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.

Previous research suggests that bystanders of conflicts use a range of intervention strategies. Yet, much less is known about other actions—beyond intervention—that bystanders might engage in during conflicts. Further, while prior studies reveal that gender differences emerge in bystander behavior, few studies have assessed the ecological validity of such potential differences in bystander actions during real-life conflicts. Addressing this concern, we systematically observed the diverse bystander behaviors of individuals presenting as men and women in real-life public conflicts captured on CCTV. We observed 67 public conflicts in the inner city of Amsterdam, the Netherlands. Using a qualitative approach, we first identified the broad spectrum of actions that bystanders engaged in. We then ran linear probability models to examine the relationship between bystander’s gender presentation and bystanders’ engagement in seven bystander actions (N = 1,959), followed by a multimodel analysis to test the robustness of these findings. Results indicate that bystanders engaged in a diversity of actions, ranging from inattentive (i.e., glancing while moving) and reactive actions (e.g., laughing) to physical forms of intervention. Unexpectedly, women were not more likely to engage in affiliative forms of intervention (e.g., calming hand gestures, non-forceful touching, and practical help). In addition to physical intervention, men were more likely to react to conflicts by laughing, filming, or cheering. The only type of action that was more typical among women than men was inattention (i.e., glancing while moving). Our results show that bystander behavior in public space is carried out in gendered ways, albeit in a less clear-cut manner than expected. [ABSTRACT FROM AUTHOR]

Using longitudinal survey and space–time budget data from adolescents in the Netherlands, this study examines whether the code of the street affects violent delinquency and victimization, and whether these effects vary by lifestyle. Values similar to the code (i.e., conflict positioning) were related to higher risk for violent delinquency; this relationship was stronger when involvement in public unstructured socializing was high. No relationship was found between these values and victimization. The findings suggest a nuanced relationship between the code of the street and violent delinquency and speak to the code of the street theory’s applicability outside the United States. [ABSTRACT FROM AUTHOR]