In celebration of APA’s Planning Month, Project Manager and Senior Technical Advisor Alex Bell reflects on the Accessibility-Based Travel Model & West Station Area Pilot Study project.
When asked what I do for a living, I often fumble for words. I try to explain that I’m an urban planner who knows little about conventional planning (thankfully my colleagues know a ton about it!). But I’m really an urban analyst striving to use quantitative insights – generated through data analysis and custom tools – to deliver a sense of clarity for my peers, partners, and clients.
My contribution to planning arises from the development of novel analytical frameworks and the demonstration of fresh metrics to offer richer insight into the performance of and interactions among the complex systems that are frequently the subjects of plans and studies. After all, if we want to create and live in cities that work, we need to understand how they work.
In my role as an analyst, I often generate materials that support project stories, but my work isn’t really the story itself. In fact, the analytical process is often deemed too technical or inaccessible to belong within a larger project narrative. When projects are almost exclusively technical in nature, there’s a temptation to overemphasize innovation, novelty, and discovery. These tendencies obscure the art and craft of the analytical process – and the expertise and human motivations of the analysts designing and executing it.
The Accessibility-Based Travel Model & West Station Area Pilot Study Project
The Accessibility-Based Travel Model & West Station Area Pilot Study project illustrates Renaissance’s evolving analysis frameworks that reflect shifting priorities within the planning profession. This evolution always reflects our best efforts to get to grips analytically with the trending issues that resonate with planners and the communities they serve. Our introduced innovations need not impose themselves as a new way of doing things, but they can reframe prevailing thinking on key topics to better address changing questions and concerns.
In 2019, the Metropolitan Area Planning Council (MAPC) of Massachusetts set out to develop a sketch travel planning model based on multimodal accessibility, an emerging analytical lens that combines land use and transportation factors to describe the ease with which travelers can reach destinations by different travel modes. Renaissance was retained to lead the study, and I took the reins as Project Manager. The goal was to offer planners an alternative to the regional travel demand model for local area analyses that would be sensitive to fine-grained demographic, land use, urban design, and transportation network and policy details.
It’s worth pausing to remind ourselves how difficult it is to understand how these aspects of place influence travel behaviors. Intuitively, we reckon they affect trip generation, mode choice, average trip length, etc. But tying them together in a comprehensive behavioral model? Let’s just acknowledge that no one on earth has perfected that yet.
Of course, perfection isn’t demanded. Sound, reasonable insight that can inform policy decisions is deeply valuable even when imperfect, especially in complex settings where numerous interests are represented. Our sketch travel model was to be pilot tested in the West Station planning area. Sticking strictly to the technical aspects of model development, Renaissance’s intention was to demonstrate sensitivity to alternative land use scenarios, walk and bike network configurations, transit service schedules, and pricing considerations in travel behavior forecasts. This would give planners critical insights to guide discussions with policymakers and stakeholders.
Our approach built on some of our prior experience using multimodal accessibility measurement and comparing the ease with which travelers could reach key destinations by walking, biking, driving, and transit. Accessibility offers several attractive features for this kind of sketch modeling application.
It’s pretty intuitive: the access offered by each mode relative to competing modes has a significant influence on whether travelers opt to use it.
It can incorporate a lot of other variables: access scores generally reflect proximity to destinations and mobility provided over travel networks. In this way, it combines land use and transportation factors in a coherent framework. It can be sensitive to travel time, parking charges, transit fares, tolls, vehicle operating costs, network connectivity, non-motorized facility characteristics, land use mix, and more.
We can flex geographies: when we want to understand how highly localized changes like walking directness or land use mix influence travel patterns, we need a granular analysis. Since it can be measured at various scales, accessibility offers the opportunity to achieve that detailed granularity for a local study area, all the while retaining an awareness of regional travel patterns.
The modeling approach and results are summarized here. We developed rich accessibility scores for each major mode of travel that accounted for its unique operating characteristics and costs, providing granular detail in the West Station area while maintaining a regional perspective. The access scores were used to develop behavioral models that also account for traveler socio-economic and demographic characteristics to forecast trip generation, mode choice, and distribution patterns. We also created a post-processing step to estimate where trips by transportation network companies (TNC’s) like Uber and Lyft would be made.
Finally, we constructed the model in a relatively lightweight package for sharing and reuse. We programmed the entire modeling framework using Python and R, well-known data analysis languages. We delivered the full toolkit and workflow for MAPC staff to use in developing and testing alternative scenarios for the West Station area and for use in similar future efforts.
What It Means
Renaissance collaborated with the Boston Region MPO’s Central Transportation Planning Staff (CTPS) to compare our accessibility model’s results against those generated by the regional travel demand model. We found inevitable discrepancies between the two models, particularly within the probability of travelers using non-auto modes in the West Station area. To align the accessibility model results with the regional model, Renaissance introduced some guardrail factors – these were not draconian but helped ensure that discrepancies between the platforms didn’t introduce murkiness where clarity was sought.
The MAPC project demonstrated the effectiveness of multimodal accessibility analysis as a basis for providing insight into travel behavior. There are several reasons to be excited about that:
Accessibility is emerging as a common metric in transportation and land use planning applications. By demonstrating the connection to travel behavior, planners can glean at least rough insight into expected travel patterns from their accessibility maps . In turn, this insight can inform context-sensitive facility design, development review processes, and local connectivity investments.
One of the major emerging uses of accessibility is supporting equity analyses that describe who has what level of access to key destinations. Here again, the connection to travel behavior allows those analyses to move from simply describing what opportunities are reachable (which is still quite useful) to describing how transportation and land use factors interact with socioeconomic and demographic attributes – and ultimately influencing the utility offered by the complete transportation system.
Planners need ways to generate travel insights that don’t depend on travel demand models but offer a comparable level of rigor. That’s because the questions planners often want to answer are difficult to put to most regional travel models and because multimodal travel is increasingly important in transportation planning and infrastructure funding. As planning emphases evolve, it is difficult to adapt legacy travel models that are very useful for their primary intended purpose (highway planning) but often limited for other purposes. The variables to which planners seek sensitivity may not always translate into the regional model’s datasets or operative relationships. As demonstrated in the Boston area, an accessibility-based framework can offer more flexibility to expose those variables and define those relationships in robust ways.
This last point brings me back to the art of analysis – in this case, focusing on model development. It is common for analysts to presume that the best-fitting model is of paramount importance. In doing so, they can deem variables that planners and community stakeholders care about as “not statistically significant.” While we don’t want any noise in our models, we also have no real use for models that can’t respond to our human-centered questions.
For MAPC’s Accessibility Model, we tested innumerable permutations of variables and interactions, challenging ourselves to define logical steps of the process through which critical variables could be operationalized and ensure they behaved intuitively. Sometimes you can’t fit a square peg in a round hole, but we also do well to remember that our models and tools do not represent some quantitative tyranny. Rather, they are extensions of ourselves, imbued with our values, crafted by humans earnestly trying to make the world a better place.