Abstract:

According to literature, networking plays an important role when it comes to reinforce the student performance. Different theories have been deployed underpinning the idea of the need for networking for a better academic performance, among the most relevant of them the one carried out by Picciano in 2002, or that presented in 2013 by Karpinski et al. The idea underneath it is that the broader the student’s network is, the easier to specialise. Factors which might explain this are splitting tasks, sharing knowledge, or giving support each other, which may lead to lower difficulties. However, the research done so far speculates plainly with this, assuming it as a work statement rather than a verified fact. Actually very little research has been done about whether it works at any time, any kind of student or any sort of learning process.

In this paper we try to deepen in this concept, presenting data gathered within the period 2014-2018, coming from a single Human Resources subject, belonging to Tourism Faculty. Thus, data from 994 students was registered, using their final grade as the explained variable. Independent variables were students’ partial grades and whether they co-authored works or assignments through their assessment, using this for building a network model in software Gephi 0.9.2. Right after this we could execute a statistical study and calculate for any given student their measures according to the place he or she was keeping within the network. Students who co-authored more assignments received a higher score in centrality, for instance. We also calculated their respective degree, betweeness, closeness, eccentrality, clustering coefficient, etc. Furthermore, apart from the academic performance variables, we also included in our study dummy variables, such as the language of teaching, or the group they belonged. After getting the network analysis, we transferred this information to a SPSS 25 matrix, for running a linear regression multivariable model. 43 students were considered outliers (Using Cook’s Distance), eventually remaining 901 as useful data. With this we could calculate a model that suggested that not the group nor the language was relevant for predicting the final score got by the student, since no effect seemed to have on the final grades. On the other hand, their final score appeared significantly linked with four statistical measures associated with their networking behaviour (measured through the papers and assignments co-authored). Namely, these variables were eccentricity and eigencentrality (positive relationship), or degree and clustering (negative relationship). From our perspective, and according to our results, networking cannot be understood as a tool plainly useful in any way, for it needs to be qualified. The student could need to overcome several drawbacks even in the event they freely choose their networking team (e. g. lack of time, disparity of mates, disagreements with other members of the network, disengagement, etc). Turns out that the limitations of this study, like the short period of time available for analysis, or the need for including information from other fields and courses, also constrains the chances of applying it in a wider context. The study is part of a wider project which intends to improve the understanding of the student’s behaviour when it comes to organise their work on their own, the responsibilities they take, and even how they tackle their job search strategy.

Keywords:

Student's performance, academic behaviour, social network analysis, networking.