Document View

This study tests the New Urbanist claims that placing amenities such as parks and retail shops within walking distance of homes will increase pedestrian travel and thereby increase interaction among neighbors. It also examines the relative roles of physical design and personal attitudes and perceptions in predicting walking and neighboring behaviors. Surveys were conducted in eight neighborhoods (four inner-city, four suburban) with varying degrees of local access to parks and shops. Analyses were conducted at the neighborhood and individual levels and were supplemented with qualitative data. The findings provide some support for each of the tested relationships, but also underscore the significance of other variables, especially personal attitudes. [PUBLICATION ABSTRACT]

New Urbanism and other development strategies that fall under the general heading of "smart growth" make a number of claims about enhanced community life that are based on assumptions about how people will interact with and respond to their neighborhood environment. This article focuses on one of these claims-that placing amenities within walking distance of homes will increase pedestrian travel and social interaction among neighborhood residents. The study presented here tested the significance of local access (defined as no more than 1/4 mile from all housing units in the neighborhood) to parks and/or a neighborhood shopping area. This limit was used because of its acceptance in planning literature as a comfortable walking distance (Calthorpe, 1989). Three specific relationships were tested: local access and pedestrian travel, pedestrian travel and neighbor interaction, and local access and neighbor interaction. In addition, the article examines the relative roles of sociodemographic characteristics, personal attitudes, and perceptions of the local environment in predicting pedestrian travel and neighboring behaviors. Two types of pedestrian travel were considered: strolling trips (walking around the neighborhood) and destination trips (walking from home to a local business, park, or other activity site).

Before proceeding, however, it is important to note that to realize its goals of enhanced community life and reduced reliance on the automobile, New Urbanism calls for a combination of neighborhood design elements. As outlined in the charter of the Congress for the New Urbanism (1998), these elements consist of:

* compact, walkable neighborhoods with clearly defined edges;

* a clearly defined center with public space, public buildings, a transit stop, and retail businesses;

* an interconnected street network, forming coherent blocks and lined with building fronts rather than parking lots;

* a diverse mix of activities and housing options;

* civic spaces in prominent places; and

* open spaces in convenient locations throughout the neighborhoods.

In light of this comprehensive approach to neighborhood design, it would be unfair to study the potential of any one design element in isolation from the others. Therefore, this study looks only at neighborhoods that meet-to the extent possible-the first of these criteria: They are compact and walkable. This approach allowed the study to focus on the accessibility of everyday amenities without ignoring the importance of people-friendly neighborhood designs. (It is also possible, however, that the findings will underestimate the impact of New Urbanist designs since the research design holds many features of New Urbanism constant.)

Overview of the Literature

The extent to which physical environments affect, or have the potential to affect, individual behaviors has been the subject of an increasing number of studies over the past three decades. Most of this research has been centered in the field of environmental psychology, but also includes a number of seminal writings in the area of urban planning and design. Kenneth Jackson, in Crabgrass Frontier (1985), and Jane Jacobs, in The Death and Life of Great American Cities (1961), for instance, note the impact that urban planning has had on the quality of life in our neighborhoods and cities-especially after World War II. Of particular interest for both authors was the loss of "street life" as a result of both neighborhood-scale elements, such as the elimination of land use diversity, population diversity, and usable public spaces (Jacobs, 1961), and smaller-scale elements, such as the removal of parlors and porches from housing designs (Jackson, 1985). They note how this loss of street activity resulted in a loss of cohesiveness and perceived safety in our neighborhoods (Jacobs, 1961) and the privatization and isolation of life in automobile-dependent suburbs (Jackson, 1985). It is this loss of community life, whether real or perceived, to which New Urbanism is responding in its "smart growth" guidelines.

The question now is whether changing the way we design our neighborhoods-particularly their public spaces-can help revive the strong community life observed in early-20th-century neighborhoods. A number of authors and researchers, such as Alexander (1977) and Whyte (1980), have advocated strongly for the link between design and behavior. Others dispute the arguably deterministic view that people's daily lives-either their behaviors or attitudes-are guided by their surroundings. Audirac and Shermyn (1994), for instance, argue that the isolated and automobile-oriented nature of Americans' lives is the result of a shift in lifestyles and individual preferences rather than the structure and design of our cities and neighborhoods. Of particular interest for this article, however, are studies that focus on the link between design and behaviors such as local pedestrian travel and social interaction among neighbors. Unfortunately, few studies have examined both types of behaviors simultaneously. For pedestrian travel, we must turn to literature in the field of urban and transportation planning; for social interaction, we turn to environmental or community psychology.

One of the first in-depth, comprehensive studies of the link between small-scale environments (in this case, streetscapes) and the social life of a neighborhood was Appleyard and Lintell's (1972) Livable Streets. This study connected human-scale, people-friendly street designs to increased interaction among and knowing of one's neighbors, as well as increased children's play. Since then, this area of research has received a great deal of attention, including a follow-up study to Appleyard and Lintell (Bosselmann et al., 1999) that provided general support for the earlier study's findings. In addition to streetscape designs, casual contact and interaction among neighbors has also been frequently connected to the presence of semipublic spaces, particularly front porches (Abu-Ghazzeh, 1999; Bothwell et al., 1998; Brown et al., 1998; Skjaeveland & Garling, 1997), the presence of trees and other vegetation (Kuo et al., 1998), and the spaciousness and arrangement of open spaces (Abu-Ghazzeh, 1999; Skjaeveland & Garling, 1997).

