Search

Social scientists increasingly use video data, but large-scale analysis of its content is often constrained by scarce manual coding resources. Upscaling may be possible with the application of automated coding procedures, which are being developed in the field of computer vision. Here, we introduce computer vision to social scientists, review the state-of-the-art in relevant subfields, and provide a working example of how computer vision can be applied in empirical sociological work. Our application involves defining a ground truth by human coders, developing an algorithm for automated coding, testing the performance of the algorithm against the ground truth, and running the algorithm on a large-scale dataset of CCTV images. The working example concerns monitoring social distancing behavior in public space over more than a year of the COVID-19 pandemic. Finally, we discuss prospects for the use of computer vision in empirical social science research and address technical and ethical challenges.

As intimate partner violence (IPV) often remains unknown to police, bystanders can play a crucial role in prevention and further escalation of IPV. However, little is known about what brings them into action by reporting incidents of IPV to authorities. As such, we use statements of bystanders who filed reports about IPV incidents to an official domestic violence agency in the Netherlands (N = 78), to investigate the reasoning and motivations for reporting their suspicions. Results show that the reasons for bystanders to report IPV differ depending on the relational dynamics between partners. In situations perceived as intimate terrorism, involving a hierarchical abusive relationship between a man offender and a woman victim, bystanders primarily reported when previous helping initiatives proved inefficient, and they did so to prevent further harm, often particularly in relation to the woman victim. In situations perceived as situational couple violence, involving a symmetrical abusive relationship, bystanders primarily reported when escalation appeared, and they did so to prevent further harm to involved children. We conclude that bystanders report IPV incidents when the need for help is clear, and their motivation for acting concerns the well-being of victims. Our findings add to the growing body of evidence about real-life bystander intervention in emergencies and highlight the need for understanding intervention as context-specific in order to design effective intervention initiatives. [ABSTRACT FROM AUTHOR]

The causes of riots have been extensively researched. Comparatively, little is known about how they are prevented from occurring. We address this gap with a qualitative study of the role of formal and informal guardians in Amsterdam in January 2021, when public disorder was widespread across the Netherlands after the government had announced a curfew to curb the spread of the Covid−19 virus. We used CCTV footage of two gatherings that had elements of disorder and two that occurred without them. We also rely on interviews with 40 so-called "intimate handlers" who were present during these gatherings to understand how they managed them. We find that the presence and actions of intimate handlers collaborating with the police during the gatherings, effectuated through frequent affiliative contacts with the crowd, had direct de-escalatory effects, operative because of their well-developed social community bonds: participants in gatherings avoided jeopardizing these bonds of attachment, which also created a more positive image of the police through citizen-police collaborations. Our findings thus stress the importance of social bonds for the effectiveness of riot prevention and we consider practical implications for public disorder and large-scale crowd management. [ABSTRACT FROM AUTHOR]

Social scientists increasingly use video data, but large-scale analysis of its content is often constrained by scarce manual coding resources. Upscaling may be possible with the application of automated coding procedures, which are being developed in the field of computer vision. Here, we introduce computer vision to social scientists, review the state-of-the-art in relevant subfields, and provide a working example of how computer vision can be applied in empirical sociological work. Our application involves defining a ground truth by human coders, developing an algorithm for automated coding, testing the performance of the algorithm against the ground truth, and running the algorithm on a large-scale dataset of CCTV images. The working example concerns monitoring social distancing behavior in public space over more than a year of the COVID-19 pandemic. Finally, we discuss prospects for the use of computer vision in empirical social science research and address technical and ethical challenges. [ABSTRACT FROM AUTHOR]

Animal ethologists suggest that non-human primates console victims of aggression in a manner similar to humans. However, the empirical basis for this cross-species comparison is fragile, given that few studies have examined consolation behavior among humans. To address this gap, we revive and apply the underappreciated ethological branch of micro-sociology, which advocates the study of human interactions by applying ethological observation techniques. We thus systematically observed naturally occurring human consolation captured by video surveillance cameras in the aftermath of violent public assaults. Consistent with prior human and non-human primate research, social affiliation promoted consolatory helping. By contrast, we found no main effect of sex. A further exploratory analysis indicated an interaction effect between social affiliation and sex, with female affiliates having the largest probability of providing consolation. We discuss implications for the cross-species study of primate consolation and advocate that micro-sociology should reappraise ethological perspectives. [ABSTRACT FROM AUTHOR]

Third parties tend to take an active role and intervene in interpersonal conflicts in public. Previous research has shown that the level of aggression of these interventions determines how they influence the conflict. No previous study has, however, systematically investigated whether the aggression of third-party interventions is influenced by the development of the conflict situation. The objective of this study is twofold. First, the study determines the extent to which the aggression level of intervening third parties changes during the course of interpersonal conflicts. Second, the study identifies and investigates the factors that affect the aggression levels displayed by intervening third parties. We systematically observed and coded CCTV footage of 46 interpersonal conflicts in public space, recorded by surveillance cameras in Amsterdam, the Netherlands. The data included 565 intervention behaviors by 125 third parties. We recorded the levels of aggression of the individuals involved in the conflict and conducted a multinomial logistic regression analysis to investigate what influenced the aggression level of the third-party interventions. We found that the aggression levels of the preceding intervention behaviors by the third parties predict aggression levels of their subsequent interventions. This shows a consistency in third-party interventions over the course of a conflict. We also found that the aggression levels of the conflict parties that are the targets of the interventions influence the aggression levels of third-party intervention. This finding demonstrates that the development of the conflict situation influences how aggressive the third parties are. Our study emphasizes the importance of taking the interactional dynamics of interpersonal conflicts into consideration when explaining third-party behavior. [ABSTRACT FROM AUTHOR]

Guardians are a potential resource of conflict de-escalation but we still know little about their actual behaviour. In this article we investigate whom among the antagonists a guardian selects as a target when they intervene in an interpersonal conflict. We investigate this using CCTV footage from Amsterdam (the Netherlands) of 46 interpersonal conflicts in public spaces involving 641 interventions by 176 individuals. We find that guardians are more likely to target antagonists: (1) who have performed the most aggressive behaviours, (2) who are not simultaneously targeted by other guardians, (3) who are from their own social group, (4) who are men. The analysis shows that the behaviour of intervening guardians is shaped by multiple aspects of the complex and often ambiguous conflict situations. [ABSTRACT FROM AUTHOR]

Despite the risk of employee victimization, research on deviant citizen behavior in frontline encounters is limited. When passengers are issued a fine on the bus, they sometimes respond aggressively toward the ticket inspector. This article identifies how citizens present themselves in ticket fining events and examines how their self-representations are related with aggressive acts toward frontline workers. Data comprise video footage of 40 events recorded with occupational body-worn cameras from public buses in Denmark. Using in-depth qualitative video analysis, the study systematically analyzes the encounters for citizen self-presentation and aggression. Five self-presentation types are identified, comprising passengers appearing innocent, honest, wronged, defiant, or superior in the inspector encounters. In aggressive events, passengers in particular perform the defiant and superior self-presentation types and typically perform more types than in the nonaggressive events. The findings yield insights into aggressive frontline encounters with an enlarged understanding of the range of reactions and counterattacks during these events. Recognizing these presentations and changing responses could help frontline workers prevent exposure to aggression on the job. [ABSTRACT FROM AUTHOR]