Fri 2 Feb 2007
A walkable neighborhood isn’t walkable unless it has a well-connected thoroughfare network. A well-connected network, composed of direct, convenient routes, is one of the key ingredients of walkability. Well-connected neighborhoods have a host of advantages for residents and for the greater community.
A large and growing collection of research is finding that street connectivity is associated with more walking, less driving, greater safety, less crime, better physical fitness, and fewer per capita emissions. This post reviews the research on neighborhood-scale relationships between connectivity and walking.
Connectivity on the neighborhood scale is about connectivity within neighborhoods. It’s about the routes and connections from building to building, from lot to lot, and from block to block. For more about the key ingredients of walkability, see the frequently asked questions page.
A Few Words About Research
By the 1980s, researchers were seeing urban form and its relationship to social, environmental and other outcomes as worthwhile topics of investigation. The research has been both “popular” — i.e., publications and methods using more subjective judgments and intuitive observations — and “formal,” using the format of peer-reviewed statistical analysis. Both types can be rigorous and authoritative.
Connectivity research begins with basic definitions. There are a variety of metrics that can characterize connectivity, such as density of intersections, block size, number of 3- and 4-way intersections and cul-de-sacs, and route directness. Measuring Network Connectivity for Bicycling and Walking by Jennifer Dill (2004) provides a good review of techniques for measuring connectivity.
It’s important to recognize several caveats about connectivity analyses. Personal socioeconomic factors like income, age, ethnicity, gender, and family status usually influence travel behavior more than urban form. Non-design factors like hills, weather, local crime and maintenance play a role. Urban form itself includes a variety of elements such as proximity to mixed use, density, width of thoroughfares, and presence of sidewalks. Well-designed studies take into account some combination of these factors.
Finally, academic research can make your eyeballs glaze over. But in our technocratic society, for better or worse, policy is often based on research. So it’s useful to be a little familiar with the broad trends.
Connectivity and Walking
Over the past 15 years, an extensive body of literature has investigated urban design and its effect on walking. Well-connected streets are associated with more walking, and the magnitude and strength of the association varies with study methodology and the data that are used.
Great Streets (1993) by Allan B. Jacobs was an outstanding example of popular-style research. The book delved into the nitty-gritty details of pedestrian-oriented streets around the world, showing a wealth of plans and measurements useful to designers. Great Streets analyzed street networks around the world using a standardized scale, area and connectivity metrics. The most walkable districts — such as London, Rome, Savannah, Seoul and Tokyo — had intersections per sq mi ranging from 400 to 1,000. Car-free Venice was an extreme at 1,725 intersections per sq mi. At the auto-oriented end of the scale, places like Walnut Creek, San Fernando Valley and Irvine in California, and Brasilia in Brazil, had intersections per sq mi ranging from 15 to 120.
One of the first large-scale, comprehensive studies of pedestrian design and travel choices was the LUTRAQ report, launched in 1990 and completed in 1997. The Pedestrian Environment chapter developed the Pedestrian Environmental Factor (PEF) which included ease of street crossings, sidewalk continuity, connectivity and topography. Although LUTRAQ was not primarily focused on physical activity, its models projected that residents of well-connected, transit oriented developments were twice as likely to walk to work, and children were twice as likely to walk or bike to school.
LUTRAQ was based on Portland, OR, travel surveys and was intended as a model for further studies. Accordingly, the SMARTRAQ study in Atlanta and the LUTAQH study in King County, WA, emulated and refined the methodology. Both were large, multi-year studies with unusually comprehensive scope, and both were headed by Lawrence Frank.
The 5-year SMARTRAQ study used travel surveys from 8,000 households in Atlanta; the methodology used a combination of urban form measures (net residential density, land use mix, and connectivity) and controlled for sociodemographic characteristics. For the walking portion of the study, participants wore a device like a pedometer that objectively measured physical activity. Compared to the least walkable neighborhoods, people in the most walkable neighborhoods were more than twice as likely to do thirty minutes of moderate daily exercise.
As noted above, results showed that residents of the most walkable areas of the Atlanta Region are 2.4 times more likely to get the level of activity necessary to maintain health after controlling for demographic factors (Frank et al 2005).
– SMARTRAQ Summary Report (p. 23)
(By the way, the SMARTRAQ final summary report, with lead authoring duties performed by David Goldberg, was released on Jan. 19, 2007. It’s quite difficult to write in an engaging way about formal statistical research, and David and his crew have done a terrific job. The report is full of all sorts of vital information about urban form and its relationship to health, walking, driving, pollution, market preferences, etc. The results are especially interesting because they’re from Atlanta, one of America’s most sprawling cities.)