Neighborhood and streetscape environments also affect the frequency with which people walk in their neighborhoods. Pedestrian travel is higher, for instance, in neighborhoods with adequate pedestrian facilities and amenities, such as continuous sidewalks and safe street crossings (Kitamura et al., 1997; Moudon et al., 1996), higher densities (Frank & Pivo, 1995; Kitamura et al., 1997), and access to local amenities (Gordon & Peers, 1993). This last factor is of particular significance for this study and, as Handy has concluded in multiple studies (1992, 1996a, 1996b), is related to both the presence of those amenities and the directness of the routes that lead to them. Handy's work is also supported by Steiner (1996), who identifies distance as the strongest determinant of a decision to walk to a store, and Shriver (1996), who attributes higher walking rates in traditional neighborhoods to the more direct routes and the number of destination choices compared to conventional suburban developments. More automobile-oriented studies, such as Ewing, Haliyur, and Page (1995) and Cervero and Radisch (1995), have also found traditional neighborhoods, with their internalized facilities and services, to be less dependent on automobiles than their suburban counterparts.

Finally, a small number of studies have taken a more holistic approach to the link between community design and resident behaviors, albeit in a more qualitative manner. Through interviews conducted in New Urbanist and/or older, traditional neighborhoods, Langdon (1997) and Plas and Lewis (1996) both concluded that people were more likely to walk in these communities and that there were higher levels of interaction among neighbors on account of the physical layout of the communities, particularly the accessibility of retail shops, open spaces, and other everyday amenities and the pleasant walking environment.

It is also important to note, however, that "pleasant walking environment" is very subjective and that such perceptions also have a significant effect on individual behaviors. In a survey of the neighboring behaviors of multifamily housing residents, for instance, Skjaeveland and Garling (1997) found that residents who perceived their surroundings as more open or spacious were more likely to engage in neighboring behaviors. Similarly, from a household survey of residents' pedestrian travel patterns, Handy (1996b) found that respondents' perceptions of their local environment relate significantly to pedestrian trip frequencies, with (as you might expect) more trips made by those with more positive perceptions. Of particular importance-to both strolling trips and destination trips-are perceptions of safety and comfort when walking. Demetsky and Perfater (1975) also link pedestrian travel to perceptions of safety, with "fear of attack" serving as a significant impedance to walking, especially for women. Based on these findings, it is clearly important to consider subjective factors in any study of individual behaviors.

This article adds to the existing body of literature in two ways. First, it bridges two interrelated fields of research that have been explored largely in isolation from one another. One is urban and transportation planning, which has focused on the relationships between urban form and travel behavior; the other is community and environmental psychology, with its focus on the relationships between physical environments and more socially oriented behaviors. To truly understand communities and how they function, we need to be taking a more multidisciplinary approach; a bridging of fields is therefore critical. Second, this article further examines two under-researched areas: the relative role of attitudes and design elements in predicting behaviors and the link between pedestrian travel and neighbor interactions. In addition, the study avoided adding to the plethora of measurement tools (which results in non-comparable data and less reliable results) by using existing scales developed by Handy (1996b) and Skjaeveland et al. (1996).

Research Design

To evaluate these New Urbanist claims, the study examined pedestrian travel behavior and neighbor interaction in eight neighborhoods of varying design in the Portland, Oregon, metropolitan region. Four are inner-city neighborhoods and four are suburban developments. Individual-level data on walking and neighboring behaviors, personal attitudes, sociodemographic characteristics, and perceptions of the local environment were collected through household surveys consisting of both quantitative and more exploratory qualitative questions. This study also addressed the issue of neighborhood self-selection (whether observed differences in behavior are attributed to differences in design or to residents selecting neighborhoods that support their preferred behaviors) by examining the relative influence of local access and personal attitudes on walking and neighboring behaviors.

Research Limitations

The most significant limitation of the research is the study's narrow sociodemographic focus. The suburban developments are nearly homogeneous, populated by middle-income, primarily non-Hispanic White homeowners. In order to control for the influence of these nondesign factors to the greatest extent possible, it was necessary to match these neighborhoods with inner-city neighborhoods of similar sociodemographic characteristics. The study therefore under-represents minority residents, renters, and residents with very low and very high incomes. Factors that contribute to community life in neighborhoods of a different ethnic or economic character may differ from those identified in this study and should be the focus of separate research.

A second limitation is the potential self-selection of respondents. The most effective way to minimize this limitation is to maximize the survey response rate. This was attempted by using a four-stage mail-out/mail-back data collection process and by surveying all households (rather than a sample of households) within each neighborhood. Even with these efforts, the data collection process yielded an overall response rate of just 34%. On the positive side, however, these responses reflect 34% of the entire neighborhood populations rather than 34% of a sample. Through an analysis of respondent characteristics, it was also determined (where census data were available) that the only significant difference between respondents and the total populations of their respective neighborhoods was the presence of children. Survey respondents were more likely to be living with children, suggesting that the results presented in this study may be most applicable to households with children.

