Figure 1.
Networks are based on student recall of interactions.
Links in these networks are based on the assumption that students remember interactions they had as a part of studying (a). Once a week for nine weeks, sessions were set up so students could indicate the names of fellow students with whom they remembered having communicated about problem solving in physics. Students were asked to consider only the preceding week, and they responded online (b). For each session, the responses were used to create a directed network that models the interactions between students. In each network a directed link exists from a node B to a node A if student B indicated student A in the corresponding session. (c)
Figure 2.
Four networks from different weeks indicate how student interactions develop.
The density for each network. Colors represent different sections of students. Females are represented by large circles, males by smaller circles. For each week right panel shows the degree distributions (in and out). The middle panels show the accumulated link weight distribution, indicating the extent to which links are reused. The total numbers of nodes in the networks are 161, 152, 154, and 139 respectively.
Figure 3.
Student interaction networks develop over time.
(a) Number of total, new, and reestablished links. New links, , calculated based on the preceding week.
is the number of links in a given week which are also present in at least one of the preceding weeks.
for week 2 (not shown). (b) Average degree with error bars for each week. Numbers at data points indicate the size,
, of the network and the diameter.
Figure 4.
The fraction of completely new links relative to the total number of links, , decreases over time.
This is not due to students saturating connections, since summed over all weeks; students use a total of 5% of all available links.
Table 1.
Infomap results and performance.
Figure 5.
Segregation Z-scores for gender, grade, and section show different behaviors.
The shaded area indicates the non-significant region, where . Students segregate according to section, and somewhat according to gender but not according to end-of-course grade.
Figure 6.
Large scale structures show between-week student movement and within-week information flow.
Top: Alluvial diagram for communities of the four networks displayed in Figure 2. The height of a block indicates the number of students in the community (see scale). The thicknesses of the gray streamlines between groups in different weeks indicate between group movements; thicker lines indicate that more students moved together. The color of a box representing a community indicates whether it is significantly segregated () with respect to the given attribute. Bottom: Maps of community structure for the same networks. Node sizes are proportional to accumulated flow rate for a particular community. Labels on the map correspond to labels in the alluvial diagram, allowing for comparison between community size and flow. Arrow sizes are proportional to the information flow between groups as calculated by Infomap. Color codes in the maps have the same meaning as in the alluvial diagram. The total number of communities each week is 28, 28, 22, and 20, respectively.