Peddling Progress: Greenway Expansions and Increasing Multimodal Commute Behaviors

Term Project for Travel Behavioral Informatics Spring 2024 by Enola Ma and Hannah Bonestroo

1. Peddling Progress: Greenway
Expansions and Increasing
Multimodal Commute Behaviors
Hannah Bonestroo and Enola Ma
04.30.24

2. 2
Outline
●Motivation
●Research Question
●Method
○Survey
○Models
●Result
●Limitations and Next Steps

3. 3
Motivation
●Cycling is a healthy, affordable, and environmentally friendly mode choice
●Also a good mode for first/ last-mile connection
●The count of frequent bike users in the city is on the rise
○But the pace of safe cycling infrastructure expansion is not matching this
●Recently the Adams administration proposed to expand/ establish greenways
○Greenways are a type of bike infrastructure isolated from motor vehicles,
function similarly as protected bike lanes

4. 4
Greenway Example
Image Source: Curbed

5. 5
City’s
Expansion
Plan
Image Source: NYCDOT

6. 6
City’s
Expansion
Plan
Image Source: NYCDOT

7. 7
City’s
Expansion
Plan
Image Source: NYCDOT

8. 8
Current View of East New York Ave
Image Source: Google Maps

9. 9
Research Question
●How does the proximity and accessibility of greenway corridors affect the likelihood
of individuals incorporating cycling into their daily commute?
●Will bike users prefer protected bike lanes over traditional bike lanes?
●How can policymakers leverage insights from commuter behavior and preferences to
optimize the design and management of greenway networks for maximum social and
economic impact?

10. 10
Method: Survey
50 useable responses

Key Questions:
●Current commute mode?
●Do you own a bike?

Issues:
●Did not ask demographic questions such as income
●Heavily skewed toward NYU students and bike commuters
●Only 1 participant in target location

11. 11
Method: Survey (Stated Preference Scenario 1)
Imagine you live in Brownsville (A) and work in Downtown Brooklyn (B).

12. 12
Method: Survey (Stated Preference Scenario 1)
Option A:
Bike (more direct)
Option B:
Bike (better protected)
Option C:
Transit
Option D:
Car
Time 21 minutes 24 minutes 25 minutes (14 minutes walking
+ 11 minutes subway)
Drive 26
minutes
Cost $0.00 $0.00 $2.90 $0.35
DetailsBike Michael Griffin St
(unprotected bike lane) to
Bergen St (unprotected bike
lane) to Flatbush Ave (no bike
lane)
Bike Eastern Parkway (first
mile is shared bike lane and
remainder is protected bike
lane) to Flatbush Ave (no
bike lane)
Walk 10 minutes to Rockaway
Ave Station, take C train to
Lafayette Ave Station, walk 4
minutes to destination
Drive 26
minutes

13. 13
Method: Survey (Stated Preference Scenario 1)
Option A

Option B

14. 14
Method: Survey (Stated Preference Scenario 1)
New Option B (protected bike lane has been constructed on East New York Ave)

15. 15
Method: Survey (Stated Preference Scenario 2)
Imagine you live in Brownsville (A) and you work in Union Square in Manhattan (B).

16. 16
Method: Survey (Stated Preference Scenario 2)
Option A:
Transit (C to 5 trains)
Option B: Transit
(L train)
Option C:
Multimodal
(Bike to L train)
Option D:
Other (Bike, Car, Ride
Hail Service Etc)
Time41 minutes 43 minutes 31 minutes N/A
Cost $2.90 $2.90 $2.90 N/A
DetailsWalk 10 minutes to Rockaway Ave
station, Ride C train 20 minutes to Fulton
St station, transfer to 5 train and ride 5
train for 5 min to 14 St-Union Sq station,
walk 1 minute to destination
Walk 16 minutes to
Broadway Junction
station, ride L 25
minutes to 14
St-Union Sq station
Bike 5 minutes to
Broadway Junction
station, ride L 25
minutes to 14
St-Union Sq
station
Any option that is not
transit such as biking,
driving, or ride hail
service
Protected bike lane has now been added to East New York Ave.

