Learning, Integrated theory of Text and Picture Comprehension, Cognitive Load Theory, and the Information Processing Theory of MM Learning. Collectively, these theories consider individual differences, characteristics of multimedia Stimuli, learning Tasks, learner strategy use/Inference, and features of the Content. Our new theory--the MiSTIC model--encompasses all of these, and makes specific predictions about how individual differences such as working memory and prior knowledge act as critical moderators of the effects of stimuli, tasks, and content on strategies+inference.


We propose to conduct a thorough meta-analytic review of peer-reviewed research articles published 2000-2016 on learning from educational MM with STEM content to identify major themes and gaps. Studies include regular populations in 1st grade – UG, formal or informal STEM learning contexts, and we expect ~500 articles. We will search for studies that examine effects of variability in features of learners, stimuli, learning tasks, learning strategies/inference, and to-be-learned content. The planned approach is to: 1) use PsycINFO, ERIC, and Scopus databases to locate articles, 2) check for missed articles from among those analyzed in meta-analyses and other lit reviews, and 3) check for missed articles citing already-located articles and key authors through the Social Sciences Citation Index as well as references cited in the already-located articles. We will restrict the review to studies that include an effect size or provide enough descriptive statistics that an ES for an outcome variable can be calculated.


Broader Impacts.

Both basic and applied research in STEM multimedia have been hampered by the lack of an integrative theory of multimedia learning. Integrating across multiple theories of multimedia learning and multiple communities of practice (e.g., static diagrams and simulations) has the potential to yield better research designs in studies of educational multimedia and more powerful practical applications of results with diverse samples. Specifically, our focus on learners and learner individual differences holds huge promise for the design of future learning environments tailored to specific learner characteristics (e.g., high spatial, low working memory). Our model may also be able to explain null findings from interventions—e.g., if participants include low- and high-spatial students and there is a spatial x treatment interaction, an intervention that is actually successful with some students will appear to have no effect. The project also has the potential to disseminate sound findings about learning with multimedia to broad audiences, such as science and math teachers and postsecondary instructors, as well as Discipline-Based Education Researchers. The results will be informative for those involved in designing educational multimedia, such as electronic textbooks and Web-based materials. 

Both basic and applied research in STEM multimedia have been hampered by the lack of an integrative theory of multimedia learning. 

Dr. Jennifer Cromley, PI 



Natasha Beretvas