Reviewing the differences in size, composition and structure between the personal networks of high‐ and low‐performing students

Abstract

An interesting aspect in the current literature about learning networks is the shift of focus from the understanding of the “whole network” of a course to the examination of the “personal networks” of individual students. This line of research is relatively new, based on small‐scale studies and diverse analysis techniques, which demands for more empirical research in order to contextualize the findings and to meta‐analyze the research methods. The main objective of this paper is to review two research questions posed by a previous British Journal of Educational Technology contribution by Shane Dawson in order to know whether the differences in personal network composition impact on the performance of students. The two questions were defined by Dawson as follows: (1) Are there significant differences in personal network composition between high‐ and low‐performing students? and (2) Do high‐performing students have larger personal networks than their low‐performing peers? In addition, the “clustered graphs” method used in this study allows the inclusion of the structural analysis of personal networks. In doing so, a new research question is addressed: (3) Are there significant differences in personal network structure between high‐ and low‐performing students? This paper tries to answer these questions in the context of two undergraduate, inter‐university and fully online courses, and two different technology‐enhanced learning environments (a virtual learning environment and a personal learning environment) where interactions took place indirectly through shared resources. The results show that the network behaviors of high‐ and low‐performing students' are strongly correlated, and that high‐performing students developed larger personal networks than low performers.

Exploring a Personal Social Knowledge Network (PSKN) to aid the observation of connectivist interaction for high‐ and low‐performing learners in connectivist massive open online courses

“…In summary, learning performance can be predicted by PSKN (network densities and structures) to some degree, especially by CU node and degree. The results supported the studies of Dawson () and Casquero et al () about high‐performing learners developed larger networks while low‐performing learners formed small ones. We used different approaches (knowledge creation), though, to observe interactions of network building in this study, which reinforce than the PSKN approach is feasible.…”

“…In this study, direct interaction means that the sending or receiving message can be identified (eg, in an email) between friends, while indirect interaction refers to collaboration based on topic, sharing and reflection, etc, which contributes to the CU number and tag. The results of this study further indicate that “when public spaces based on indirect interactions are set up in online courses, learners’ selection procedures for interaction are not focused on the individuals, but rather on those collaborative activities”(Casquero et al, , p. 26; Dawson, ). In summary, learning performance can be predicted by PSKN (network densities and structures) to some degree, especially by CU node and degree.…”

“…On the other hand, the relationship and predictive ability of both the friend node and social tag is less conclusive and weaker. Casquero et al () argues that there are two types of interactions in technology enhanced learning, direct interaction and indirect interaction. In this study, direct interaction means that the sending or receiving message can be identified (eg, in an email) between friends, while indirect interaction refers to collaboration based on topic, sharing and reflection, etc, which contributes to the CU number and tag.…”

“…We defined high‐performing learners as these in the top 50th percentile and low‐performing learners as those in the bottom 50th percentile. Contrarily, other past studies have chosen to rank the high‐performing learners in the above 75th percentile and low‐performers ranked from the 25th percentile and below, and we choose a different approach with the aim to differentiate connectivist interactions that could enrich the analysis for all learners (Casquero et al, , p. 22). Further, in order to verify if learning outcomes can be predicted by connectivist interactions displayed in PSKN, the performance was examined by various statistical tests.…”

“…Wang et al ’s study and their use of an overall analysis approach to examine connection interaction between groups, results in difficulties in diagnosing learners' interaction behaviors and dropout of individual learners. Therefore, analyzing the connectivist interactions based on connection‐building and the network‐forming process (Wang et al, ) from the perspective of individual, is an essential component (Casquero, Ovelar, Romo, & Benito, ) in cMOOCs.…”