17. 17
Method: Survey (Stated Preference Scenario 2)
Option C: Multimodal (Bike 5 minutes to Broadway Junction station to L train)

18. 18
Survey Results
Do you own a bike?

What is your ideal
commute mode?

No

Yes

Public Transit
(short walk)

Bike

Long walk +
Public Transit

Bike + Public
Transit

19. 19
Survey Results- Scenario 1
Bike 1

Bike 2

Transit

Car

20. 20
Survey Results- Scenario 2
Transit 1

Transit 2

Transit 3

Other

21. 21
Model - Scenario 1 Model - Scenario 1
The utility U
ij
that an individual i derives from choosing alternative j:


For an MNL model, the deterministic part of the utility, which is a linear combination of the observed
attributes, can be specified as:





Our model was as follows:


where Bike_lane is a dummy variable set to 1 only when the choice was Bike2 for the second choice
response, Cost (the cost of the transit option in dollars), Mode_2 (indicating current mode choice: 1 if
biking, 0 if other), and Bike (indicating bike ownership: 1 if yes, 0 if no)

22. 22
Model - Scenario 1
McFadden R^2
●Constant Model: 0.1487
●Null Model: 0.2979

23. 23
Results - Scenario 1 - Interpretation
●The significant positive coefficient for Bike_lane indicates that the
presence of a bike lane increases the likelihood of being chosen
●Only public transit and car modes had costs; the significant positive
coefficient for cost could indicate some other factor influencing choice
for these modes is missing
●The significant negative coefficient for bike ownership for Bike1 and
Car indicates that owning a bike increases the likelihood of choosing
Route 2 over these other options

24. 24
Results - Scenario 1 - Compensating Variation
●Mean Compensating Variation:
○Bike1 = -0.35
○Bike2 = -.49
○Car = -0.069
○Transit = -0.43

25. 25
Results - Scenario 1 - Predicted Probability
Mode Predicted Probability (without
protected bike lane)
Predicted Probability (with
protected bike lane)
Bike2 0.294 0.371
Bike1 0.276 0.244
Car 0.070 0.064
Transit 0.360 0.321

26. 26
Model - Scenario 2
McFadden R^2
●Constant Model: 0.234
●Null Model: 0.259

27. 27
Results - Scenario 2 - Compensating Variation
●Mean Compensating Variation:
○Transit 1 = -0.505
○Transit 2 = -0.256
○Transit 3 = -0.359
○Car = -0.229

28. 28
Results - Scenario 2 - Predicted Probability
Mode Predicted Probability (without
protected bike lane)
Predicted Probability (with
protected bike lane)
Transit 3 0.130 0.261
Transit 1 0.411 0.353
Transit 2 0.267 0.223
Car 0.191 0.163

29. 29
Scenario 1 - Extended
●Cost of $0.35 used for the car option was probably much lower than the actual cost of
using a car if considering parking fees
●Reran the model with a hypothetical increased parking cost of $12 to reflect parking
and gas costs combined
○This yielded insignificant results, so we experimented with duplicating our
dataset for hypothetical results, focusing on Scenario 1

30. 30
Model - Scenario 1 - Extended
McFadden R^2
●Constant Model: 0.15327
●Null Model: 0.30163

31. 31
Results - Scenario 1 Extended-
Compensating Variation
●Mean Compensating Variation:
○Bike1 = $2.78
○Bike2 = $3.75
○Car = $2.43
○Transit = $0.25

32. 32
Results - Scenario 1 Extended - Elasticity
●Elasticity of 0.2569 indicates that a 1% increase in the "Bike_lane" attribute
results in an approximately 0.26% increase in the probability of selecting
Bike2 as the transportation mode

33. 33
Conclusions
●Proximity (accessibility) to bike lanes does increase bike usage for commute
purposes
●Protected bike lanes become the preferred choice when available
○Build more protected bike lanes
●Increase bike accessibility to further promote bike usage

34. 34
Limitations and Next Steps
●Could have better researched bike routes - Flatbush Ave
●More samples, especially from target location
●The large presence of bikers out of survey takers led to biased results
●Could have asked more detailed demographic data
●Future work may try other models such as nested logit and hybrid choice

35. 35
Q&A
Thank you!
Enola Ma - [email protected]
Hannah Bonestroo - [email protected]