residing in particular environments, does that mean that they are more physically active, overall? These questions are of great current interest in the public health field where there is much concern about low levels of physical activity and increasing obesity, with Design and Destinations: Factors Influencing Walking and Total Physical Activity Ann Forsyth, Mary Hearst, J. Michael Oakes and Kathryn H. Schmitz [Paper first received, March 2007 in final form, August 2007] Abstract Do people walk more, or less, depending on the physical character of their residential areas rather than merely their individual characteristics? This paper reports fi ndings for the Twin Cities, Minnesota, about how walking and total physical activity are affected by street pattern, ���pedestrian-oriented��� infrastructure and amenities, and mixed use or destinations���in shorthand, design and destinations. The effects of density are dealt with in less depth. Like earlier studies, it fi nds that walking for specifi c purposes (i.e. travel or leisure) varies in relation to the physical characteristics of places. However, this study using multiple measures of overall walking and physical activity suggests that socially similar people do the same total amount of physical activity in different kinds of places and that level of activity is, on average, low. Introduction Do people walk more, or less, depending on the physical character of their residential areas rather than merely their individual characteristics? If they do walk more when 0042-0980 Print/1360-063X Online �� 2008 Urban Studies Journal Limited DOI: 10.1177/0042098008093386 Ann Forsyth is in the Department of City and Regional Planning, Cornell University, 106 West Sibley Hall, Ithaca, New York, 14853, USA. Fax: 607 255 1971. E-mail: af16@cornell.edu. Mary Hearst is in the Division of Epidemiology, University of Minnesota, Room 300 WBOB, 7525, 1300 S 2nd Street, Minneapolis, Minnesota, 55454, USA. Fax: 612 624 0315. E-mail: onei0085@umn.edu. J. Michael Oakes is in the Division of Epidemiology, University of Minnesota, Room 431 WBOB, 7525, 1300 S 2 nd Street, Minneapolis, Minnesota, 55454, USA. Fax: 612 624 0315E-mail: oakes@epi. umn.edu. Kathryn H. Schmitz is in the Department of Biostatistics and Epidemiology, University of Pennsylvania, 921, 8th Floor, Blockley Hall, 423 Guardian Drive, Philadelphia, Pennsylvania, PA 19104 6021, USA. Fax: 215 573 5315. E-mail: schmitz@mail.med.upenn.edu. 45(9) 1973���1996, August 2008 1973-1996 USJ_093386.indd 1973 1973-1996 USJ_093386.indd 1973 6/5/2008 5:10:29 PM 6/5/2008 5:10:29 PM Process Black Process Black

1974 ANN FORSYTH ET AL. implications for many health conditions from blood pressure to mental health (Warburtun et al., 2006 Spanier et al., 2006). If walking is a signifi cant part of total physical activity and higher levels of total physical activity are important for health, then increasing walking could have important public health benefi ts. Drawing mainly on work from transport and urban design, and focusing on walking for transport rather than exercise, four broad dimensions of the built environment have been proposed as likely correlates of walking and physical activity (PA). It has been suggested that walking is higher in areas with elevated residential and employment densities, more connected street patterns, quality pedestrian infrastructure and amenities such as side- walks and street trees, and the presence of a variety of destinations (mixed use) (Cervero and Kockelman, 1997 Frank and Pivo, 1994 Handy et al., 2002 Handy, 2003 Saelens et al., 2003). Of course, none of these studies has fully accounted for self-selection into neighbourhoods based on preferences for a built environment that supports walking. The complexities of disentangling signifi - cant correlations of the built environment and physical activity from those individuals that comprise the neighbourhood highlight issues of endogeneity, or unmeasured choices, in all such work. In addition, some of these characteristics are diffi cult to change in exist- ing areas���for example, street patterns. Others are easier and cheaper to retrofi t���for ex- ample, pedestrian amenities such as street lamps. Transport research has emphasised the importance of destinations���a trip is defi ned as movement between two destinations and so such destinations have been seen as key to movement. This paper reports findings about how both walking and total physical activity are affected by these factors in residential areas, focusing on three sets of factors: street pat- tern, ���pedestrian-oriented��� infrastructure and amenities, and mixed use���in shorthand, design and destinations.1 The effects of density are reported separately (Forsyth et al., 2007b). The study examined associations between physical activity (measured by survey, seven- day travel diary and accelerometer) and the built environment (measured using computer mapping and survey) for 715 participants in 36 environmentally diverse areas in the Twin Cities in Minnesota. It also examined the relative importance of these constructs compared with other social and psychological variables. Like most earlier studies, we found that walking for specifi