Table 3.
Neighborhood Level Characteristics in 259 Empirical Quantitative Studies of Neighborhood Effects and Health
| No. of Studies | % of total studies | |
|---|---|---|
| Multiple level of geographies | ||
| 1 level | 243 | 89.58 |
| 2 or more levels | 15 | 9.65 |
| Neighborhood definition | ||
| Census tracts | 135 | 52.12 |
| Block groups | 57 | 22.01 |
| Neighborhood clustersa | 21 | 8.11 |
| ZIP codes | 15 | 5.79 |
| Othersb | 20 | 7.72 |
| More than one definitionc | 8 | 3.09 |
| No description | 3 | 1.16 |
| Is neighborhood geographic vs spatial | ||
| Geographic | 208 | 80.31 |
| Spatial | 14 | 5.41 |
| Both | 37 | 14.29 |
| Is neighborhood variable proximity vs prevalence | ||
| Prevalence | 234 | 90.35 |
| Proximity | 5 | 1.93 |
| Both | 20 | 7.72 |
| Neighborhood level variables | ||
| Census-based aggregated | 112 | 43.24 |
| Survey-based aggregated | 31 | 11.97 |
| Non-aggregatedd | 14 | 5.41 |
| Combinatione | 99 | 38.22 |
| Not reported | 3 | 1.16 |
| Explicit mention of MAUP/UGP | ||
| None | 249 | 96.14 |
| UGP | 2 | 0.77 |
| MAUP | 8 | 3.09 |
Abbreviations: MAUP= modifiable areal unit problem as originally described in Gehlke & Biehl (1934) ; UGP=uncertain geographic context problem as originally described in Kwan (2012); only explicit reference to these terms was acknowledged.
All 21 studies using neighborhood clusters (NCs) as neighborhood definition were analyzing data from Project on Human Development in Chicago Neighborhoods (PHDCN). 847 census tracts in the city of Chicago were collapsed to form 343 NCs that were “ecologically meaningful” (http://www.icpsr.umich.edu/icpsrweb/PHDCN/sampling.jsp).
Other definitions of neighborhoods included primary care service areas, “buffers” or “radius circle”, geopolitical units, and minor civil division.
Studies that used more than one definition of neighborhood tested their main effects models using multiple definitions. 6 studies used census tracts and block groups; 2 studies used census tracts and ZIP codes.
Non-aggregated neighborhood variable refers to truly contextual features of the environment, including number of convenience stores, availability of recreational centers, and air quality.
Combination refers to any mix of census-/survey-based aggregated variables and non-aggregated variables used in the studies as neighborhood level predictors and/or covariates.