Collaborative Research: The Geography of Information: Testing the Effects of Unequal Information in the Market for Rental Housing

在美国，由于种族、民族和社会阶层的不同，城市居民之间存在着相当大的隔离，这对社区的形成和不平等以及学校隔离都有影响。我们知道，在美国，大多数在城市里找房子的人现在都把互联网作为他们寻找新住处的主要来源。尽管有了这样的发展，但我们对租房者如何选择他们的住房知之甚少，因为关于可用住房的信息越来越多地转移到网上。虽然这种技术变革在很大程度上没有得到检验，但之前的一些研究表明，租房广告并不都是一样的；相反，它们会根据正在做广告的房屋所在社区的人口统计数据系统性地有所不同。该项目分析了网上发布的租房广告，以调查这些差异是否对人们在寻找住房时产生影响。了解个人如何解释他们在网上住房市场上看到的信息是解释为什么人们搬到某些地方而不是其他地方的关键，这对未来的居住不平等和种族/民族隔离有影响。研究结果将促进对住宅选择过程的理解，并帮助决策者扩大和平等地获得寻求住房者的信息，从而改善城市地区的社会和经济福祉。考虑到在线租赁广告在促进社区构成方面所起的作用，值得注意的是，我们对个人如何解释这些信息知之甚少。利用自然语言处理分析了美国50个最大城市在Craigslist上发布的数以百万计的租房广告，之前的工作已经确定了不同种族/民族和贫困率的社区中不同类型信息的分布模式。本提案将使用这些信息来测试这些现实世界差异的因果效应，这些差异是在五个大城市地区（洛杉矶、旧金山湾区、纽约、芝加哥和休斯顿）对个人住房和社区偏好的广告方式上产生的。该项目将在每个地区实施三项调查实验，以测试在线广告如何影响住房决策和居民对当地社区的看法。该项目将比较住房广告中的信息和对社区的看法与其他类型的信息（包括社区人口统计数据）对个人对住房单元的兴趣的影响。这五个大区域的选择允许该项目对黑人、拉丁裔和亚洲少数民族的受访者进行抽样调查。每个调查实验首先将被构建为代表城市地区（n=1,000/地区），但随后将从每个地区的特定少数民族人口中收集额外的受访者：在旧金山湾区额外收集100-300名亚洲受访者的超样本；纽约和芝加哥的黑人受访者；以及洛杉矶、芝加哥和休斯顿的拉丁裔受访者（包括说西班牙语的人）。通过对特定城市地区的少数民族居民进行代表性调查，该项目将通过差异经济状况调查和多元回归模型的结合，分析不同民族和种族对住房广告的反应是如何不同的。该项目将为社区和单元选择过程、住宅分类、地方声誉的形成以及潜在租户如何形成对其住宅环境的印象等社会学理论做出贡献。该项目还将通过展示大数据源和计算技术的效用，帮助改变调查实验在社会科学领域的开发和实施方式。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

There is considerable urban residential segregation in the United States by race, ethnicity, and social class, with implications for community formation and inequality, as well as school segregation. We know that the majority of urban home-seekers in the United States now use the internet as their primary source to find new places to live. Despite this development, we know relatively little regarding how renters select their housing at a time when information about available housing is increasingly moving online. While this technological transformation has largely gone unexamined, some prior research shows that advertisements for rental housing are not all the same; rather, they differ systematically depending on the demographics of the neighborhood where the housing being advertised is located. This project analyzes rental housing advertisements posted online to investigate if these differences matter for people during their housing search. Understanding how individuals interpret the information they see in the online housing market is key to explaining why people move to certain places and not others, which has implications for the future of residential inequality and racial/ethnic segregation. The findings will advance understanding of residential selection processes and aid policy makers looking to expand and equalize access to information for home-seekers, with implications for improved social and economic well-being in urban areas. Given the role that online rental advertisement plays in promoting neighborhood composition, it is notable that we know so little regarding how individuals interpret this information. Using natural language processing to analyze millions of advertisements for rental housing in the 50 largest U.S. cities posted on Craigslist, prior work has identified patterns in the distribution of different types of information in neighborhoods that vary by race/ethnicity and poverty rate. This proposal will use this information to test the causal effects of these real-world differences in the ways units are advertised on individuals’ housing and neighborhood preferences in five large urban areas: Los Angeles, SF-Bay Area, New York City, Chicago, and Houston. The project will implement three survey experiments in each area to test how online advertisements shape housing decisions and residents’ perceptions of local neighborhoods. The project will compare the effects of information in housing ads and perceptions of neighborhoods to the effects of other kinds of information, including neighborhood demographic data, on individuals’ interest in housing units. Selection of these five large areas allows the project to oversample Black, Latino and Asian minority respondents. Each survey experiment will first be constructed to be representative of the urban area (n=1,000/area), but will then collect additional respondents from specific minority population(s) within each area: oversamples of 100-300 additional Asian respondents in the SF Bay Area; Black respondents in New York and Chicago; and Latino respondents (including Spanish-speakers) in Los Angeles, Chicago, and Houston. By using representative surveys of specific urban areas with oversamples of minority residents, the project will analyze—through a combination of difference of means tests and multiple regression models—how reactions to housing advertisements vary across ethno-racial groups. The project will contribute to sociological theory regarding neighborhood and unit selection processes, residential sorting, the formation of place reputations, and how prospective tenants form impressions of their residential contexts. The project also will help to transform how survey experiments are developed and implemented across the social sciences by demonstrating the utility of big data sources and computational techniques.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.