— Resources for Researchers
GIS and Audits
Introduction
The IPEN goal is to study people from a broad range of socioeconomic strata who live in communities with a broad range of built environment patterns. It is essential to have wide variation in walkability (land use mix, residential density, street connectivity, sidewalks) associated with walking and cycling for transportation. It is also desirable to have good variation in environmental features that may be related to recreational physical activity, such as access to recreational facilities, parks and other recreation promoting environments.
Neighbourhood Selection
Since the goal is to study associations, not to estimate prevalence rates, it is not necessary to have representative population samples. The sampling frame can be a metropolitan area, a region of a country, or a whole country. The sampling goal can be achieved in several ways, depending on circumstances in each country.
A. The preferred study design is the one used by the US (NQLS), Australian (PLACE), and Belgium (BEPAS) studies. Four cells are created by high/low walkability and high/low income or SES. This design allows an examination of the association of walkability and physical activity in different socioeconomic groups. GIS is the preferred approach used to estimate walkability. Where adequate GIS spatial data are not available, high/low walkability neighborhoods can be selected based on input from local experts such as city planners, geographers, or public health officials.
B. Random sample of the population (most applicable when there is a good distribution of SES and land use patterns)
C. Combination of random sampling and oversampling of people who live in rare land use types
D. Other methods to achieve a wide range of SES and land use types. If specific neighborhoods are targeted, there must be a total of at least 8 neighborhoods in the study.
IPEN Adolescent Neighborhood Selection Presentation by Marc A. Adams, presented in Ghent, Belgium 2013.
IPEN Adolescent Neighborhood Selection Presentation — GIS Details, calculating macro built environment measures, by Larry Frank, Jim Chapman, Jared Ulmer, Urban Design 4 Health, Inc., presented via webinar January 23, 2014.
GIS Measures of Built Environment
Geographic Information Systems (GIS) is a useful tool for assisting with measures of the built environment. GIS is used with available spatial and census data in the neighborhood selection process to identify administrative units (e.g. U.S. Census Block Groups, Australian Census Collection Districts, UK Super Output Areas) that best represent high and low walkability and high and low income.
Table 1 shows the walkability variables are calculated using GIS.
| Measure | Definition | Equation |
| Residential density | Number of residential units per residential acre | # of housing units/acres of land in residential use |
| Intersection density | Number of intersections per square kilometer | # of intersections/kilometers |
| Land use mix | Evenness of distribution of building floor area of residential, commercial, and office devlopment | Mix2 |
| Retail FAR (floor area ration) | Ration of retail building floor area to land area | Retail building square footage divided by retail land square footage |
- The first measure included in the walkability index was net residential density; the ratio of residential housing units to the land area devoted to residential use per block group.
- The second measure is or intersection density measured the connectivity of the street network, represented by the ratio between the number of true intersections (3 or more legs) to the land area of the block group in acres. A higher density of intersections corresponds with a more direct path between destinations.
- The third measure was the land use mix, or entropy score, indicating the degree to which a diversity of land use types were present in a block group. For this project, the mix measure considered five land use types: residential, retail (excluding region-serving or “big box” uses of 300,000 square feet or larger), entertainment (including restaurants), office, and institutional/civic (including schools and community institutions). Land use area values were normalized between 0 and 1, with 0 being single use and 1 indicating a completely even distribution of floor area across the 5 uses.
- The fourth component, the retail floor area ratio, was the retail building square footage (from all floors) divided by retail land area (square footage). This variable indicated the density of retail development. The rationale was that a low ratio indicated a retail development likely to have substantial parking, while a high ratio (often times higher than 1) indicated smaller setbacks, and less surface parking; two factors thought to impact pedestrian access. Note that not every country will have data to calculate this variable.
The four calculated values were normalized for each block group using a Z score. For more information on the calculating walkability, please see our manuscript available here.
Example: NQLS Study Neighbourhood Selection
Neighborhoods were selected based on walkability and income.
The walkability of the neighborhoods was established using an index based on:
Walkability = | [(2 x z-intersection density) + (z-net residential density) + (z-retail floor area ratio) + (z-land use mix)] |
U.S. block groups were ranked and divided into deciles based on the normalized walkability index. The top four and bottom four deciles represented “high walkability” and “low walkability” areas.
Similarly, the median household income data for each block group were deciled and categorized into “high income” and “low income”. Household income values less than $15,000 and greater than $150,000 were not included in the deciling process in order to avoid skewing the data with outliers. The second, third, and fourth deciles constituted the “low income” category, and the seventh, eighth, and ninth deciles made up the “high income” category.
Thus, a walkability-income quadrant, shown in table 2, was created. For each participant, therefore, a walkability index score is available as well as a walkability quadrant.
NQLS Study Design: Walkability and income quadrants
| Low Walkability | High Walkability | |
| Low Income | 8 neighbourhoods | 8 neighbourhoods |
| High Income | 8 neighbourhoods | 8 neighbourhoods |
For more information on the NQLS neighborhood selection, please see our manuscripts available here and here.
Individual buffers
While GIS and census data are used in the neighborhood selection process to identify block groups which best represent high and low walkability and high and low income, each participant’s address can also be geocoded and given an individual walkability index score. GIS is also used to calculate environmental variables for individual buffers around participants’ residences. For IPEN, aspects of the environment within 500- and 1000-meter street network buffers around individuals’ residences are investigated further. See the templates section below to assist with this process. We also encourage countries to measure other built environment features not listed here that are expected to be related to physical activity.
We used different methods for creating individual buffers for GIS variables in the IPEN adult and adolescent studies. This paper led by Dr. Frank uses IPEN Adult data to examine the “sausage” buffer method that was then used to create IPEN Adolescent GIS-based variables. Read paper.
IPEN Adolescent Presentation: Creating internationally-comparable built environment variables in GIS for the IPEN studies by Marc A. Adams, presented in Ghent, Belgium 2013.
Templates/Specific
developed a series of documents to guide the GIS decision process. These documents are not meant as a step-by-step guide, but rather provide conceptual guidance by providing operational definitions for environmental variables and a survey of GIS procedures that may differ by GIS analyst. Please see the following document to help guide measurement and decision process:
This document includes the following:
- Template instructions
- Variable naming instructions
- Neighborhood buffers
- Residential density and land use
- Commercial/retail land use
- Civic and institutional land use
- Entertainment land use
- Recreational land use
- Food related and restaurant land use
- Intersection density
- Public transportation
- Private recreation facilities
- Public parks
- 500 meter street network buffers variable names
- 1 km street network buffers variable names
- 500 meter pedestrian enhanced buffers variable names
- 1 km pedestrian enhanced buffers variable names
- Park distance variable names
- Character key for variable names
Audits
Neighborhood audits are not a required part of IPEN, but they may be another way to understand your study area. The following audit tools have been used by the US group, and are in progress with regards to scoring, etc.
Food Audits: Nutrition Environment Measures Survey (NEMS)
Park Audits: Environmental Assessment of Public Recreation Spaces (EAPRS)
Street Audits: Microscale Audit of pedestrian streetscapes (MAPS)
Links
GIS Methods paper for IPEN Adult