Hypotheses

The research hypotheses, as well as the direction of each hypothesized relationship, are based on the New Urbanises' claims that mixed-use, pedestrian-friendly neighborhoods have higher rates of walking and neighboring than single-use, automobile-oriented neighborhoods. With regard to pedestrian travel, the study hypothesized that, controlling for relevant sociodemographic factors, measures of walking trip frequencies and neighboring behaviors will be highest in the neighborhoods with access to local parks and retail shops and lowest in the neighborhoods with neither of these amenities. At the individual level, walking trip frequencies were hypothesized to have a significant positive relationship with objective and subjective physical variables, and neighboring behaviors were hypothesized to have a significant positive relationship with objective and subjective physical variables and walking trip frequencies.

The study also proposed that there would be a link between neighborhood age and behavioral outcomes, with older neighborhoods experiencing higher levels of neighboring behavior (because residents are more likely to have lived in the neighborhood for many years and to have established relationships with their neighbors), as well as walking (because longer-term residents are more likely to feel comfortable in their neighborhood and have had more time to establish connections to local shops).

Neighborhood Selection Criteria

For the purposes of this study, neighborhoods were defined by functional rather than political boundaries. The primary criterion was the level of access (see Note 1) to local parks and retail shops. The final selection of neighborhoods included two with access to both a park and neighborhood shopping area, two with access to only a park, two with access to only a shopping area, and two with access to neither parks nor retail shops (of any kind). In each neighborhood with local access, all households were walking distance from the same park(s) and/or shopping area.

Potential neighborhoods within the Portland metropolitan region were identified using a geographical information system (GIS). All residential properties that met each of the above criteria were selected from a regional database, producing easily identifiable clusters of housing, or "neighborhoods." Figure 1 shows the locations of the neighborhoods in the Portland metropolitan region.

Using GIS, planning documents, and site visits, these neighborhoods were then evaluated for the following characteristics:

* Route directness. Neighborhoods were eliminated from the access categories if there were no direct pedestrian routes between the housing units and the destinations.

* Quality of the pedestrian environment. All neighborhoods selected were required to have sidewalks, fairly level terrain, shallow setbacks, narrow lots, and buffers between the sidewalk and street (planting strips, on-street parking, or both).

* Quality of the local park/shopping area. Both parks and retail shopping areas had to be well maintained and to demonstrate an observable level of use (based on subjective evaluations conducted on multiple site visits).

* Neighborhood era. For each category of access, one new suburban development (developed within the last 5 years) and one older inner-city neighborhood (developed prior to 1945) were selected.

* Median property value. Each neighborhood selected had a median property value (house and land) of approximately $200,000, slightly above the median for the entire Portland metropolitan region.

These selection criteria produced three independent variables (local park access, local retail access, and era of development) and two control variables (pedestrian environment and property value).

The Neighborhoods

Each selected neighborhood contains approximately 150-200 single-family housing units. (The study was intentionally limited to single-family units because only two of the eight neighborhoods have any multifamily housing.) As a result of the requirement for narrow lots in the suburban developments, the neighborhoods are also similar in density. Figures 2 and 3 show typical sites in the neighborhoods.

The New Subdivisions

The four new subdivisions selected for this study are Orenco Station, Bethany Village, Jones Farm, and Arbor View. They are located on the west side of the Portland region in Washington County, where development regulations actually require some of the neighborhood selection criteria described above, such as sidewalks and shallow setbacks (see Figure 2).

Orenco Station. Orenco Station was developed (starting in 1997) as a New Urbanist community and represents the new pedestrian-oriented subdivision. The neighborhood is dominated by traditional bungalow-style homes. All housing units are within 1/4 mile of a variety of retail shops and services and a large park with benches and other amenities, as well as at least one small open space. Pedestrian-oriented streetscapes and alleys help reduce the dominance of the car, and street trees, while not mature enough yet to provide shade to walkers, create a more natural and human-scale feeling.

Enlarge 200% Enlarge 400%

Bethany Village. The Bethany Village subdivision was developed from 1995 to 1997 and meets two selection criteria: pedestrian environment and local retail access. Each of its housing units has local access to a shopping area of the same name that was designed and marketed as a New Urbanist retail development. The Bethany Village shopping area is pedestrian friendly and has a variety of shops and services but was not developed in conjunction with, or fully integrated into, the Bethany Village subdivision. Residents do not have access to a local park.

Jones Farm. The Jones Farm subdivision was developed during the same time period as Orenco Station and is similar in character, but lacks local access to retail shops. Residents do have local access to two parks with varying amenities. At the center of one development phase is a fairly large landscaped park with benches. To the south is a more child-oriented park with benches and playground equipment. The entire development is pedestrian oriented, with sidewalks, paths, and ample landscaping.

Arbor View. The Arbor View subdivision was constructed from 1997 to 1998 and is similar in character to Bethany Village. While the streetscapes are pedestrian oriented, none of its housing units has local access to a park or retail shop.

The Inner-City Neighborhoods

The four inner-city neighborhoods selected for this study are Ladd's Addition, Beaumont, Alameda-33rd Avenue, and Alameda-Bryce. Each is a well-preserved pre-World War II neighborhood in Portland's "inner eastside," which is composed almost entirely of its original, turn-of-the-century homes (see Figure 3).

