Decoding the Political and Social Landscape of Perceived Neighborhood Context

• In Event: The Built Environment, Urban Context, and Political Behavior

Abstract

This project aims to investigate the multifaceted nature of neighborhood context and its impact on political behavior by integrating physical features of the built environment with subjective perceptions. While traditional research has predominantly relied on demographic data to operationalize neighborhood context, there is a growing recognition that context extends beyond demographic factors alone. This study acknowledges the importance of the built environment in shaping individuals' experiences. It departs from the traditional paradigm by incorporating subjective measures and physical features, providing a more nuanced understanding of the intricate dynamics between neighborhood context and political behavior.

To provide a more comprehensive measure of context, we employ a machine learning approach that combines physical features extracted from "street view" imagery with subjective perceptions. The project utilizes 360-degree street-level images obtained from Mapillary, a non-proprietary platform for geotagged street-level photos. We then use object detection models to identify physical features in the images and combine these data with human coders on Amazon Mechanical Turk classifications of these images along dimensions such as disorder, dilapidation, wealth, and safety. This classification process enables the creation of a model capable of predicting perceptions of new places, providing a comprehensive measure of perceived neighborhood context, as well as an exploration of the relationship between physical features and perceptions of places. By integrating physical features and subjective perceptions, this research explores the relationship between environmental attributes and individuals' perceptions of their surroundings. The dataset generated through this integration will be analyzed to assess how physical features of the built environment predict subjective perceptions. The findings will contribute to a deeper understanding of the multidimensional nature of neighborhood context and, ultimately, its implications for political behavior.

Authors

• Paige Bollen, Ohio State University

• Melissa Sands, London School of Economics

• Joe Higton, New York University