Online aggression from a sociological perspective: An integrative view on determinants and possible countermeasures

Sebastian Weingartner, Lea StahelACL 2019

TL;DR

This paper proposes a novel sociological framework to explain online aggression by integrating individual, situational, and social-structural factors. It addresses the lack of a unified theory and proposes an empirical strategy to link digital commenting data with offline survey data for a comprehensive understanding.

What Problem Does This Paper Address?

Online aggression, including offensive and threatening comments, has become a significant issue with detrimental effects on individuals and social harmony. While previous research has identified various determinants of online aggression, these studies often focus on single factors and lack an overarching theoretical framework to explain their interplay. Furthermore, socio-structural determinants have been largely overlooked, and there's a scarcity of studies that systematically link digital commenting behavior with offline demographic and background data of aggressors. This paper aims to bridge these gaps by developing a comprehensive sociological model and outlining an empirical approach to test it.

Methods

The paper introduces a theoretical model based on "structural individualism" and "frame-selection." Structural individualism posits that social phenomena arise from individual decisions embedded within social structures, which in turn influence actors' goals, beliefs, and opportunities. The frame-selection model (MFS) is adopted to explain how individuals define situations by synchronizing situational cues with internalized knowledge (frames). This process determines which individual beliefs (descriptive and evaluative) are activated and which situational constraints are perceived, ultimately guiding behavior. The model emphasizes that online aggression results from the interaction between individual and situational determinants, rather than from either in isolation.

For empirical testing, the authors propose a large-scale quantitative survey in German-speaking Switzerland. This survey will collect data on individual, situational, and social-structural determinants of online aggression. A key methodological innovation is the proposed two-step sampling strategy:

  1. Identifying Commentators: Collaborating with a large Swiss media corporation to analyze their news comment dataset. This involves using human/automated content analysis to categorize commentators into "frequently aggressive," "occasionally aggressive," and "non-aggressive" groups based on their submitted comments.
  2. Sampling Non-Commentators: Using a random sample of the general population to recruit individuals who do not engage in online commenting.

The survey data will then be matched with the commentators' digital behavior, allowing for a comprehensive analysis of determinants.

Datasets

The paper does not explicitly list datasets used for training or evaluation in the traditional machine learning sense. However, it proposes the use of:

  • A large dataset of news comments from a Swiss media corporation, including meta-data such as submission time. This dataset will be used for content analysis to identify and categorize online aggressors.
  • Survey data collected from four groups: frequent OA commentators, occasional OA commentators, non-OA commentators, and non-commentators. This data will contain information on individual, situational, and social-structural determinants.

The paper does not introduce a new benchmark dataset but proposes a novel approach to link existing digital data with newly collected survey data.

Key Insights and Ideas

  • Integration is Key: The most significant insight is the emphasis on integrating individual, situational, and social-structural determinants. The paper argues that understanding online aggression requires looking beyond isolated factors to their complex interplay.
  • The Role of "Definition of the Situation": The frame-selection model highlights that how individuals perceive and define a situation is crucial. This subjective interpretation, influenced by both internal beliefs and external cues, mediates the relationship between determinants and aggressive behavior.
  • Social-Structural Factors Matter: The paper strongly advocates for the inclusion of socio-demographic and socio-economic factors, which have been historically underrepresented in online aggression research. It suggests these factors can shape individual beliefs and opportunities, influencing online behavior.
  • Bridging the Digital-Offline Divide: The proposed empirical strategy of matching digital commenting data with offline survey data is a novel approach to overcome a significant methodological gap, enabling a more holistic understanding of aggressors.

Key Results

As this paper introduces a theoretical model and an empirical strategy, it does not present quantitative results from an executed study. The "results" are conceptual:

  • A theoretical model that integrates individual, situational, and social-structural determinants of online aggression.
  • A proposed empirical strategy for testing this model, which involves a novel data linkage approach.

Limitations and Future Work

The authors acknowledge several limitations and suggest future research directions:

  • Theoretical Gaps: The paper highlights that existing theories of cyberbullying are "sparse and piecemeal" and lack conceptual development. A major future task is to develop comprehensive theoretical models that integrate micro, meso, and macro levels of explanation.
  • Systematic Link to Social-Structural Factors: There is a need to systematically relate online aggression to social-structural factors, especially exploring potential "civility divides" and how socio-economic status, gender, and age might influence online behavior.
  • Methodological Challenges: The paper notes the difficulty in sampling online aggressors due to their relative rarity. The proposed two-step sampling strategy is an attempt to address this, but its effectiveness will need to be evaluated.
  • Empirical Testing: The paper outlines an empirical strategy but does not present the results of its execution. Future work will involve conducting the proposed survey and analyzing the data to empirically validate the theoretical model.
  • Data Protection and Ethics: The authors discuss the ethical considerations for their proposed empirical approach, particularly regarding data privacy and participant recruitment, and state that ethics approval will be sought.

Summary updated

BibTeX
@inproceedings{weingartner2019online,
  author = {Sebastian Weingartner and Lea Stahel},
  title = {{Online aggression from a sociological perspective: An integrative view on determinants and possible countermeasures}},
  booktitle = {Annual Meeting of the Association for Computational Linguistics},
  year = {2019},
}
Figure