Ladd's Addition. Ladd's Addition, in southeast Portland, meets all three selection criteria. Like Orenco Station, Ladd's Addition has a network of alleys and many street trees, although Ladd's trees are much older and their shade is more dense. There is one large open space in the center of the neighborhood, plus four smaller gardens scattered throughout the site. All housing units have local access to a variety of retail shops and services in a nearby main-street-style shopping area.

Beaumont. The Beaumont neighborhood, located in northeast Portland, has pedestrian-oriented street-scapes and local access to retail shops but not to a park. Beaumont also has mature street trees that are not as dense as in Ladd's Addition but do provide moderate shade for pedestrians. As in Ladd's Addition, local retail shops and services are located in a main-street-style shopping area.

Alameda-33rd Avenue. The Alameda-33rd Avenue neighborhood has narrow, pedestrian-oriented streets lined with large shade trees and bungalow-style homes. This neighborhood has local access to a park but not to retail shops. A large park located just east of the neighborhood has benches and picnic tables, playground equipment, a playfield, and a running trail.

Alameda-Bryce. Alameda-Bryce is an established neighborhood with pedestrian-oriented streets but no local everyday amenities. This neighborhood borders the Alameda-33rd Avenue neighborhood and is thus similar in character, but none of its housing units are within easy walking distance of a park or retail shops.

Data Collection and Variables

Data Collection

Data were collected in early fall using a mail-out/mail-back survey. Every single-family housing unit in each selected neighborhood-a total of 1,454-received a four-page questionnaire that was stamped with a numerical geocode. A cover letter asked that the questionnaire be filled out by an adult member of the household. The overall response rate was 34%-slightly higher in the inner-city neighborhoods (37-41%) and slightly lower in the suburban developments (24-36%). Refer back to the section on research limitations for a discussion of the potential for self-selection bias.

Dependent Variables

Pedestrian Travel Behavior. Pedestrian travel behavior consists of two variables: frequency of strolling trips and frequency of destination trips (frequency is the number of times in the previous week the respondent reported the behavior). These forms were separated because, according to Handy (1996b), each is associated with a different set of motivations.

Neighboring Behaviors. To capture the multiple dimensions of neighboring (see Skjaeveland et al., 1996), neighboring behavior consists of three variables: frequency of unplanned interactions (chance encounters) with one's neighbors, local social ties (number of acquaintances within close proximity of home), and supportive acts of neighboring (frequency with which one gives/receives assistance to/from neighbors). Frequency of unplanned interactions is the number of times in the previous week respondents waved or said hello to neighbors, stopped and chatted with neighbors, and/or invited neighbors inside their home. The other two variables were measured using Skjaeveland et al.'s (1996) Multidimensional Measure of Neighboring (MMN) scale. The study also included a "neighbor annoyance" variable (frequency and degree to which one was irritated by neighbors), which was also measured using the MMN scale; those findings are not presented here because they were not significant.

Independent Variables

The independent variables are grouped into three categories containing a total of five sets: personal variables (sociodemographic and attitudinal characteristics), neighborhood variables (objective and subjective evaluations of the physical environment), and behavioral variables (walking trip frequencies). These are the same sets used for the hierarchical regression models. Not all variables described here are included in every regression equation.

Personal Variables. Two sets of variables focused on personal characteristics of the respondents: sociodemographic and attitudinal. Both sets served as control variables in the regression models. Sociodemographic characteristics include age group, gender, race, number and ages of children, and whether the respondent identifies as a "homemaker." (Other sociodemographic variables collected, but not included in the regression models due to a lack of overall variation among respondents, include approximate household income, home ownership, and a more detailed breakdown of race/ethnicity.) Attitudinal characteristics measure respondents' attitudes toward the importance of walking to daily activities, interacting with one's neighbors, and feeling "at home" in the neighborhood. Each was measured on a three-point scale based on the research of Kitamura et al. (1997).

Enlarge 200% Enlarge 400%

Enlarge 200% Enlarge 400%

Neighborhood Variables. Variables representing the physical environments of the neighborhoods include both objective evaluations and respondents' subjective evaluations. The objective physical variables are defined by the neighborhood selection criteria and include dichotomous variables for local access to retail only (no parks), local access to parks only (no retail), local access to parks and retail, and location in the inner city. Respondents' subjective evaluations of the neighborhood were measured in three ways: satisfaction with local parks, satisfaction with the local shopping area (measured separately, each using a four-point scale), and "perception of walking in neighborhood" (measured using an 11-item scale developed by Handy, 1996b).

Behavioral Variables. While pedestrian travel was treated primarily as a dependent variable in this study, walking trip frequencies were also included as independent variables in each of the regression models for neighboring behaviors. This allowed the study to examine the impact that walking in one's neighborhood has on neighbor interactions (after controlling for all other variables).

Analyses and Results

Each of the dependent variables was examined using first an analysis of covariance (to detect neighborhood-level variations) and then a hierarchical regression model (to explain individual-level variations).

Neighborhood-Level Analyses

The analyses of covariance (ANCOVAs) tested the hypotheses that walking trip frequencies and neighboring behaviors would be highest in the neighborhoods with local access to parks and retail shops and lowest in the neighborhoods with no local access to either (see Tables 1 and 2). For each ANCOVA, sociodemographic variables that correlate significantly with the dependent variable were included as covariates.

