Hegarty, Mary; Quilici, Jill; Narayanan, N Hari; Holmquist, Selma; Moreno, Roxana (1999). Multimedia Instruction: Lessons from Evaluation of a Theory-based Design. Journal of Educational Multimedia and Hypermedia, 2 (8), 119-150.


Abstract :

3 experiments were carried out to assess a model for designing multimedia manuals that explain how machines work. Results show that animated and hypermedia instructions provide a significant advantage over static text and diagram presentations when both text and diagram are designed according to the model.

A model of machine comprehension

Machine comprehension from text and diagram involves the following processes :
- machine decomposition by diagram parsing
- constructing a static mental model by making representaional connections (between text document and text representation, and graphic doc and graphic representation)
- making referential connections between the digrams and text information
- determining the causal chain of events in the machine
- constructing a dynamic mental modelby mental simulation and rule-based inference

Designing multimedia instructions using this model

- Section 1 : facilitating decomposition and referential connections by proving schema with labeled parts
- Section 2 : facilitating the decomposition into to functional units by an exploded view that spatially separate the two units
- Section 3 : facilitating the construction of a static mental model of the machine, by proving users with information on units in the real world, and its linkages with other components
- Section 4 : encourage users to reason about the causal chain of events before viweing the animation, since self-generated explanations improve learning (Chi, deLeeuw, Chiu & LaVancher, 1994). Means : multiple choice questions with appropriate feedback
- Section 5 : explanantion to help understand the propagation of causality, with auditory comments and changing highlights on a static schema
- SEction 6: animation of the whole machine
- Section 7 : explanation of underlying physics principles

Experiment 1

Comparison between hypermedia manual, full text (equivalent in text and diagrams without hyperlink nor animation), and causal text condition (about 50% correct)
Results : No difference on any learning outcomes, not even for mental animation or causal questions

Experiment 2

3 conditions on paper : full text, causal text and diagram only
Results : people in the full text condition no more able to animate the system that users in the causal text condition. No difference either with these two conditions and diagram only, except for the bicycle tire pump, for which the text was giving an unvisible information.

Expe 3 : replicates expe 2 with open ended questions instead of multiple choice
Same results


Explanation of why is animation not better than static instructions : when the machine has several causal chains that merge, subjects repeatedly animate components near the merging point. Animations that continuously run from strat to end prevent people to do so.
Faraday & stuttclife (1997) : using eye movement recording, they found that it was necessary to draw users'attention to part of the diagram by highlighting components that were going to animate.


Animation should not been used when people could easily infer the information from static diagram and text. Similarly, text is useful only when the information can is not visible and can not be inferred from the diagram .