The 5-year LUTAQH study analyzed urban form and its relationship to walking, transit use, pollution and obesity. The study found that street connectivity, when considered in isolation, had a significant but weak correlation with total walking trips. When controlling for demographics, each quartile increase in intersection density corresponded with a 14 per cent increase in the odds of walking for non-work travel. LUTAQH also found that street connectivity was moderately but significantly correlated with recreational walking. The authors speculated that the greater psychological variety inherent in connected street patterns was the reason for this result:
In a gridiron street network with short block lengths, walkers have numerous opportunities to vary their route, to investigate interesting activities or features glimpsed up side streets, and to shorten or lengthen their walk without retracing their steps along the same roads. In contrast, in hierarchical street networks with curvilinear streets and cul-de-sacs, walkers have fewer route options, opportunities to change direction are some distance apart (and often out of sight around curves), and varying the length of a walk often means simply turning around and walking back along the same route. (p. 150)
The LUTAQH study concluded that connectivity works in concert with mixed use to increase walking:
Likewise, more intersections — greater street connectivity — do not automatically translate into more places to walk to. The results of this research show that intersection density and increased mix of uses play complementary roles in encouraging people to choose alternatives to the car. (p. 154)
Researchers have also performed studies with a narrower scope, looking specifically at connectivity and its association with walking. Greenwald and Boarnet (2001) used the Pedestrian Environmental Factor and found it was significant in determining the probability of non-work walking travel at the neighborhood level. Rajamani et al (2003) analyzed travel diaries and found that a street design with few cul-de-sacs and a grid-like geometry had the potential to encourage walking. The authors wrote that additional design factors must be analyzed in conjunction with street geometry in order to gain a comprehensive view.
Using a San Francisco-area travel survey, Cervero and Duncan (2003) found that well-connected streets, small city blocks, mixed land uses, and close proximity to retail activities induced nonmotorized transport. But the effect was modest, and other factors, such as topography, weather and demographics were far stronger predictors of walking and bicycling choice than built-environment factors. The authors emphasized the need for stronger evidence. In a later study, Cervero (2007) explored the factors that influenced the choice to commute by transit:
… the only neighborhood-design variable that provided significant marginal explanatory power was the level of street connectivity at the destination (among several dozen variables representing densities, land-use mixes, and design features that were available for model entry). When exiting a station en route to work, having a walkable grid-street pattern with high connectivity matters to station-area residents when deciding whether to commute via transit.
– Transit Oriented Development’s Ridership Bonus (p. 8)
Schlossberg et al (2006) studied children’s trips to school in two Oregon cities, using surveys from 292 families. In neighborhoods with high intersection density, walking to school was three to five times more likely than in neighborhoods with low intersection density (10% likelihood vs. 2-3% likelihood). Using the density of dead ends as a measure yielded similar results. Distance to school and urban form together were highly associated with whether middle school children walked to and from school, predicting 32-41% of the variability in survey results.
A consistent theme running through the studies is that connectivity works synergistically with other urban design elements to support walkability. When tested by itself, connectivity may show little relationship with walking. When tested in concert with other design elements like proximity to mixed use, connectivity has a significant impact on walking. The LUTAQH study put it well:
Still, it is not the presence of individual characteristics, but the collective effect of land use mixing and density that influences walking. By simultaneously increasing a number of the measures, including density, the square feet of retail space, the number of restaurants and parks, and street connectivity, planners may be able to bring about dramatic transformations in the number of household trips accomplished by walking. (p. 110)
Another theme in the studies is that urban form and self-selection both play a role in walking behavior. In other words, to some degree people walk more in walkable neighborhoods because they already prefer to walk, and choose neighborhoods where they can do so.
A few commentators have advised that the presence of self-selection means that government officials should be wary of pursuing pro-walkability policies. But it’s not clear why that should be so. If a sizable portion of the population prefers a more pedestrian oriented lifestyle, then there should be an adequate stock of housing available in such neighborhoods to meet the demand. In addition to freedom of choice, planning policy should consider the many public benefits of walking — more energy efficiency, less pollution, better health, less traffic, etc.
This argument would follow that individuals who are predisposed to walk, choose to live in environments with more destinations accessible on foot. … If this is the case, then the variety of land uses in these neighborhoods is only in part motivating people to walk more; that is, people who like to walk are moving to (or staying in) these neighborhoods, if they are available and affordable. The latter part of the last sentence is the key; emerging research suggests that many residents of auto oriented environments would prefer an environment with walkable destinations, but have traded it off because it is undersupplied, and this undersupply has driven the cost up to the point where locating there is economically illogical, or even infeasible …
– LUTAQH (p. 96)
Some policymakers and researchers are interested in resolving the self-selection question. If urban form causes greater physical activity, then public health institutions would have a professional and perhaps regulatory interest in city planning. According to Susan Handy in Critical Assessment (2005), two avenues of investigation may prove most revealing. One is longitudinal studies — that is, studies that track people over time as they move from one neighborhood to another, paying attention to how their travel choices change. The other is intervention studies; an example would be a before-and-after study of a neighborhood where sidewalks were installed, and the resulting effect on walking.
