Internet Learning Volume 3, Number 2, Fall 2014 | Page 86

Visualizing Knowledge Networks in Online Courses Interactive 1. Comparative Corpus Diagrams. Finally, in Figure 7, we show comparative corpora for responseLevel – the level of the conversation tree at which the discussant contributes each response. The slightly cooler cast to Renlit’s corpus indicates that Renlit tends to participate somewhat later in a thread than Loret – compare, for example, at L2 (orange) and L4 (light green). Note also that Loret and Renlit each have four responses at L1, indicating that they have each initiated four threads. Naya, on the other hand, has no posts at L1 because instructors do not typically respond directly to their own discussion prompts. This may seem self-evident, but it is encouraging to see an intuitive result illustrated so plainly in the data. Finally, to the question of Loret as a student who presents as instructor-like in certain ways, what happens if we disregard the L1 responses in the Loret and Renlit diagrams? The remainder of Loret’s corpus falls somewhat between Naya’s and Renlit’s for wordCountAvg, as well as for the distribution of responseLevels. For example, Loret’s proportion of L2 to L4 posts is much more similar to Naya’s than it is to Renlit’s. D. RQ1 Discussion A simple graph traversal, derived from the schema shown in Figure 3, can yield a participant corpus data structure that is amenable to visual and statistical analysis. The examples above show that comparative corpus diagrams can be used as exploratory tools for generating rough insights about individual differences among discussants, and as useful models to support reasoning about individuals. They 85