The Impacts of an Urban Cable Car System on Liveability: A Mixed Methods Study in Bogotá, Colombia

14.2.1

Study Setting

The TrUST study was conducted in the city of Bogotá, Colombia. The intervention area encompasses the Ciudad Bolívar neighbourhoods, and the control area includes the San Cristóbal neighbourhoods where another cable car is scheduled to be implemented in 2024. In 2018, Bogotá’s overall Multidimensional Poverty Index was 4.1; this value rose to 7.1 in 2020 (DANE, 2020) with some neighbourhoods experiencing more extreme conditions than others. The marginalised neighbourhoods of Ciudad Bolívar and San Cristóbal are characterized by precarious planning and dense populations illustrating the effects of accelerated urbanization driven in part by internal, conflict-related displacement. The past five decades has resulted in self-built neighbourhoods located in areas along the outskirts of the city built on extreme slopes land facing severe economic, employment, health, and mobility challenges. The intervention and control areas are difficult to reach from the centre of Bogotá. These areas were selected for this study because they have similar geospatial and sociocultural conditions and crime levels yet are separated by geographical barriers that should limit contamination (Sarmiento et al., 2020).

14.2.2

The TransMiCable Cable Car Project

TransMiCable is 3.43 kilometres long, includes four stations, and runs on clean energy. There are 163 cabins with a capacity of 10 passengers per cabin. During its first year of operation, approximately 7.5 million people used TransMiCable. Currently, approximately 21,000 people use TransMiCable on workdays and around 17,500 on Sundays and public holidays. TransMiCable is deemed to be a sustainable and healthy transportation system. TransMiCable cabins have better air quality within the cabins compared to any other public transportation systems in Bogotá (Morales-Betancourt et al., 2023). Additionally, TransMiCable fosters or maintains active travel within combined transport modal shares. More information about TransMiCable and the rest of Bogotá’s public transport system can be found online (Transmilenio, 2022). The TransMiCable cable car system is the main component of a wider intervention that also includes a library, a tourism office, a local history museum, a citizen service office, two trails, two sport and recreation centres, three local markets, three community centres, and 11 parks. The TransMiCable project also includes a program to support improvements to local homes and a project to reduce landslides and other local environmental hazards (Sarmiento et al., 2020).

14.2.3

Study Design

The TrUST study is a natural experiment using mixed methods with a simultaneous, bidirectional integration approach (Moseholm & Fetters, 2017) including dimensions of the urban liveability framework (Badland et al., 2014). To assess the effect of the urban transformation in Ciudad Bolívar, we used an array of methodologies to capture the interplay of changes in different liveability domains. The quantitative component focused on measuring the impact of the intervention on liveability, while the qualitative component was intended to document the residents’ experience of the urban transformation.

The study design is described in detail elsewhere (Sarmiento et al., 2020). Overall, the intervention area in Ciudad Bolívar included households located within an 800-metre radius (or buffer) of each of the four TransMiCable stations. The area of influence in the control neighbourhood of San Cristóbal included households located within an 800-metre buffer of the potential locations of the planned stations. An 800-metre buffer is greater than the walking buffer normally used in transport studies, based on the hypothesis that people with lower incomes would be willing to walk long distances to take advantage of public transport (Sarmiento et al., 2020). Participants were recruited from 225 blocks within the intervention area and 228 blocks within the control area. Blocks were selected with a probability proportional to the density of parcels. Every third household was systematically selected and one adult that fulfilled the inclusion criteria was selected per household (Sarmiento et al., 2020). For the quantitative study, we included cohort data. ‘Citizen science’ data was obtained from convenience samples in the intervention and control areas, while REM data was obtained from a convenience sample from the intervention area. For the qualitative study, the overall sample differed in baseline and follow-up periods. Figure 14.1 shows the timeline of quantitative and qualitative data collection; Table 14.1 describes the various data sources.

Photos and sampling of buffer areas from the intervention (a) and control (b) areas. Purple dots correspond to the areas where pictures were taken by citizen scientists. Figures of the TrUST study 2017–2021

Full size image

Full size table

14.2.4

Quantitative Component

14.2.4.1

Household Survey and Built Environment Characteristics

Trained interviewers surveyed participants in their households using a structured questionnaire. Baseline surveys were conducted from January 2018 to December 2018 and included 1031 adults in the intervention area and 1021 in the control area. Follow-up surveys were conducted after the inauguration of TransMiCable from July 2019 to March 2020 and included 825 adults in the intervention area and 854 adults in the control area (Table 14.2).

Full size table

14.2.4.1.1

Sociodemographic and Transport Characteristics

Sociodemographic characteristics included sex, age, educational attainment, occupation, and monthly household income. Transport accessibility included mode choice, travel time, and distance to the nearest bus-rapid transit station based on the shortest path from the household to the BRT station through the street network.

14.2.4.1.2

Built Environment Characteristics

Built environment characteristics included slope, intersection density (i.e., crossroads density), and parks density, measured using secondary official data (IDECA & Gobierno de Colombia, 2022). We created a 500-m street network buffer around each georeferenced household, computing slope using the triangulated irregular network that represents terrain surfaces irregularly distributed to accommodate areas of high variability in the surface every five meters. Intersection density was calculated as the number of intersections per square kilometre. Park density was calculated as the area (in square metres) of parks within each buffer area. All analyses were conducted using ArcGIS® software (Esri Inc, n.d.).

