Several instruments have been developed to measure neighborhood environments for physical activity. These are survey measures in which respondents report on features of their neighborhoods, including both items that are ‘factual’ or mainly objective (e.g., "how long would it take you to walk to the nearest park?") and items that relate to perceptions of one’s neighborhood (e.g., "How satisfied are you with the amount and speed of traffic in your neighborhood?").
Brownson and colleagues (2004) tested the reliability of three of these measures, one of which is described in detail here—the Neighborhood Environment Walkability Scale (NEWS). The NEWS, developed by Saelens and colleagues (2003), is a 66-item instrument that assesses the following neighborhood features hypothesized to be associated with walking:
- Residential density (6 items)
- Proximity to nonresidential land uses (23 items)
- Ease of access to nonresidential uses (7 items)
- Street connectivity (5 items)
- Walking/cycling facilities (5 items)
- Aesthetics (6 items)
- Pedestrian traffic safety (8 items)
- Crime safety (6 items)
Most of the items are assessed with a 4-point Likert scale with 1=strongly disagree and 4=strongly agree. The exceptions measure residential density and land uses. The NEWS is available, along with scoring procedures and detailed information on inter-rater reliability (See Appendix A).
When the NEWS was administered to a national sample of both urban and rural residents through telephone interviews the test-retest reliability across a period of one to three weeks, calculated by a 1-way intraclass correlation coefficient (ICC), was at the moderate level or higher (0.4 to 0.6 agreement) for all major constructs. Reliability was highest for land use mix-diversity (0.93) and lowest for street connectivity (0.41). Detailed information on ICC and % observed agreement are presented in Brownson et al. (2004).
Saelens et al. (2003) tested the reliability and construct validity of the NEWS in two census tracts in San Diego using a mailed, self-administered survey. The two neighborhoods differed in objective measures of walkability. The high walkability neighborhood had a large concentration of restaurants and stores, short blocks with few cul-de-sacs, and both single and multiple family residences. In contrast, the low walkability neighborhood was primarily residential with single-family houses, and had lower street connectivity as characterized by longer blocks, and more cul-de-sacs. Test-retest scores ranged from .63 to .80 for subscales, suggesting good test retest reliability. Construct validity was assessed by comparing the mean scores on the NEWS dimensions between residents of the low and high walkability neighborhoods. The two neighborhoods differed in the expected direction on six of the eight dimensions, thus supporting strong construct validity. Physical activity and obesity rates also differed in the two neighborhoods in the expected direction.