c purposes (i.e. travel or leisure) varies in relation to the physical char- acteristics of places. However, the contribution of this research is to show that total physical activity is similar for similar people in different places. Further, that level of activity is, on average, low. The few associations of the built environment with overall physical activity have unclear policy implications.2 Results for specifi c sub-populations may differ from the larger population, but preliminary analyses show a similar lack of variation (Forsyth et al., 2007a). The apparent deviation in fi ndings from some earlier work is likely to be due to two factors: a focus in this paper on total physical activity rather than walking for specifi c pur- poses and, methodological innovations in this second generation of work on the rela- tionship between the built environment and physical activity. This study combines a survey, seven-day travel diary and accelerometer measures to assess reliably and objectively total physical activity rather than walking for a specifi c purpose. It adds objective meas- ures of the built environment to surveys of environmental perceptions and social and psychological characteristics. The study looked at a wider variety of defi nitions of neighbourhoods than previous studies, such as buffers of different sizes and shapes. The 1973-1996 USJ_093386.indd 1974 1973-1996 USJ_093386.indd 1974 6/5/2008 5:10:36 PM 6/5/2008 5:10:36 PM Process Black Process Black

DESIGN AND DESTINATIONS 1975 study also uses a sample design that minimises confounding by socioeconomic characteristics in that similar people in the study live in environmentally different areas, maximising exchangeability (Oakes, 2004). As with pre- vious studies, we were unable to account for the problem of endogeneity, in this case the potential that some measured variables refl ect unmeasured choices or characteristics of individuals. These fi ndings have important implications ��� They add further evidence to an emerging proposition that there is a physical activity budget, akin to a travel time budget. People who do more leisure time physical activity do less for other purposes and vice versa (Rodr��guez et al., 2006 Krizek et al., 2004). ��� It may well refl ect the daily mobility of most metropolitan residents who have choices about where and why to be active. Walking outdoors for travel and leisure in one���s neighbourhood���the typical focus of studies to date���may have less infl uence on total physical activity than has been assumed. The paper fi rst outlines debates about the association between walking, physical activity and features of the built environment de- scribes study methods and then provides an overview of select variables from the surveys, travel diaries and accelerometry, comparing these with design and destination features. It concludes with implications for the diffi - cult task of building physical activity into daily life. The fi nding that regular activity is not normative may require consideration in future intervention approaches. Encourage- ments beyond merely changing the resid- ential environment may well be needed and are likely to include typical public health strategies such as education campaigns and policy changes (like pricing). Conceptual Issues In considering physical activity, it is import- ant to distinguish between walking and other forms of physical activity and between utilitarian walking (primarily for transport) and leisure walking (primarily for exercise). This study drew on a base of research from transport and urban design that indicated that travel walking���measured by survey and diary���was affected by the built environment and particularly by four dimensions: density, street pattern, pedestrian infrastructure and amenities, and mixed use (Cervero and Kockelman, 1997 Crane, 2000 Ewing et al., 2003 Handy et al., 2002 Saelens et al., 2003). Environmental variables affecting leisure walking were less clear. Earlier research mostly relied on self-reported, rather than measured, characterisations of the built environment and thus focused on perceptions. While per- ceptions are important, one of the motiv- ations behind the fl urry of research on this topic is to see if urban planning interventions into the actual physical environment can increase physical activity. Fortunately, new developments in geographical information system (GIS) software and databases enable more sophisticated measurements of these variables at a variety of scales going beyond most earlier work (although see Moudon et al., 2004 for extensive use of GIS also earlier work by Cervero and Kockelman, 1997 Krizek, 2003a, 2003b Frank and Pivo, 1994). In this study, we used the variables listed in Table 1 with detailed protocols for how we measured the variables available on-line.1 The specifi c importance of the four categories of variables is outlined in what follows. In the results, we focus on those variables with signifi cant associations with a measure of physical activity and deal only briefl y with non-signifi cant variables. 1973-1996 USJ_093386.indd 1975 1973-1996 USJ_093386.indd 1975 6/5/2008 5:10:36 PM 6/5/2008 5:10:36 PM Process Black Process Black