Purpose: Develop Tool to Estimate Demand for Rural Intercity Bus Services

2. Purpose: Develop Tool to Estimate Demand for Rural Intercity Bus Services Goals Included: • Easy to Use • Not Requiring Access to Specialized Data or Software • Appropriate for Rural Intercity Bus Services • Sensitive to Variety of Factors Potentially Affecting Demand

3. Methodology: • Literature Review • Gather Data to Identify Rural Intercity Bus Services: • Survey State DOT’s to Identify Funded Operators • Contact or Survey Operators to Obtain Data on Ridership, Operating Characteristics • Classify Services • Develop Full Data for Each Service • Points Served • Frequency • Route Length • Population • Demographic Data • Service to Key Destinations (Major Medical, Airports, Passenger Rail, Correctional Facilities, Universities) • Connectivity to the Intercity Bus Network - Loudoun County, VA, - Charles County, MD, - Prince William County, VA, - Outer Banks, NC , - Rutland VT, - Springfield, VT, - Nampa/Caldwell, ID - Williamsburg, VA

4. Methodology (continued): • Classification Revised Following Panel Meeting • Population Alternatives: • Municipal populations at stops • Urbanized Area/Census Designated Place population • Test Relationships Among Variables • Develop Statistical Models—Regression • Develop Alternative Models Using Data from the National Personal Transportation Survey • Develop User-Friendly Toolkit

5. Data on Services: • Classified as: • Regional: Not operated by an intercity bus carrier, part of the national network (local fares, information not national) • ICB: Operated by an intercity carrier that is part of the national network (allows for interline ticketing and information) • Key Issue: Definition of Intercity • Not all Section 5311(f) funded routes included • Connectivity to National Intercity Network a key characteristic used in classifying services • Non-intercity services deleted (commuter characteristics, long rural transit routes with no connectivity, etc.) • Many issues with data: • Ridership (one full year, for this particular route/service) • Service Characteristics (round-trips per week) • Fares (Cents per mile calculated from end-to-end fare)

6. ResultingDatabase: • 135 routes identified in survey • 120 routes with basic data (ridership) • 57 routes that met final definition of intercity and had complete data

7. Population Data: • 2000 Census • Use of Census-designated areas, rather than GIS • Initially used Municipal populations for each stop • Summed for the route • Issues led to use of Urbanized Area/Census Designated Place populations • Population of Destination/Origin in Major Metro Area—how to address?

8. Resulting Models: • Regression Model • Trip Rate Model

9. Regression Model: • Basic Model: Annual Ridership = -2,803.536 + 0.194(Average Origin Population) + 314.734(the number of stops on the route) + 4971.668(yes to airport service/connections) + 5783.653(yes to service provided by an intercity provider) R2 = 0.712, Adjusted R2 = 0.690 All variable significant a 5% level or better • Signs are plausible • Shows positive impact of connectivity to air and national network • Use of average population and number of stops positively related to ridership

10. Trip Rate Model: • Used data from National Household Travel Survey • Base Data was Long Distance Trips (50 miles or more one-way) • By Urban and Rural • By Region (Census Divisions) • By Income Group: • Under $30,000 • Under $75,000 • Over $75,000 • Converted to a Per Capita rate for Rural Trips, by region • Bus Mode Share of .09% rounded to 1% produced best estimates

11. Accuracy of Models: • Trip Rate with 1% Mode Share: • Within 50% of actual ridership 45.60% • Within 10% of actual ridership 14.00% • Adjusted 1% Trip Rate Model (used regression to predict error terms, which were then subtracted from Trip Rate predictions): • Within 50% of actual ridership 54.40% • Within 10% of actual ridership 15.80% • Regression Model: • Within 50% of actual ridership 59.60% • Within 10% of actual ridership 17.50%

12. Limitations: • Continued need for judgment in application of the models: • Which estimate to use • Diligence in entering plausible data • Deciding whether or not a service is operated by an “Intercity Bus Carrier”, or serves an Airport if some type of transit connection is required • Neither model is sensitive to changes in fare or frequency • Neither model includes overhead or through ridership that might result from being part of a network

13. Potential Next Steps: • More data, try again • Stop level models—predict ridership at a single stop • Further efforts to include fare and frequency effects • Impact of terminals and park and ride • Include demand tools in an overall intercity planning and procedures guidebook • Develop a national network model to include network effects (may be more important as Section 5311(f) is used to fill network gaps, rather than replace dead-end branches)

14. Toolkit Development: • Simple and User-Friendly • Not Requiring GIS Software • Self-Contained—Includes Data • Includes Both Models • Provides Reference to Ridership on Comparable Routes • Allows for User Adjustments to Reflect Special Conditions

15. Toolkit: • Product is a CD, Requires Excel • All Directions and References are Included • Includes Introduction, Application Steps • Example of Application • Includes All Required Census Data

