The Effect of Mixed Visualizations on Learning From Science-Related Texts on the Internet

• In Event: 35.030 - Multiple Representations and Multimedia: Student Learning and Instruction

Abstract

Objectives and theoretical framework
Web-based science-related documents are increasingly used in higher education classrooms. Such documents frequently include graphs, photographs or both. Previous experiments showed that graphs can increase the plausibility of web-based textual information, depending on learners’ amount of knowledge about visualization conventions, and at the same time lead to stronger text comprehension (Isberner, Richter, Maier, Knuth-Herzig, Horz & Schnotz, 2013). Photographs have been shown to increase the plausibility of the information as well but depending on learners’ amount of web experience and leading to a weaker text comprehension, (Knuth-Herzig, Horz, Isberner, Maier & Richter, submitted).
The question, remaining hitherto unanswered is, which of the two kinds of visualizations stronger determines information processing if both of them are present in a web-based text and which learner characteristics play a role in the plausibility assessment.

Methods and materials
In the present experiment, participants read two web-based texts (approximately 1200 words each) on a scientific topic while the presence of visualizations (line graphs and photographs) was systematically varied. As dependent variables, situation model strength was measured as indicator for text comprehension using a verification task adapted from Schmalhofer and Glavanov (1986) and perceived credibility of the texts was assessed using a 6point likert scale (Cronbachs a > .90) based on aspects of web-based credibility assessment according to Flanagin & Metzger (2000). Knowledge about scientific visualization conventions, prior domain knowledge and the amount of web experience were used as covariates.

Results
An ANCOVA for within-subject designs (N = 35) with perceived credibility as dependent variable yielded an interaction of knowledge about visualization conventions and the presence of graphs, F (1, 29) = 7.0, p = .01, hp2 = .20. Simple slopes showed that the perceived credibility of the texts with mixed visualizations increased with participants’ amount of knowledge about visualization conventions (B = 0.49, SEB = 0.16, p = .002) whereas the slope was not significant for texts without graphs. The more knowledge about visualization conventions participants possessed, the more they were inclined to perceive the information provided by the text with mixed visualizations as credible.

A second ANCOVA with situation model strength as dependent variable revealed an interaction of the difference in perceived credibility with the presence of graphs, F (1, 28) = 4.9, p = .04, hp2 = .15. The credibility difference exerted a positive effect on the situation model for texts with mixed visualizations (B = 0.52, SEB = 0.09, p = .01). The more credible participants perceived texts with mixed visualizations, the stronger was their situation model for them.
Additional analyses revealed an indirect effect of the knowledge about visualization conventions on differences in situation model strength via differences in perceived credibility.

Scientific significance
Web-based texts with mixed visualizations have long since found their way into the higher education classroom. Hitherto such texts did not get as much attention in text-picture and text comprehension research as texts with only one type of visualization. The present results hint towards a dominance of graphs concerning text processing and comprehension.

Authors

• Katja Knuth-Herzig, Goethe University

• Holger Horz, Goethe-Universität Frankfurt