Chapter 6 Discussion

6.1 Research significance

6.1.1 Global development goals

This study contributes to the wider agenda of sustainable development by addressing how active travel can be systematically evaluated and promoted. Cycling is explicitly connected to several United Nations Sustainable Development Goals (SDGs), including SDG 3 on good health and well-being, SDG 11 on sustainable cities and communities, and SDG 13 on climate action. By operationalising a composite measure of cycling quality, the research provides a framework for monitoring progress toward these goals. The findings highlight the need to improve accessibility, reduce exposure to environmental hazards, and ensure equity in mobility provision—key principles that resonate with international objectives for low-carbon, inclusive urban futures.

6.1.2 Local policy

At the local level, the dissertation provides evidence directly relevant to London’s ongoing transport and environmental strategies. The identification of low-quality segments in the Central Activity Zone suggests that current investments in cycling infrastructure have yet to resolve issues of continuity and environmental stress in the city’s busiest areas. The mapping of high-quality but disconnected corridors in outer boroughs also underscores the importance of linking peripheral areas to central employment hubs. These insights can inform initiatives such as the Mayor’s Transport Strategy, the Healthy Streets Approach, and borough-level cycling action plans. By integrating infrastructural, environmental, and network perspectives, the CECI framework offers a tool for prioritising interventions where they can deliver the greatest impact.

6.1.3 Academic research

Academically, this study advances the growing field of urban cycling assessment by bridging methodological divides between infrastructure analysis, environmental perception, and spatial network science. While previous work has often focused on one dimension at a time, the Cycling Environment Composite Index (CECI) demonstrates the value of synthesising diverse indicators into a single evaluative framework. The approach responds to calls in the literature for multi-scalar, holistic assessments of urban mobility systems and contributes empirical evidence on London, a city that is often studied but rarely through an integrated, composite perspective. Beyond cycling research, the framework also illustrates how geospatial data, environmental metrics, and graph-theoretic measures can be combined in the analysis of urban accessibility.

6.2 Limitations

The estimation of edge betweenness via stochastic source–target sampling imposes an inherent constraint on strict reproducibility. The sample size KK is chosen to balance run-time and stability, yet the procedure remains probabilistic; small numerical deviations and local rank shifts can arise across runs, and may be accentuated by differences in software versions, parallel execution order, or hardware. Under these conditions, the index is most appropriately interpreted through spatial patterns and relative ordering rather than exact equality of segment-level values across replications.

Temporal alignment across inputs constitutes a further limitation. The component datasets are not fully contemporaneous—street-level greenery is inferred from imagery circa 2015, modelled NO\(_2\) fields originate from LAEI 2016 projections, and the OSM network is continuously updated. This asynchrony introduces the possibility of local timing mismatches, particularly where recent greening initiatives, traffic-calming schemes, or redesigns have occurred. As a result, fine-grained differences should not be read as current, time-specific conditions in all locations but rather as a structural and perceptual baseline.

Limitations also arise from cross-city parameter transfer within the structural rideability module. Thresholds and mappings for elements such as effective width, separation, and buffer rules derive from a framework calibrated in Berlin. Although the metadata in both contexts are sourced from OSM via Overpass Turbo, country-specific tagging practices, attribute completeness, and data conventions differ. During transfer, this heterogeneity can increase the frequency of missing or partially specified attributes. To avoid optimistic bias where attributes are absent, a conservative penalty mechanism is adopted that applies uniform downward adjustments under defined missingness conditions; this improves internal consistency but may produce a systematic downward shift in areas with sparser tagging.

The dependence on OSM further entails issues of coverage and tagging consistency. Volunteer-contributed data are uneven across space and over feature types, and fine-grained fields of particular relevance to cycling—cycleway subtype, buffer presence and width, surface and smoothness—may be omitted or inconsistently recorded. Such variation can propagate to factor assignment and to the stability of LTS classification, especially in neighbourhoods where infrastructure is evolving faster than tagging activity.

Design choices embedded in the network analysis also shape results. Centrality is computed on a primal, undirected, length-weighted graph, which abstracts from signal delay, turn costs, grade at junctions, and micro-routing preferences; the selected closeness radii and the approximate betweenness sampler represent defensible but not unique specifications. Similarly, distributional treatment through p1–p99 winsorisation and min–max scaling reduces the influence of extremes but compresses tail variation, with potential consequences for the relative prominence of isolated high- or low-scoring segments.

Besides, validation provides limited temporal coverage, while autumn counts only cover Cycleways. The results should therefore be interpreted as indicative rather than comprehensive.

In addition, the composite weighting schemes encode normative assumptions regarding the relative salience of structural quality, environmental comfort, and network position. Alternative analytical or policy perspectives might reasonably prioritise these dimensions differently, yielding variation in spatial rankings. Taken together, these considerations indicate that the index should be interpreted as a comparative, spatial diagnostic rather than as a definitive, time-specific measure of present conditions.

6.3 Transferability

A further significance of this research lies in its transferability to other urban contexts. The CECI framework is built on widely available datasets and reproducible geospatial methods, making it adaptable to cities beyond London. While local calibration would be necessary to reflect different infrastructural standards, environmental conditions, or policy priorities, the general approach of integrating structural rideability, environmental perception, and network centrality is applicable to diverse settings. This opens avenues for cross-city comparisons and benchmarking, enabling researchers and policymakers to assess cycling environments in a consistent way. By doing so, the framework supports broader efforts to mainstream active travel within sustainable mobility agendas globally.