Understanding Public Perceptions of Affordable Housing in the United States through Quantitative and Qualitative Data

Graduate Researcher: Isabella Douglas

Faculty Advisor: Sarah Billington

For the past two decades, the cost of housing in the United States has increased relative to income, making housing affordability precarious for the average household. This housing precarity has been exacerbated by an ongoing shortage of housing, as well as challenges to affordable housing acceptance. The current health and financial crises brought on by COVID-19 have highlighted the importance of housing and in particular affordable housing, for personal and community wellbeing. While awareness of affordable housing need is definitive among the general public, it is less clear how uniform or variable the public’s understanding of what counts as affordable housing is. This variability in public understanding can impact actions taken for or against the support of affordable housing. We deployed a nation-wide online survey (n = 540) consisting of closed- and open-ended questions that focused on public perceptions and personal exposure to affordable housing projects in addition to acceptance. This survey was designed to answer the main research question: What is the public’s baseline understanding of affordable housing? And specifically, what is the public's mental imagery involving people, policies, and buildings associated with affordable housing (Figure 1)? Our analysis approach combines statistical analysis of the quantitative survey questions with hand coding and natural language processing (NLP), including closed-vocab analysis, open-vocab analysis, and topic modeling, of the qualitative free response answers. By triangulating among these three analysis approaches, a more nuanced understanding can be obtained. For example, while the closed-ended responses show an awareness of a broad range of affordable housing building typologies, in participants’ free-responses, the typologies of single-family homes and apartments dominate. Participants’ free-responses highlight that their mental images of affordable housing buildings have not yet caught up with current-day reality (i.e., participants often think of stark, modernist "projects" rather than examples of more recent affordable housing projects). Furthermore, there seems to be a dichotomy between how participants describe their current existing picture of affordable housing (mainly negative), and how they describe their ideal future picture of affordable housing (mainly positive). To compare with these explicitly probed opinions in a survey-setting, we will scrape social media data to generate a large text dataset and run NLP analyses to capture unprompted opinions. In addition to furthering knowledge about public perceptions of affordable housing and how these relate to demographics, exposure, and self-reported acceptance, this research highlights the potential of semi- and unstructured text data and emerging methods like NLP to efficiently capture the opinions of the broader community often not present at planning meetings.