Methods: Following a 3-day training session, observers (n 8) were provided with a map of a predetermined walking route constructed through the joining of ten randomly selected street blocks. Then, using an 18-item observation grid, pairs of observers performed ratings of 112 neighborhoods. Resulting observations produced a hierarchically structured data set including 4032 observations nested within observers, which in turn were nested within neighborhoods. Data from the 2001 Canadian census were linked to the neighborhood data.

 

Results: Application of ecometric multilevel modeling analyses showed that once interitem and interobserver variability were statistically controlled, about one third of the variability in observations were at the between-neighborhood level. Reliability estimates were 0.78 for items measuring activity-friendliness, 0.76 for safety, and 0.83 for density of destinations. Assessment of the convergent validity of the instrument identified that safety of the environment was positively associated with neighborhood affluence. Density of destinations was negatively associated with affluence and positively associated with higher proportions of persons in the neighborhood walking to work.

 

Conclusions: The three dimensions of the neighborhood active-living potential measure have good reliability and convergent validity and are able to capture between neighborhood differences. Measurement characteristics would have been difficult to ascertain without the ecometrics methodology.