Short Report Relationships between neighborhood walkability and adults’ physical activity: How important is residential self-selection? Delfien Van Dyck a,b,n , Greet Cardon b , Benedicte Deforche b,c , Neville Owen d , Ilse De Bourdeaudhuij b a Research Foundation Flanders (FWO), Brussels, Belgium b Ghent University, Department of Movement and Sports Sciences, Belgium c Vrije Universiteit Brussels, Department of Human Biometrics and Biomechanics, Belgium d Baker IDI Heart and Diabetes Institute, Melbourne; The University of Queensland, School of Population Health, Brisbane, Australia article info Article history: Received 6 December 2010 Received in revised form 5 April 2011 Accepted 6 May 2011 Available online 13 May 2011 Keywords: Belgium Exercise Accelerometers GIS Physical environment abstract The study’s aims were to examine whether residential self-selection differed according to socio- demographic characteristics and objectively assessed neighborhood walkability; and, whether objec- tively assessed walkability was a significant correlate of physical activity (PA) beyond residential self-selection. In total, 412 adults (aged 20–65 years) completed a socio-demographic questionnaire, the long IPAQ, a neighborhood selection questionnaire and wore an accelerometer for seven days. Walkability characteristics were an important reason for selecting the current neighborhood and were more important for women, older and less-educated adults, but not for high-walkable neighbor- hood residents. Both in the total sample and in participants with high residential self-selection scores, walkability was positively related to active transportation and objectively measured moderate-to- vigorous PA. Designing walkable neighborhoods may help to increase adults’ PA, even in those for whom walkability is an important criterion when choosing their neighborhood. However, findings from studies with longitudinal and controlled designs are required to provide more strongly causal evidence. & 2011 Elsevier Ltd. All rights reserved. 1. Introduction Several studies have found neighborhood walkability (higher residential density, land use mix and street connectivity) to be positively associated with adults’ physical activity (PA) behaviors, primarily physically active transport (Saelens et al., 2003; Sallis et al., 2009; Van Dyck et al., 2010). However, because almost all studies examining the associations between neighborhood char- acteristics and health outcomes have used observational designs, inherent differences (background characteristics, personal prefer- ences, etc.) between residents of different neighborhoods are likely to bias these associations (Oakes, 2004, 2006). Ideally, studies should use controlled trials when investigating neigh- borhood effects on health outcomes; in absence of relevant control conditions, numerous possibilities for confounding exist. However, because of practical and ethical reasons and since most neighborhood effects are generated by the internal dynamics of neighborhoods and cannot be examined experimentally, observational designs are usually employed (Merlo and Chaix, 2006; Black, 1996), and studies aim to measure and control for potential confounding factors. In environmental health research, ‘residential self-selection’ has been put forward as a possibly important confounder of the positive associations between walkability and PA, but the significance of self-selection in this context remains unclear. Residential self-selection implies that individuals are likely to select their neighborhood according to their lifestyle and personal preferences, so those already active or who want to be active may choose to live in a high-walkable neighborhood (Ewing and Cervero, 2010; Owen et al., 2007). Consequently, higher PA participation among high-walkable neighborhood residents might not be caused directly by walkability characteristics in these neighborhoods; those who attach importance to a physically active lifestyle might self-select a high-walkable neighborhood. Some studies have controlled for residential self-selection in their analyses, resulting in mixed findings ranging from attenua- tion of the associations between walkability and PA to minimal effects on the associations (Cao et al., 2009; Owen et al., 2007; Pinjari et al., 2007; Sallis et al., 2009). One study found residential self-selection to strengthen the relationship between neighbor- hood walkability and PA (Chatman, 2009). Based on this pattern Contents lists available at ScienceDirect journal homepage: www.elsevier.com/locate/healthplace Health & Place 1353-8292/$ - see front matter & 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.healthplace.2011.05.005 n Corresponding author at: Ghent University, Faculty of Medicine and Health Sciences, Department of Movement and Sports Sciences, Watersportlaan 2, B-9000 Ghent, Belgium. Tel.: þ32 9 264 63 23; fax: þ32 9 264 64 84. E-mail address: Delfien.VanDyck@UGent.be (D. Van Dyck). Health & Place 17 (2011) 1011–1014