Handy’s literature review on the built environment, physical activity and self-selection was commissioned by the National Academy of Sciences for its book Does the Built Environment Influence Physical Activity? Handy found that the literature so far leaves many unanswered questions, and that causal relationships cannot be conclusively determined. However, the evidence shows a “distinct possibility” that the built environment causes some measure of physical activity.
The LUTAQH study furnishes a suitable final word:
While the self-selection argument poses an interesting logical puzzle for planners, the reality is that when walkable places are created, people living in them walk. If the self-selection argument is true, then the supply of walkable neighborhoods has not yet met demand, as the higher property values for walkable neighborhoods, such as Capital Hill and Queen Anne Hill in Seattle, demonstrate. If the self-selection argument is wrong — if in fact, building walkable environments does encourage people to walk more — then planners can and should facilitate the creation of walkable environments whenever possible.
– LUTAQH (p. 113)
Active Living Research has assembled an extensive bibliography of peer reviewed statistical research on the built environment and physical activity. The sections that are most relevant to connectivity are Measuring the Environment and Health and Environment. Active Living summarized the bibliography in its pamphlet Designing for Active Transportation.
Another bibliography of peer reviewed statistical research, this one annotated, has been assembled by the Sightline Institute in its Sprawl and Health database.
FHWA Course on Bicycle and Pedestrian Transportation. Lesson 1: The Need for Bicycle and Pedestrian Mobility provides examples of Allan Jacobs’ street network diagrams.
Cervero, Robert and Michael Duncan, Walking, Bicycling, and Urban Landscapes: Evidence From the San Francisco Bay Area. American Journal of Public Health, 2003 September; 93(9), pp. 1478-1483.
Cervero, Robert, Transit Oriented Development in America: Contemporary Practices, Impacts, and Policy Directions. International Planning Symposium on Incentives, Regulations, and Plans — The Role of States and Nation-States in Smart Growth Planning. National Center for Smart Growth Research and Education, University of Maryland and Habiforum Foundation, The Netherlands. September 30-October 1, 2004.
Cervero, Robert, Transit-oriented development’s ridership bonus: a product of self-selection and public policies. Environment and Planning A 39(9), pp. 2068-2085, 2007.
Dill, Jennifer, Measuring Network Connectivity for Bicycling and Walking, TRB 2004 Annual Meeting CD-ROM, 2004.
Greenwald, Michael J. and Marlon G. Boarnet, The Built Environment as a Determinant of Walking Behavior: Analyzing Non-Work Pedestrian Travel in Portland, Oregon. Institute of Transportation Studies, University of California, Irvine, July 2001.
Handy, Susan, Critical Assessment of the Literature on the Relationships Among Transportation, Land Use, and Physical Activity, Resource paper for Does the Built Environment Influence Physical Activity? Examining the Evidence, Special Report 282, Transportation Research Board and Institute of Medicine Committee on Physical Activity, Health, Transportation, and Land Use, 2005.
LUTAQH: A Study of Land Use, Transportation, Air Quality and Health in King County, WA. Prepared by Lawrence Frank & Co., Inc. et al for King County, WA, December 2005.
Making the Land Use, Transportation, Air Quality Connection (LUTRAQ), Technical Reports Vols. 1-8, prepared by Cambridge Systematics, Calthorpe Associates, Parsons Brinckerhoff Quade & Douglas and ECONorthwest for 1000 Friends of Oregon, 1991-1997.
National Academy of Sciences, Does the Built Environment Influence Physical Activity? Examining the Evidence, Special Report 282 . Transportation Research Board and Institute of Medicine Committee on Physical Activity, Health, Transportation, and Land Use, 2005.
Rajamani, Jayanthi, et al. Assessing the impact of urban form measures in nonwork trip mode choice after controlling for demographic and level-of-service effects. Transportation Research Board 2003 Annual Meeting.
Schlossberg, Marc, et al, “School Trips: Effects of Urban Form and Distance on Travel Mode.” Journal of the American Planning Association, Vol. 72, No, 3, Summer 2006.
SMARTRAQ Summary Report (January 2007). New Data for a New Era: A Summary of the SMARTRAQ Findings Linking Land Use, Transportation, Air Quality and Health in the Atlanta Region by David Goldberg, et al.
SMARTRAQ Final Report (April, 2004). Integrating travel behavior and urban form data to address transportation and air quality problems in Atlanta, by Jim Chapman and Lawrence Frank. Georgia Regional Transportation Authority and Georgia Department of Transportation.