Examining the spatial distribution and relationship between support for policies aimed at active living in transportation and transportation behavior Daniel Fuller Lise Gauvin Yan Kestens
Introduction Researchers examining psychological factors do not often integrate a geographic perspective Potentially related to assumption that these factors are antecedent variables Information could by gained from a geographical approach
Introduction Integration of geography with behavioral medicine encourages multidisciplinary research and practice
Objective Present the application of spatial analysis to a measure of support for policies aimed at active living in transportation(pal-t) and examine the association between the PAL-T and active transportation using two regression approaches
Methods Repeated cross sectional population based samples Random digit dialing Over sampling: 25% in the neighbourhoods were a bicycle share was implemented Design Spring Bike share November Bike share November O X O X O n=2001 n=2502 n=2509
Participants Pooled sample n=6861 (2% missing data) Mean age = 48.7 years Female = 59% Response rate= 35.7% Fuller et al., 2012
Variables PAL-T (Policies for active living in transportation) 17 items measure agreement with policies for Active Living in Transportation Would you say that you 1) completely agree, 2) somewhat agree, 3) somewhat disagree or 4) completely disagree with the following potential changes that municipal or government authorities might implement in your neighborhood. Validated measure of agreement with active living friend policies Fuller et al., 2012
Variables Walking and cycling International Physical Activity Questionnaire Self-reported age & sex Road network Kernel Density of road network in Montreal Land Use Sum of parks, grocery stores, banks, pharmacies, and medical services within a 500m road network buffer
Non spatial analyses Analysis Ordinary least squares (OLS) Geographical analyses Kernel densities Moran's I Getis-Ord General G LISA(Local indicators of spatial association) Geographically weighted regression (GWR)
Kernel Density of PAL-T A smoothly curved surface is fitted over points.
Measures of spatial autocorrelation Moran s I Index=0.34 (p<0.01) Getis-Ord General G G=0.01 (p<0.01) Both measures suggest significant spatial clustering of PAL-T Moran, 1948; Getis & Ord, 1995
Local indicators of spatial association Anselin, 1995
Ordinary Least Squares Walking Coefficient Standard Error p Age -2.20 0.55 0.01* Female -34.16 21.16 0.10 Street Connectivity 0.42 0.68 0.54 Land Use -0.95 2.17 0.66 PAL-T 18.56 21.16 0.38 Intercept 189.76 71.35 0.01* Note: R 2 =1.7%. Dependent variable is total number minutes walked for active transportation in the past week
Ordinary Least Squares Cycling Coefficient Standard Error p Age -0.17 0.08 0.54 Female -17.36 3.00 0.03* Street Connectivity -0.05 0.09 0.53 Land Use 0.63 0.31 0.04* PAL-T 7.45 3.00 0.01* Intercept 13.80 10.13 0.17 Note: Note: R 2 =0.6%. Dependent variable is total number of minutes cycled in the past week
Heteroscedasticity Wiki Commons
Mapping Heteroscedasticity Moran s I for cycling residuals Index=0.01 (p=0.33)
Mapping Heteroscedasticity Brunsdon, 1996
Mapping Heteroscedasticity OLS R 2 =0.6% GWR R 2 =0.8%
Applications ArcGIS: GWR Tool Stata: gwr R: spgwr SAS: PROC MIXED (with x,y coordinates) SPSS: Compute spatial weight in ArcGIS, import into SPSS and run weighted regression GWR3: Stand alone program for GWR
Discussion Integrating a spatial perspective into understanding psychological and behavioural constructs is relevant Interdisciplinary silos Talk to me but make friends with a geographer Fuller et al., 2012
Thanks Daniel Fuller fuller.daniel@gmail.com www.walkabilly.net
References Moran, P.A.P. (1948). The interpretation of statistical maps. Journal of the Royal Statistics Society B, 10: 243-251. Anselin, L. (1995). Local indicators of spatial association LISA. Geographical analysis, 27(2): 93-15. Getis, A. & Ord, J.K. (1995). Local spatial autocorrelation statistics: Distributional issues and an application. Geographical Analysis, 27(2): 287-306. Fuller, D., Gauvin, L., Fournier, M., Kestens, Y., Daniel, M., Morency, P., & Drouin, L. (2012). Internal Consistency, Concurrent Validity, and Discriminant Validity of a Measure of Public Support for Policies for Active Living in Transportation (PAL-T) in a Populationbased Sample of Adults. Journal of Urban Health. Fuller, D., Hobin, E. P., Hystad, P. & Shareck, M. (Joint Authorship) (2012). Challenges to interdisciplinary training for junior space, place and health researchers, Critical Public Health, 22(1), 1-7. Brunsdon, C. A., Fotheringham, S., & Charlton, M. E. (1996). Geographically Weighted Regression: A Method for Exploring Spatial Nonstationarity. Geographical Analysis, 28(4), 281 298.