Four groups were defined according to each neighborhood's local access characteristics: Group 1 = park and retail access, Group 2 = retail access only, Group 3 = park access only, and Group 4 = no local access. Differences in mean values were analyzed overall (with the F-statistic) and among specific means. For the analysis among specific means, Group 4 was specified as the control group; Groups 1, 2, and 3 were compared to Group 4 using a simple contrasts model. A second set of ANCOVAs (not presented in the tables) was also conducted with groups defined according to each site's era of development (Group 1 = inner-city neighborhoods; Group 2 = suburban developments).

Two variables vary significantly across local access groups, but not across development eras: destination trips and frequency of unplanned interactions. Compared to the neighborhoods with no local access, destination trips are significantly higher in the neighborhoods with local access to retail shops, either alone or in combination with local access to parks. Unplanned interactions, on the other hand, are significantly higher in the neighborhoods with local access to parks, either alone or in combination with access to retail shops.

Differences across development eras, but not across local access groups, are significant for destination trip frequencies (Inner-city: mean = 3.12, mean standard error [MSE] = 0.17; Suburban: mean = 1.97, MSE = 0.22) and number of local social ties (Inner-city: mean = 4.74, MSE = 0.06; Suburban: mean = 4.30, MSE = 0.08). In each case, mean values for the inner-city neighborhoods are significantly higher than for the suburban developments at the 95% confidence interval. Strolling trip frequencies do not vary significantly across local access groups or development eras. This finding is consistent with Handy (1992, 1996a).

For the remaining variable-supportive acts of neighboring-the ANCOVA findings are less straightforward, with interaction plots revealing an interaction between development era and local access group. This interaction is explored further in the regression analysis for supportive acts of neighboring.

Individual-Level Analyses

Individual-level analyses used a hierarchical regression technique to examine the relative effect of sets of variables, as well as that of individual variables. The order of the hierarchy is structured first to remove potentially confounding variables (in this case, sociodemographic and attitudinal attributes) and then to reflect the causal priority of the remaining variables. These remaining sets were entered in the hierarchical regression equation so that-to the extent possible-no variable set entering later is a cause of an earlier set. Objective physical variables were entered first, followed by subjective physical variables, and then, in certain models, behavioral variables. Due to space constraints, however, only the final models are presented in Tables 3 and 4.

Enlarge 200% Enlarge 400%

Enlarge 200% Enlarge 400%

Overall, the regression model explains about one third to nearly one half the variation in supportive acts of neighboring, number of local social ties scores, destination trip frequencies, and frequencies of unplanned neighbor interactions. The model is less useful, however, in explaining variations in strolling trip frequencies, accounting for just over one tenth of the total variation.

Pedestrian Travel. Separate regression models were conducted for destination and strolling trips.

* Destination trips. The number of destination trips an individual made during the week was linked most significantly to attitudinal factors (R^sup 2^ change = 0.18, F change = 49.92, p < .01). Those who placed importance on this activity were more likely to walk to local stores than those who did not. Destination trips were also linked to objective environmental factors (R^sup 2^ change = 0.11, F change = 17.41, p < .01), primarily local access to retail shops. The sociodemographic and subjective environmental variable sets each increased the explanatory power by just 2%.

* Strolling trips. As mentioned previously, strolling trips are not well explained by the factors included in this study. Attitudinal variables were the most powerful in predicting strolling trip frequencies (as was the case for destination trips), but even this set explained less than 5% of the total variation. Individual variables that correlated significantly with strolling trips were placing importance on walking to daily activities and identifying oneself as a "homemaker," both of which had a positive impact on strolling, followed by "perceptions of walking in neighborhood" (positive), "retail access only" (negative), and presence of children ages 5 to 12 in a household (negative).

Enlarge 200% Enlarge 400%

* Summary. These pedestrian travel findings provide strong support for the hypothesis that destination trip frequencies have a significant positive relationship with objective physical variables, some support for the hypothesis that destination trip frequencies have a significant positive relationship with subjective environmental variables, and no support for the hypothesized relationships between these environmental variables and strolling trips. Of greatest significance, however, were respondents' attitudes toward walking.

Neighboring Behaviors. Separate regression analyses were conducted for frequency of unplanned interactions, local social ties, and supportive acts of neighboring.

* Frequency of unplanned interactions. The frequency with which each respondent had unplanned, or chance, encounters with neighbors (see Table 4) is explained most significantly by walking trip frequencies, represented by the behavioral variable set (R^sup 2^ change = .10, F change = 30.37, p < .01). Personal characteristics are also highly significant, with both the sociodemographic variable set and the attitudinal variable set adding 9% to the total explanatory power (p < .01). Contributing the least amount of explanatory power (2% each) are the objective and subjective environmental variable sets. Individually, the variables of greatest significance in the final model include, respectively, strolling trip frequencies, the importance one places on neighbor interaction, having children ages 5 to 12 in the household, and having local access to parks only. Each relates positively to unplanned interactions.

* Local social ties. Variation in the number of social ties a respondent had with neighbors is attributed almost entirely to personal characteristics (see Table 4). Sociodemographic variables explain 14% of the total variation (R^sup 2^ = 0.14, F = 8.67, p < .01); most significant are age group and the presence of children ages 0 to 4, both of which related positively to local social ties. Attitudinal variables, entered next, increased the explanatory power of the model by 20% (R^sup 2^ change = 0.20, F change = 67.98, p < .01). The remaining variable sets-objective environmental, subjective environmental, and behavioral-together increase the total explanatory power by just 5%.

