Text analysis of studies using ERGMs or SAOMs in educational contexts

Basic data for this talk

Information about the event

Abstract

Social network analysis has frequently been used in educational contexts, with a notable increase in published articles in the last decade. Despite the increased interest and application of the method, only a few sources provide an overview of its use in educational research (e.g. Robins, 2015). This study aims to give an overview of this growing field by identifying relevant topics. Especially in educational contexts social structures, their formation and influences are relevant topics. For example, the intertwined link between social embeddedness, wellbeing and academic achievement is well documented. However, to influence pleasant and unpleasant tie formation (as educators often wish to do), it is necessary to reveal all relevant network effects and their interplay. In educational research, only few investigate the interdependency of networks – even though they are highly relevant to the field and practitioners. In our study, we therefore focus on studies providing this information of interplay using Exponential Random Graph Models (ERGM) and Stochastic Actor Oriented Models (SAOM) as two common advanced methods of social network analysis. While ERGMs are (mainly) used to analyze cross-sectional data, SAOMs can be used to examine longitudinal data. Both allow endogenous and exogenous network effects to be mathematically controlled. This science mapping aims to close this gap by quantifying the frequency of research topics and reveal the field’s publication structure in studies using social network analysis, especially ERGMs and SAOMs, in educational contexts. Using a selected Boolean search strategy, empirical, English-language, peer-reviewed studies were searched for in the "Web of Science" database of which publications between 2013 and 2023 are considered. We identified 223 studies on social school class networks in the “k-12” and university context using ERGMs or SOAMs. For each study, characteristics such as year and source of publication were analyzed using bibliographic evaluation procedures. Also, the abstracts were used for text analysis to identify topics within the subject area. Content obtained through bibliometric information abstracts and titles are categorized and presented, as follows: - How? What are main publication sources? - Where? In which countries and regions are studies conducted? - What? Which topics arise (e.g. What terms arise most frequent? Which topic clusters can be modeled)? Results for a preliminary sample of 160 studies indicate that social network analysis is a growing field in educational research. Using the “stm” R package we identified 11 thematic clusters from the most frequent terms in the abstracts: 1) “friendship formation in high schools”; 2) “friends and identification processes” ; 3) “peer selection and influence processes in adolescence”; 4) “adolescents’ friendships and mental health”; 5) “peer selection and influence by academic achievement”; 6) “ethnic minorities and formation of friendship”; 7) “peer selection processes by gender”; 8) “bullying and victimization processes”; 9) “collaboration ties in school and university”, 10) “students and teachers”, 11) “adolescents and health”. Advanced text analytics procedures and bibliometric methods are planned with the total sample of studies giving a whole picture of what is conducted with the two main model approaches of social network analysis in educational contexts.

Speakers from the University of Münster

Marticke, Sophie

Professorship for methods of empirical educational research (Prof. van Ophuysen)