Methodology

The Bicycle Network Analysis (BNA) is data analysis software that measures how well the bike network in a given area connects people to the places they want to go, automating the otherwise labor-intensive task of network analysis. The BNA produces network analysis results for each census block in a designated area, as well as an overall score for the area. The software is freely available for anyone to run from source code via the Brokenspoke Analyzer.

Methodology

The BNA relies on the concept of a low-stress bike network as developed by Maaza Mekuria, Peter Furth, and Hilary Nixon at the Mineta Transportation Institute and approximately aligns with NACTO street and bikeway design standards. The BNA's low-stress determination generally translates to a Level of Traffic Stress 1 or 2 rating on the Mineta Transportation Institute's original scale. In practical terms, this corresponds with the comfort level of a typical adult with an interest in riding a bicycle but who is concerned about interactions with motor vehicles.

The BNA scores cities in four steps:

Data collection
Traffic stress analysis
Destination access analysis
Score aggregation

Each step is described below. Click through the tabs to learn more about our process.

Data Collection
Traffic Stress Analysis
Destination Access Analysis + Score Aggregation

The types and sources of data used in the BNA are listed in Table 1.

Table 1. Data Types and Sources in the Bicycle Network Analysis

Data Type	Country
Data Type	United States	Australia	Canada	All other international
Bike + street infrastructure	OpenStreetMap via Geofabrik
Destinations	OpenStreetMap via Geofabrik
City boundary	U.S. Census TIGER/Line Files or individual submissions	Australian Bureau of Statistics	Statistics Canada	OpenStreetMap via Overpass Turbo or individual submissions
Default residential speed limit	Based on state law or individual submissions	25 mph (40 km/h) or individual submissions
Unit of analysis	Census block	Statistical area level 1	Dissemination area	1 km square
Population	2010 Decennial Census	2021 Census	2021 Census of Population	2020 WorldPop (unconstrained, UN adjusted)
Jobs	LEHD Origin-Destination Employment Statistics (except U.S. territories, where jobs are excluded)	Excluded

Although most places in the BNA database are cities, the BNA can measure any area within its computational limits. For the purposes of this methodology, we will refer to the measured area as a city. Take caution when comparing BNA results between cities with different population data sources, as smaller geographic units increase the stringency of the analysis.

The BNA obtains information on bike infrastructure, street, and intersection characteristics as well as the location of common destinations from OpenStreetMap. OpenStreetMap employs a system of key-value pairs called 'tags' to represent metadata about map features. For a comprehensive review of the tags recognized in the BNA, please refer to our tagging guidelines. When any needed information is missing from OpenStreetMap, the BNA makes assumptions about street characteristics based on the functional class of the street in OpenStreetMap as described in Tables 2, 3, and 4.

Table 2. Street Segment Assumptions

» For two-way streets, each side is evaluated separately

Functional class	Speed	Number of lanes	Parking	Parking lane width	Roadway width
Primary	40 mph	2	Yes	8 ft	N/A
Secondary	40 mph	2	Yes	8 ft	N/A
Tertiary	30 mph	1	Yes	8 ft	N/A
Unclassified	25 mph	1	Yes	N/A	27 ft
Residential	25 mph	1	Yes	N/A	27 ft

Table 3. Bicycle Facility Minimum Width Assumptions

» Gutter does not count towards bike lane width

Bike Infrastructure	Multi-Use Path	Buffered Bike Lane	Bike Lane (with parking)	Bike Lane (no parking)
Width	8 ft	6 ft	5 ft	4 ft

Table 4. Intersection Signal Assumptions

» Uncontrolled intersections assume a low-stress crossing for travel along the higher order roadway

Functional Class	Primary	Secondary	Tertiary	Residential
Primary	Yes	Yes	No	No
Secondary	Yes	Yes	No	No
Tertiary	No	No	Yes	No
Residential	No	No	No	No

The BNA downloads OpenStreetMap data and census data then clips it to match the city's boundary plus a buffer distance around the boundary equivalent to the default trip distance, which is 2,680 meters or 1.67 miles, a distance roughly equivalent to a person biking 10 minutes at 10 mph.