* Supportive acts of neighboring. The regression model for supportive acts of neighboring also attributes most of its explanatory power to personal variables: sociodemographic variables explain 11% of the total variation (R^sup 2^ = 0.11, F = 7.10, p < .01), and attitudinal variables explain an additional 31%. Length of residency, the presence of children ages 5 to 12, and the importance one places on neighbor interaction were the most significant variables. Each relates positively to individual acts of neighboring.

As a whole, there is not a significant direct relationship between the objective environmental variables and acts of neighboring. The interaction between "inner-city neighborhood" and "retail access only" is significant, however, even when all other variables are controlled. This suggests that local access to retail shops contributes positively to acts of neighboring in inner-city neighborhoods, but not in suburban developments. Possible explanations for this interaction are explored in the discussion section below. The remaining variable sets-subjective environmental (i.e., "perceptions of walking in neighborhood") and walking behaviors-increase the explanatory power of the equation by, respectively, 3% (p < .01) and a nonsignificant amount.

* Summary. These findings provide some support for the hypothesis that neighboring behaviors are positively and significantly related to objective physical factors; this is true, however, only for frequency of unplanned interactions and supportive acts of neighboring. The hypothesis that neighboring behaviors are positively and significantly related to subjective physical factors is also somewhat supported, this time for all three measures of neighboring. Finally, the hypothesis that walking in one's neighborhood positively and significantly contributes to neighboring is strongly supported for unplanned interactions, somewhat supported for local social ties, and not supported for supportive acts of neighboring.

Discussion

The results presented in this article provide some support for each of the three relationships tested: between local access and pedestrian travel, between pedestrian travel and neighboring behaviors, and between local access and neighboring behaviors. The results also provide a strong indication, however, that there are nondesign factors, particularly personal attitudes, which are also of significance and must be considered in future discussions and research.

Local Access and Pedestrian Travel

In the case of destination trips, this study supports the New Urbanist claim that local access contributes to increased levels of pedestrian travel. Local access to retail shops appears to be of particular importance-at both the individual and neighborhood levels. Residents do appear to be using-and walking to-their local shopping area, if there is one, a finding that supports past studies by Handy (1992, 1996b), Shriver (1997), Steiner (1996), and Rutherford et al. (1998). It is possible that the limited availability of parking in inner-city shopping areas also contributes to higher destination trip frequencies in these areas. Future research on the potential for New Urbanist designs should analyze the relative roles in encouraging pedestrian travel of incentives for walking (e.g., local access and pedestrian-friendly designs) and disincentives for driving (e.g., limited parking availability and parking costs).

There is not strong support, however, for the relationship between local access (even to parks) and strolling trips, with one interesting exception: Strolling trips were significantly lower in neighborhoods with local access to retail only than in neighborhoods with no local access to either parks or retail. While this appears to support Shriver's (1996) finding that walking trips in retail-accessible neighborhoods are more likely to be destination oriented, another likely explanation is that respondents are not necessarily strolling less, but that they are simply more likely to stop at a local store, restaurant, or coffee shop before returning home. Respondents' own reports of why they walk in their neighborhood provide some support for this explanation. In the retail-accessible neighborhoods, for instance, respondents were actually more likely to report strolling-related walking motives such as "exercise, fresh air, and relaxation" than respondents from other neighborhoods. However, they were also more likely to report that they walk to local shops.

Pedestrian Travel and Neighboring Behaviors

The quantitative analyses suggest that respondents who walk more are in fact-as New Urbanism literature suggests-more likely to engage in unplanned interactions with their neighbors and to form social ties with nearby neighbors. In both cases, it appears that strolling trips were more conducive to these neighboring behaviors. This is understandable, due to the nature of the trips themselves: whereas destination walkers are likely to be walking out of necessity or under time constraints, strollers are more likely to have time to stop and chat. They may even be walking for the purpose of socializing with their neighbors. Contrary to New Urbanism literature, however, the link between design and neighboring behaviors is not direct; while strolling in one's neighborhood does appear to contribute to neighboring behaviors, it is not strolling trips but rather destination trips (as discussed above) that are influenced by design.

Also of interest is the finding that individuals who walk in their neighborhood were not more likely to engage in supportive acts of neighboring (on either the giving or receiving end). This suggests that walking increases the frequency of casual interactions, as we would expect, but that more intense forms of interaction depend on other influences, such as (for example) attitudes toward neighboring, demographics, similarities with neighbors, or extent of integration into the neighborhood.

Local Access and Neighboring Behaviors

Even when controlling for individuals' walking crip frequencies, neighboring behaviors were still more frequent among respondents who had local access to parks and/or a retail shopping area. This suggests that these destinations may serve as places of contact for neighbors, regardless of how they get there. For instance, having parks within walking distance increased the likelihood that neighbors would come into casual contact (i.e., unplanned interactions) with each other. And in neighborhoods without parks, the presence of local retail shops contributes to a second neighboring behavior: supportive acts of neighboring. This relationship holds true, however, only for the inner-city neighborhoods. A likely reason is that these shops may have developed a connection to the community that has not yet taken shape in the suburban developments. It could also be that the newer shopping areas draw customers from a larger geographic area than the more established ones, reducing or eliminating their "neighborly" atmosphere. This relationship is an area deserving more research attention.