14.2.4.1.3

Liveability Outcomes

Liveability outcome indicators included measures for transport, neighbourhood public open space, social cohesion, local democracy, and security domains. Satisfaction with public transport was measured on a scale of 1–10 (where 1 is extremely unsatisfied and 10 is extremely satisfied). Potential barriers to better liveability in the neighbourhood included issues surrounding the obstruction or lack of parks and sports centres, disruption of sidewalks and streets, improper disposal of waste, the presence of rodents and insects, bad odours, and air pollution. Social cohesion and local democracy indicators included participation in community organizations and trust in public institutions. Security indicators included perceived security and reports of being a victim of robbery in the last 12 months.

14.2.4.2

Quantitative Data Analysis

First, we described the sociodemographic, transport, and built environment characteristics of the intervention and control groups. Next, we investigated the impact of TransMiCable on liveability outcome indicators by conducting multilevel regression models (linear and non-linear) with random intercepts for individuals. Models included main effects of time (T0/T1) and urban area (intervention/control), as well as a time by urban area interaction. The time by area interaction term was used to assess the effect of the intervention. The models were adjusted for sociodemographic characteristics (age, sex, occupation, marital status, and education), distance to the BRT station, and slope.

14.2.5

Qualitative Component

The qualitative component of our evaluation was intended to capture in-depth insights regarding the implementation of TransMiCable on liveability and the ongoing transformation of Ciudad Bolívar (Sarmiento et al., 2020). Our participatory approach acknowledged the longstanding relationship between community leaders, policymakers, and other stakeholders (Sarmiento et al., 2020) who since 2007 have been mobilizing and advocating for a cable car system in this area.

14.2.5.1

The Our Voice Citizen Science Method

The Our Voice citizen science method involved four stages, each implemented at baseline and again at follow-up (King et al., 2019). In the first stage, residents walked around their community and used the Our Voice mobile phone app to capture what they deemed to be relevant information (24 citizen scientists participated at baseline and 30 at follow-up) (Fig. 14.2). In the second stage, we facilitated community meetings to discuss residents’ findings and to establish local priorities (7 citizen scientists at baseline and 16 at follow-up). In the third stage, we held discussions between residents, policymakers, and other stakeholders to facilitate their engagement in the definition of potential actions (discussions included 16 volunteers from the intervention group and 15 from the control group). Finally, local changes were implemented and evaluated (two volunteers took part in comprehensive interviews and three expert panels were convened). Participating policymakers and stakeholders included representatives of the Secretariats for Mobility, for Women and for Urban Planning, the District Institute of Recreation and Sports, the Ministry of Health TransMilenio, and TransMiCable.

Mixed methods approach: timeline of quantitative and qualitative data collection for the TrUST study 2017–2021

Full size image

14.2.5.1.1

Analysis of the Our Voice Citizen Science Data

Five researchers with backgrounds in social sciences and public health analysed citizen scientist data from the mobile phone app. First, citizen scientists presented their data during the facilitated community meetings. Second, researchers reviewed notes from community meetings and prepared a list of potential themes to consider across the baseline and follow-up periods. Third, all entries were coded in Excel matrices according to consolidated themes. Fourth, entries were analysed using a content analysis approach to characterise themes (Elo & Kyngäs, 2008). Fifth, following a grounded theory approach, themes were ranked by frequency (Carlin & Kim, 2017). Finally, all themes were discussed and consolidated during five separate meetings with research team specialists as part of the integration strategy described in Sect. 14.2.6. The research team included specialists in transport, air quality, crime, quality of life, and physical activity.

14.2.5.2

Ripple Effects Mapping Methodology

Ripple Effects Mapping was applied to further capture and map the diverse ‘ripples’ of the TransMiCable intervention (Chazdon et al., 2017). We conducted a Ripple Effects Mapping session in February 2020 with nine citizen scientists, including community leaders, who had already taken part in the citizen science data collection in the intervention area. First, participants were divided into groups of two and interviewed each other using a questionnaire reflecting the interview component of Ripple Effects Mapping. Next, each participant reported the insights expressed during their interviews to the larger group. One researcher asked questions to deepen the reports, and a recorder took note of each account using the XMind mind mapping tool (XMind, n.d.). Participants collectively organised their insights by themes, giving form to the mind map. The session was recorded and transcribed verbatim. Following the session, two researchers independently streamlined the map in XMind, regrouping the Ripple Effects Mapping themes using the Our Voice themes using a grounded theory and content analysis approach. The map was exported to Excel to obtain a final count of frequencies for each subtheme.

14.2.6

Integration of Quantitative and Qualitative Data

We used a simultaneous, bidirectional approach to integrate the various quantitative and qualitative data collected regarding effects per each assessed liveability domain (Moseholm & Fetters, 2017). Thus, quantitative and qualitative data were analysed separately and then merged for interpretation according to the liveability framework (Badland et al., 2014). First, researchers organized results in Excel, with liveability domains providing the framework for analysis. Second, through the comparison of liveability outcomes in Excel, we assessed the convergence, divergence, and complementarity of quantitative and qualitative data. Third, results were discussed and consolidated by the research team to drive final interpretation.