16. Model Usage: • Insert disk (requires Excel) • Initial screen has links to text on applicability, background, and steps in use • Second screen is for input—use pre-loaded Census data to input route, check boxes for service attributes • Press input button, third screen gives ridership estimates for each model • Third screen also gives route characteristics and ridership for the four most comparable routes in the database • Fourth screen allows for manual adjustments on Trip Rate model to account for availability of other intercity bus service at points on the route in question, or presence of a major university or other generator

17. Demonstration: Western Maryland • Imagine that you are a planner working in the State of Maryland • A statewide intercity bus needs study has identified the existing network, and compared it to areas of the state with transit needs and likely key destinations • A gap in service is identified in far western Maryland, and so you want to see if there might be sufficient demand to support a Section 5311(f) project • You have the TCRP Project B -37 Toolkit disc, which you insert in your computer, it opens to the following page:

18. Demonstration (continued): Home Page

19. Demonstration (continued): Home Page • You click on the link and read about how the Toolkit was developed. • You click on the link and read about the applicability of the Toolkit. • You click on the link and read about the steps in applying the Toolkit. • The first steps involve developing a proposed route to test. • Identify existing services in the area of concern • Develop a proposed route and identify the potential stops

20. Demonstration of Toolkit: Identify Existing Service

21. Demonstration (continued): • Your goal is to connect the eastern and western rural areas of the state with the populated center, and connect to the frequent intercity bus services in the Washington-New York corridor—so the service will need to connect to Baltimore • Points identified for this possible service include: • Baltimore • Frederick • Hagerstown • Hancock • Cumberland • Frostburg • Grantsville • Morgantown, West Virginia • You click on the button to go to the input page

22. Demonstration of Toolkit: Define Proposed Route

23. Demonstration (continued): Input Page • Most of the route will be in Maryland, so you enter “Maryland” in the blank for state. • You enter the stops for the proposed route. You can begin by typing the place name, and in the blank it will auto-fill with the place name. Or you can scroll through the list to find each stop. • The Toolkit finds the population for each stop you identify. • In this case, you are anticipating that the proposed route will be operated by an intercity bus carrier, so you check that box. You are not planning to serve an airport, so you leave that blank. You are not aware of federal or state correctional facilities at any of the stops on the route, so you leave that blank. You need to enter the one-way route length, so you jump to a mapping site, and enter the route to obtain the mileage. It is best to check that output to make sure that the route length includes all the proposed stops. • You click on the button to obtain the results, shown on the next page.

24. Demonstration (continued): Input Page

25. Demonstration (continued): Outputs Page • On this page you find information restating the input parameters you entered previously. • Below that are the results of the two demand models: • For the Regression model, there is an estimate of route level demand accompanied by some information about the 95% Confidence Interval for that estimate. This gives you some information about the potential range around that estimate. • For the Trip Rate Model there is also an estimate of demand. • Below the demand estimates is a box labeled “Comparable Routes” with ridership and service characteristics for the four routes in the Toolkit database that are most similar to the proposed route. This may validate your estimates, or raise questions. More information is available for each of the comparable routes by clicking on the link.

26. Demonstration (continued): Output Page

27. Demonstration (continued): Manual Adjustment Page • Below the Trip Rate Model is a box stating that adjustments can be made to the trip rate model results. You click on that, and read the information provided at the link. • A new “Manual Adjustment” page opens. On this page you find the estimated demand for each stop on the proposed route, as estimated by the Trip Rate model. There is no demand estimated for the largest population point, as it is assumed to have a significant amount of additional bus service already.

28. Demonstration (continued): Adjustment Page

29. Demonstration (continued): Manual Adjustment Page • Following the directions, you make changes in the estimated demand for several of the stops. • For Frederick and Hagerstown you find out that the proposed service would only be a fourth of the total service, and so it is likely to capture only a fourth of the predicted ridership. You enter the new demand numbers in the blanks on the “Manual Adjustment” page. • In Frostburg there is a major state university that is likely to increase the demand. You adjust the ridership up and enter the new figure. • In Morgantown there is both a university that will increase ridership, and other intercity bus service, necessitating a downward adjustment in the estimate. You enter the new estimated figure. • At the top of the “Manually Adjusted Trip Rate Demand” column a new total appears, summing the adjusted demand estimates by stop. • As can be seen, the manual adjustments can make a significant difference in the estimated ridership. This information may also be used to develop alternative revenue estimates, which may be needed to develop project budgets.

30. Demonstration (continued): Adjustment Page

31. For Further Information: • To obtain the Toolkit and/or the Final Report: Transit Cooperative Research Program (TCRP) website: • http://www.tcrponline.org/publications_home.shtml • To contact the study team: • Fred Fravel or Reyes Barboza KFH Group, Inc. 4920 Elm Street, Suite 350 Bethesda, MD 20814 • Voice: 301-951-8660 Fax: 301-951-0026 • Email: ffravel@KFHGroup.com or rbarboza@KFHGroup.com