The Relationship between Bicycle Accidents
And Lanes of Travel at Downtown Ann Arbor Intersections
Overview
This is an exploration into how the “complexity”, or number of lanes, of an intersection varies with bicycle accidents. This considers bicycle accidents with motorized vehicles in an urban setting.
Research Question
What is the effect of intersection complexity on bicycle accidents?
Hypothesis
Intersections where bicycles and automobiles cross-paths the most have the most accidents
Null: Number of lanes crossed has no effect on the number of accidents at that intersection.
Results On one particular street, the number of lanes crossed is statistically significant, at 0.014. Thus, the null hypothesis is rejected.
Unit/Analysis Bicycle Intersection on the Packard St.
Data required
Location of accidents: Whether accidents occurred in intersections or alleys.
Number of lanes (and bike lanes), one-way street or two-way streets.
Sources
Study of traffic accidents involving bicycles (From Prof. Nystuen)
Intersection survey using GIS or overhead photographs
Methodology
View bicycle-accident map: specifies where bicycle accidents occurred.
Select the intersections to study. These were selected along the same street, in one of the denser parts of Ann Arbor. The number of accidents that occurred from 1983 to 1992 were tabulated at each of those intersections. A regression analysis was performed on that section. 

The fact that all accidents were on one road, within the densely populated part of town, should help reduce the variability of bicycle traffic amongst intersections. Also, automobile-counts along Packard was similar at either end of the study area. 

A brief survey was conducted at each intersection:

  • The number of lanes of traffic in each direction.
  • The angle of the intersection. (If intersection was 6-way, the angles were averaged.)
  • Dummy variables included a designated parking area, whether the intersection was a “T”, if those cars made the lane get narrower.
Determine if there is a correlation between number of vehicle-lanes on the roadway and the number of accidents. These lanes can be counted in different combinations: the total, the maximum number in any given direction, etc. If no correlation exists, determine if there is a correlation to the dummy variables: presence of bike-lanes, obtuse-angle intersection, side-alleys, or one-way streets.
Results
Although there was a correlation between the number of lanes to cross and bicycle accidents, two dummy variables had the most statistical significance: whether the intersection was a “T”, and whether the cars and bicycles had to share a narrower lane, (including the cars parked nearby).






Click here to see the interactive map.

Bike Accidents 794 bike accidents happened in Ann Arbor from 1983 to 1992.

Buffers are around every intersection and the radius is 0.02 mile. (The radius on the above map is 0.04 mile to show buffers better on the web.)

Bike Lanes
 
 
Bike Lane: On-road travel lane designated for the exclusive use of bicycles.

Designated Bike Route: Road designated by green “Bike Route” sign, which has either wide outside curb lane or low traffic volume

Suggested Bike Route: Road which has either wide outside curb lane or low traffic volume

Shared-use Path: Off-road paved path for non-motorized use

Sidewalk: Off-road facility bordering a roadway 

Suggested by the Park and Recreational Department of City of Ann Arbor, in June 2001





 
 
 
 


Spring(Mar, Apr, & May): 162 accidents, Summer(Jun, Jul, & Aug): 276 accidents,

Fall(Sep, Oct, & Nov): 282 accidents, and Winter(Dec, Jan, & Feb): 43 accidents



A Case Study



Focus Area


 

Packard Street, near Downtown Ann Arbor

 

Situation
Mostly Residential-Rental
Renter-Density appears high; (highest between Division & South University) 

2-lane road with center line 

Marked bike-lanes in both directions 

Cuts diagonally through part of a city grid 

Similar 2-way 24-hour motorized traffic count on either end (15,839 vs.15,615)

Looking at intersections only, although __ people had accidents in-between accidents when the vehicle-lane narrowed, versus __ people when the vehicle lane was normal.

 

Influence of Intersection-Design Variables on Packard Street
Between Division Str. and E. Stadium Blvd.
Bicycle Accident data from 1983 to 1992

Variable
Beta, or STRENGTH of the variable
"t" from t-test (LARGER means MORE STATISTICALLY SIGNIFICANT)
Significance (Probability that effect is due to random variation)
Lanes Crossed
0.054
0.162
0.873
T-Intersection
-3.392
-3.266
0.005
ANGLE
-0.009
-0.549
0.590
Vehicle Lane Narrows
2.451
1.301
0.211
Parking (Northbound)
1.136
0.895
0.383



Note: When considering only T-intersection and “Vehicle Lane Narrows”, R-square=0.69


Other Possibilities
 
Weather data: 

History of Road Construction, Pavement, Improvement, etc. 

Literature Research
Northeast Ann Arbor Transportation Plan, Non-motorized Component 

Successful Bicycle Planning: 

Adapting Lessons from Communities with High Bicycle Use 

to Ann Arbor and Washtenaw County 

http://www-personal.umich.edu/~jlevine/downloads/bikereport.pdf
 

Bibliography
Longley, Paul A. 2001. Geographic Information Systems and Science. England: John Wiley & Sons, Ltd. 
Davis, Bruce E. 2001. GIS: A Visual Approach. Canada: OnWord Press. 

Hewitt, Bridgett. 1997. Southeast Michigan Transportation Profile. SEMCOG. 

Bishop, Theresa and Robert Newhouser. 1993. 2015 Regional Transportation Plan for Southeast Michigan. SEMCOG. 

 
Home
Top