1. Mode Choice Survey Results

2. Utility Models Variables U(Walking) = 1.977 + 7.049(ShortDistance)
U(Bicycle)
3.797(ShortDistance) (AutoOwn) (BikeOwn) – 1.849(Gender)
U(Bus)
6.391 – 0.471(Gender)
U(HOV)
4.303 – 0.254(Gender) (ArrivalTime)
U(SOV)
(BikeOwn)
Variables
Of note:
Only 3 SOV drivers so it is difficult to get anything to be significant
Walking was the strongest indicator
One person who put down SOV indicated they did not own a car (did they steal it?)
ShortDistance =
Distance from school indicator (1 if < 1 mile, o otherwise)
AutoOwn
Car ownership indicator (1 if yes, 0 if no)
BikeOwn
Bike ownership indicator (1 if yes, 0 if no)
Gender
Gender indicator (1 if female, 0 if male)
ArrivalTime
Time arrived on campus in hours (e.g. 9am = 9)

3. Compute Probabilities
For example:

4. My Results

5. Mode Choice Survey Results
Discussion Points:
Weather wasn’t significant
Because it was raining everywhere
How might we structure the survey differently to make it significant? (ask on different days)
Why is gender significant?
Could be picking up something else not asked
Women were less likely to bike, bus or HOV
Maybe most women live close to campus so this is a surrogate for distance
Why the underpredicting of bicycle and SOV?
Only a few observations and unless they have some common characteristic the model won’t be able to predict well
Most often when bicycle had a high probability, walking was slightly higher
Didn’t ask a question that would discern between walkers and bikers
Didn’t capture the deciding factor for SOV (predictions were for 1 bus and 2 HOV)
Why is walking prediction spot-on?
Only one independent variable: distance. This is likely picking up people’s income, living style, convenience factors – they live close to campus
Some of these variables are picking up important unknown variables (e.g., income, disposable income, etc.)
Problems
Interpretation of questions
Incorrect reporting
Reporting what they would like to be rather than what they actually did
Unknown significant variables
The error term
CEE 320 Winter 2005
32 Correct Predictions