Cities are traditionally viewed as melting pots, fostering social interaction and economic opportunities. However, rapid urbanization often exacerbates social segregation, particularly between income groups. This issue is further compounded by urban planning decisions. In many cities, including Hong Kong, urban development patterns lead to the spatial separation of different socioeconomic groups, with lower-income populations often relegated to less desirable locations. This spatial separation undermines the benefits of urban diversity and inclusivity, impacting access to resources and hindering social mobility. While past research focused on residential segregation, recent studies highlight the significance of "experienced segregation" within individuals' activity spaces. This study addresses this gap by analyzing social mixing in Hong Kong using longitudinal data across two decades (1992-2011), examining the impact of various transport modes and major infrastructure projects like metro extensions on experienced segregation. The unique aspect lies in the longitudinal perspective, tracing social mixing changes over time and connecting them to urban development processes.

The literature review focuses on two key aspects: measures of social segregation and the relationship between transportation and social segregation. Regarding segregation measures, recent research has expanded beyond residential segregation to encompass experienced segregation within individuals' activity spaces, using various indexes such as "experienced isolation" and "exposure segregation." These studies, often based on mobile phone data, highlight the persistence of social segregation beyond residential areas. However, these studies are often cross-sectional. The transportation literature reveals contradictory findings. While some studies suggest a positive association between public transit access and reduced social isolation, others highlight inherent inequalities in transport accessibility and the potential for gentrification driven by transit expansion. Existing studies lack a comprehensive longitudinal perspective on the role of different transport modes in shaping social mixing within a high-density urban environment.

This study uses data from three Hong Kong Travel Characteristics Surveys (TCS) from 1992, 2002, and 2011, encompassing detailed mobility and socio-demographic data for a large number of individuals. Participants were categorized into five income groups. The study employs a people-oriented social mixing index (DMᵢ) to measure the evenness of an individual's co-location with people from different income groups throughout their daily activities. A place-based social mixing index (DMₐ) quantifies the evenness of income group distribution at specific locations (Spatial Planning Units - SBs and Tertiary Planning Units - TPUs). Also, the nighttime social mixing (NMₐ) reflects residential segregation at home locations. To address the Modifiable Areal Unit Problem (MAUP), H3 hexagon grids of different resolutions were also employed. The study also incorporates data on MTR network expansion from 2002 to 2011 and Points of Interest (POIs) to control for place attributes. Regression models are used to analyze the impact of different transport modes (buses, MTR, walking, taxis, private cars) on social mixing at both the individual and place levels. A Difference-in-Differences (DID) model assesses the impact of MTR expansion on social mixing, comparing changes in areas near new stations with areas farther away. An instrumental variable approach is employed to analyze the effects of increased MTR usage on individual social mixing.

The study's key findings are: 1. **Persistent Inequality:** Social segregation persists across income groups, with lower-income individuals experiencing less social mixing. This inequality is evident both during the day and at night (residential segregation). Individuals from the lowest income group experienced 1.4% less social mixing than group 4 during the day in 2011. 2. **Positive Role of Public Transit:** Public transit, especially buses and MTR (metro), significantly increases social mixing at both place and individual levels. At the place level, a one percent increase in bus trips was associated with a 2.8% increase in social mixing. At the individual level, increased MTR and bus trips were linked to higher social mixing. 3. **Paradoxical Metro Expansion:** New metro stations are disproportionately located in already well-mixed areas, indicating a selection bias. While metro expansion increases overall trip diversity, it also contributes to a decrease in the low-income population in the immediate vicinity of new stations, suggesting gentrification effects. A Chi-squared test showed that the changes in income group distribution near new metro stations are not homogeneous. 4. **Long-Term Trends:** Daytime social mixing shows a slightly downward trend from 1992 to 2011, while nighttime mixing had a peak in 2002 and then declined. Lower-income groups showed higher social mixing in 1992 compared to 2002 and 2011. 5. **Other Factors:** The number of food and health-related POIs strongly contributes to higher levels of social mixing at the place level. Age, gender, and primary trip purpose (e.g., employment-based versus home-based) also significantly influence individual social mixing levels. Individuals with employment-based trips were more socially mixed over time.

The findings support the hypothesis that public transit can moderate social segregation, but the effect is complex. The positive impact of public transit aligns with the "melting pot" idea, yet the spatial bias in metro expansion highlights potential unintended consequences of urban planning that could exacerbate existing inequalities. The persistent disparities in social mixing across income groups underscore the need for policies that actively promote social inclusion beyond merely improving transport accessibility. The observed gentrification near new MTR stations suggests that urban planning initiatives must consider potential displacement effects and develop strategies for equitable development. Furthermore, the importance of place amenities (like food and health POIs) in promoting social mixing emphasizes the need for a holistic approach to urban design, which integrates mobility solutions with the creation of welcoming and diverse public spaces.

This study provides robust evidence of the role of public transit in moderating social segregation in a high-density urban environment like Hong Kong. The findings highlight the importance of utilizing longitudinal mobility data to understand the complex interplay between transport infrastructure development, urban planning, and social equity. Future research could focus on comparative studies across cities with varying levels of public transit access and socioeconomic structures. Exploring further the potential mitigation strategies for gentrification caused by transportation infrastructure development should also be examined.

This study has several limitations. The data from the 1992 TCS has a lower spatial resolution compared to 2002 and 2011 data. The study does not fully control for other simultaneous changes in the transport infrastructure. The analysis focuses on Hong Kong, a city with high public transport accessibility, and the findings might not be fully generalizable to cities with lower accessibility. The study has not controlled for the effect of COVID-19 pandemic which happened in-between the periods that may affect social interaction in various ways.