Personal Attitudes and Perceptions

A particularly important finding of this study is the significant role that personal attitudes play, relative to neighborhood factors, in predicting individual behaviors. In many instances, personal attitudes toward a particular behavior (e.g., walking to daily activities, interacting with neighbors) were more important in predicting that behavior than objective neighborhood variables. This supports the findings of Kitamura et al. (1997) and suggests that personal attitudes need to be considered more carefully in behavioral studies as well as in planning efforts. This is especially true in the case of New Urbanism and other smart growth strategies that seek to enhance community-oriented behaviors such as walking and neighboring, as residents' attitudes are likely to affect the feasibility of meeting these goals.

Finally, individuals' behaviors were also related to another personal attribute: perceptions of the local environment. This was especially true for predicting walking behavior: Respondents were more likely to walk in their neighborhood if they had a favorable perception of the local walking environment. Breaking down the scale for "perception of walking in neighborhood" revealed that perceptions of comfort, opportunities for neighbor interaction, and feeling safe walking in the evening were especially important. Although not included in the regression analyses (because scores were collected only for respondents with local access to retail shops), a "perception of the local shopping area" scale revealed that feeling safe walking to local stores and perceiving the stores as being within a comfortable walking distance from home were also important factors.

Positive perceptions of the local walking environment also related to more frequent neighboring behaviors, even when controlling for actual walking behaviors. This suggests that pedestrian-oriented environments are conducive to neighboring, even if people are not using the streets for walking. This conclusion would support Appleyard and Lintell's (1972) finding that people-friendly streetscapes serve not only as places to walk, but also as gathering places for neighbors and spaces for children to play.

It is possible, however, that these causal relationships (that positive perceptions increase the likelihood of walking and neighboring) flow in the opposite direction: Strolling in one's neighborhood and interacting with one's neighbors may positively influence one's perceptions of the neighborhood. It is even more likely that these are reciprocal relationships. This is a question that could benefit from further research.

Neighborhood Self-Selection

Neighborhood self-selection is a common concern in studies of neighborhood-level phenomena, particularly those involving New Urbanist communities. The argument is that neighborhood variations in pedestrian travel, for instance, are not an indication of neighborhood design factors influencing or changing travel behaviors, but rather of the self-selection of residents into neighborhoods that enable them to continue their existing behaviors (Boarnet & Crane, 2001). Of course, if residents are choosing neighborhoods that allow them to engage in behaviors that were not feasible in their previous neighborhood, this argument is weakened.

If self-selection were the source of neighborhood-level variations in pedestrian travel or neighboring behaviors, we would expect both the attitudes toward these behaviors (assuming that these attitudes played a role in residential selection) and the behaviors themselves to be higher in the neighborhoods most conducive to those behaviors. Analyses of covariance found that this was in fact the case for destination trips, suggesting that respondents who value being able to walk to their daily activities may be self-selecting into neighborhoods with local access to retail shops. This self-selection may provide a partial explanation for the higher destination trip frequencies in these neighborhoods. With regard to neighboring behaviors, however, self-selection did not appear to be a factor. While attitudes toward neighboring were more positive in the older, inner-city neighborhoods, their residents generally did not exhibit higher levels of neighboring behaviors.

Conclusions

Implications for New Urbanism

One can conclude from the findings presented here that there is some credibility to at least two claims of smart growth or New Urbanism: (1) when combined with pedestrian-friendly streetscapes, locating everyday amenities such as parks and retail shops within a neighborhood can increase pedestrian travel and neighbor interaction within a community, and (2) people who walk around their neighborhood are more likely to interact with and form relationships with their neighbors. This study also indicates, however, that personal variables such as demographics, attitudes, and perceptions of the local environment are also important factors. What this means for planners is that there is potential for enhancing the social life of communities through New Urbanism, but that these communities will be more successful-at least in the short run-if they are able to attract individuals seeking a more community-oriented neighborhood. It is also important that residents' attitudes be taken into consideration when evaluating New Urbanist communities: Heightened levels of walking and neighbor interaction are likely to be achieved more rapidly in communities where these behaviors are valued.

Suggestions for Further Research

While this study furthers our understanding of the relative roles of personal and environmental factors in predicting behaviors, as well as the ways in which walking contributes to various forms of neighboring, it also points to a number of areas in need of further research. Most notably, we need to increase our understanding of (1) how personal attitudes and values are formed, (2) the extent to which attitudes and values can be influenced by residents' physical or social surroundings, (3) why people move into New Urbanist communities, and (4) what attributes make neighborhood parks and shopping areas appealing to local residents. Studies such as this one also need to be extended into communities of different sociodemographic makeups, not only to understand the role that factors such as income and culture play, but also to increase the chances of success in implementing New Urbanist strategies in these communities.

ACKNOWLEDGMENTS

I would like to thank Dr. Nancy Chapman for helping me to explore and understand the social and psychological side of communities and urban planning. I am also grateful to the other members of my dissertation committee at Portland State University for their valuable insights and to the anonymous referees and the JAPA editors for their extremely thorough remarks on earlier drafts of this article. Any errors or oversights are my own.

[Reference]  »  View reference page with links

REFERENCES

Abu-Ghazzeh, T. M. (1999). Housing layout, social interaction, and the place of contact in Abu-Nuseir, Jordan. Journal of Environmental Psychology, 19, 41-73.

Alexander, C. (1977). A pattern language. New York: Oxford University Press.

Appleyard, D., & Lintell, M. (1972). The environmental quality of city streets: The residents' viewpoint. Journal of the American Institute of Planners, 38(2), 84-101.

Audirac, I., & Shermyen, A. H. (1994). An evaluation of neo-traditional design's social prescription: Postmodern placebo or remedy for suburban malaise? Journal of Planning Education and Research, 13, 161-173.

Boarnet, M., & Crane, R. (2001). Travel by design: The influence of urban form on travel. New York: Oxford University Press.

Bosselmann, P., Macdonald, E., & Kronemeyer, T. (1999). Livable streets revisited. Journal of the American Planning Association, 65(2), 168-180.

Bothwell, S. E., Gindroz, R., & Lang, R. E. (1998). Restoring community through traditional neighborhood design: A case study of Diggs Town public housing. Housing Policy Debate, 9(1), 89-114.

Brown, B. B., Burton, J. R., & Sweaney, A. L. (1998). Neighbors, households, and front porches: New urbanist community tool or mere nostalgia? Environment and Behavior, 30(5), 579-600.

Calthorpe, P. (1989). Pedestrian pockets: New strategies for suburban growth. In D. Kelbaugh (Ed.), The pedestrian pocket book: A new suburban design strategy (pp. 7-20). New York: Princeton Architectural Press.

Cervero, R., & Radisch, C. (1995). Travel choices in pedestrian versus automobile oriented neighborhoods (Working Paper UCTC No. 281). Berkeley: University of California Transportation Center.

Congress for the New Urbanism. (1998). Charter of the New Urbanism. Retrieved July 2003 from http://www.cnu.org/cnu_reports/Charter.pdf

Demetsky, M. J., & Perfater, M. A. (1975). Assessment of pedestrian attitudes and behavior in suburban environments. Transportation Research Record, 540, 46-55.

Ewing, R., Haliyur, P., & Page, G. W. (1995). Getting around a traditional city, a suburban planned unit development, and everything in between. Transportation Research Record, 1466, 53-62.

Frank, L. D., & Pivo, G. (1995). Impacts of mixed use and density on utilization of three modes of travel: single-occupant vehicle, transit, and walking. Transportation Research Record, 1466, 44-52.

Gordon, S. P., & Peers, J. B. (1993). Designing a community for transportation demand management: The Laguna West pedestrian pocket. Transportation Research Record, 1321, 138-145.

Handy, S. L. (1992). Regional versus local accessibility: Neo-traditional development and its implications for non-work travel. Built Environment, 18(4), 253-267.

Handy, S. L. (1996a). Understanding the link between urban form and nonwork travel behavior. Journal of Planning Education and Research, 15, 183-198.

Handy, S. L. (1996b). Urban form and pedestrian choices: Study of Austin neighborhoods. Transportation Research Record, 1552, 135-144.

Jackson, K. T. (1985). Crabgrass frontier: The suburbanization of the United States. New York: Oxford University Press.

Jacobs, J. (1961). The death and life of great American cities. New York: Vintage Books.

Kitamura, R., Mokhtarian, P. L., & Laidet, L. (1997). A microanalysis of land use and travel in five neighborhoods in the San Francisco Bay Area. Transportation, 24, 125-158.

Kuo, F. E., Sullivan, W. C., Coley, R., & Brunson, L. (1998). Fertile ground for community: Inner-city neighborhood common spaces. American Journal of Community Psychology, 26(6), 823-852.

Langdon, P. (1997). Can design make community? The Responsive Community, 7(2), 25-37.

Moudon, A. V., Hess, P. M., Snyder, M. C., & Stanilov, K. (1996). Effects of site design on pedestrian travel in mixed-use, medium-density environments. Transportation Research Record, 1578, 48-55.

Plas, J. M., & Lewis, S. E. (1996). Environmental factors and sense of community in a planned town. American Journal of Community Psychology, 24(1), 109-143.

Rutherford, G. S., McCormack, E., & Wilkinson, M. (1998). Travel impacts of urban form: Implications from an analysis of two Seattle area travel diaries. Washington, DC: U.S. Department of Transportation, Bureau of Transportation Statistics.

Shriver, K. (1996). Influence of environmental design in pedestrian travel behavior in four Austin neighborhoods. Transportation Research Record, 1578, 64-75.

Skjaeveland, O., & Garling, T. (1997). Effects of interactional space on neighbouring. Journal of Environmental Psychology, 17, 181-198.

Skjaeveland, O., Garling, T., & Maeland, J. G. (1996). A multi-dimensional measure of neighboring. American Journal of Community Psychology 24(3), 413-435.

Steiner, R. L. (1996). Traditional neighborhood shopping districts: Patterns of use and modes of access. (Dissertation UCTC No. 37). Berkeley: University of California Transportation Center.

Whyte, W. (1980). The social life of small urban spaces. Washington, DC: The Conservation Foundation.