National Academies Press: OpenBook
« Previous: Overview and Summary
Page 5
Suggested Citation:"Response by Type of Strategy." National Academies of Sciences, Engineering, and Medicine. 2004. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 9, Transit Scheduling and Frequency. Washington, DC: The National Academies Press. doi: 10.17226/23433.
×
Page 5
Page 6
Suggested Citation:"Response by Type of Strategy." National Academies of Sciences, Engineering, and Medicine. 2004. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 9, Transit Scheduling and Frequency. Washington, DC: The National Academies Press. doi: 10.17226/23433.
×
Page 6
Page 7
Suggested Citation:"Response by Type of Strategy." National Academies of Sciences, Engineering, and Medicine. 2004. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 9, Transit Scheduling and Frequency. Washington, DC: The National Academies Press. doi: 10.17226/23433.
×
Page 7
Page 8
Suggested Citation:"Response by Type of Strategy." National Academies of Sciences, Engineering, and Medicine. 2004. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 9, Transit Scheduling and Frequency. Washington, DC: The National Academies Press. doi: 10.17226/23433.
×
Page 8
Page 9
Suggested Citation:"Response by Type of Strategy." National Academies of Sciences, Engineering, and Medicine. 2004. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 9, Transit Scheduling and Frequency. Washington, DC: The National Academies Press. doi: 10.17226/23433.
×
Page 9
Page 10
Suggested Citation:"Response by Type of Strategy." National Academies of Sciences, Engineering, and Medicine. 2004. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 9, Transit Scheduling and Frequency. Washington, DC: The National Academies Press. doi: 10.17226/23433.
×
Page 10
Page 11
Suggested Citation:"Response by Type of Strategy." National Academies of Sciences, Engineering, and Medicine. 2004. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 9, Transit Scheduling and Frequency. Washington, DC: The National Academies Press. doi: 10.17226/23433.
×
Page 11
Page 12
Suggested Citation:"Response by Type of Strategy." National Academies of Sciences, Engineering, and Medicine. 2004. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 9, Transit Scheduling and Frequency. Washington, DC: The National Academies Press. doi: 10.17226/23433.
×
Page 12
Page 13
Suggested Citation:"Response by Type of Strategy." National Academies of Sciences, Engineering, and Medicine. 2004. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 9, Transit Scheduling and Frequency. Washington, DC: The National Academies Press. doi: 10.17226/23433.
×
Page 13
Page 14
Suggested Citation:"Response by Type of Strategy." National Academies of Sciences, Engineering, and Medicine. 2004. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 9, Transit Scheduling and Frequency. Washington, DC: The National Academies Press. doi: 10.17226/23433.
×
Page 14
Page 15
Suggested Citation:"Response by Type of Strategy." National Academies of Sciences, Engineering, and Medicine. 2004. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 9, Transit Scheduling and Frequency. Washington, DC: The National Academies Press. doi: 10.17226/23433.
×
Page 15
Page 16
Suggested Citation:"Response by Type of Strategy." National Academies of Sciences, Engineering, and Medicine. 2004. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 9, Transit Scheduling and Frequency. Washington, DC: The National Academies Press. doi: 10.17226/23433.
×
Page 16
Page 17
Suggested Citation:"Response by Type of Strategy." National Academies of Sciences, Engineering, and Medicine. 2004. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 9, Transit Scheduling and Frequency. Washington, DC: The National Academies Press. doi: 10.17226/23433.
×
Page 17
Page 18
Suggested Citation:"Response by Type of Strategy." National Academies of Sciences, Engineering, and Medicine. 2004. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 9, Transit Scheduling and Frequency. Washington, DC: The National Academies Press. doi: 10.17226/23433.
×
Page 18
Page 19
Suggested Citation:"Response by Type of Strategy." National Academies of Sciences, Engineering, and Medicine. 2004. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 9, Transit Scheduling and Frequency. Washington, DC: The National Academies Press. doi: 10.17226/23433.
×
Page 19
Page 20
Suggested Citation:"Response by Type of Strategy." National Academies of Sciences, Engineering, and Medicine. 2004. Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 9, Transit Scheduling and Frequency. Washington, DC: The National Academies Press. doi: 10.17226/23433.
×
Page 20

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

9-5 Recent frequency elasticity observations have tended to group around either +0.3 or +1.0. Those that are grouped around +1.0 are suburban systems that have undertaken carefully planned, comprehensive expansion programs, in an atmosphere of good public image and a growing or at least stable economy. A majority but not all of those grouped around +0.3 are central city urban systems. The greatest concern expressed by transit riders is with dependability of service and with midday and evening service frequencies. When service is reliable, passengers make their actual waits at the transit stop less than random arrivals would imply. Waiting times are more severely affected by service irregularities than might be apparent on the basis of operating data averages, and in addition, riders have been shown to be more sensitive to unpredictable delay than predictable time requirements. Easy to remember departure times and readily available schedules appear to be significant contributors to achieving a favorable user perception of the wait for low and medium frequency transit service. Limited but consistent examples of ridership gains in response are reported. Timed transfer service design seems to improve rider satisfaction, but patronage effects are indeterminate given presently available data. Frequency changes affecting individual transit lines typically cause diversion of riders to or from other transit services when alternatives are available, such that the impact on overall transit usage is not as much, and sometimes far less, than the effect on individual route ridership. Frequency and headway elasticities are thus often, in a sense, “inflated” by this phenomenon. On the other hand, the highest observed sensitivities to frequency increases have been in circumstances where diversion from other transit services is not an issue. In business districts significant numbers of people who previously walked may be attracted by frequency improvements. In general, however, one out of every two or three new riders drawn to transit service by frequency improvements would otherwise have driven an auto, as is the case with transit fare reductions. RESPONSE BY TYPE OF STRATEGY Bus Frequency Changes Increased bus frequency normally attracts increased patronage, and vice versa but with wide variation in results. It has been suggested that available observations do not support a single numerical relationship between service frequency and patronage changes (Holland, 1974). Indeed, measured in terms of service quantity, elasticities calculated for the more recently reported frequency changes group either around an elasticity of +0.3 or around +1.0, the threshold of elastic response. Nevertheless, both historical and more recent elasticities of bus service changes exhibit a service elasticity average that is on the order of +0.5. There is not enough information to address whether there are differences in proportionate response between increases and decreases in service, though it happens that none of the highest elasticities reported pertain to service decreases. The substantial variations in reported ridership responses are attributable in part to the widely varying circumstances attending individual bus route and system headway changes. The variables involved include the pre-existing level of transit service, the geographic and

9-6 demographic environment, and the time period of the day or week. There is evidence that some of these variables affect ridership response in a predictable way, especially pre-existing frequencies and time of day (Holland, 1974; Mayworm, Lago and McEnroe, 1980). Complexities are added by the frequent presence of concurrent actions (such as fare changes or service extensions), and by other aspects of the operating environment, examined in the “Underlying Traveler Response Factors” section. (Combined impacts of concurrent actions may provide opportunities, it should be noted, for obtaining desired outcomes.) Another confounding factor is that some ridership changes in response to frequency changes reflect primarily diversion of riders from one route to another (route choice), rather than diversion from one mode to another (mode choice, such as between auto and transit). The sensitivity of overall transit usage to route frequency changes is less than would be indicated by route level elasticities derived where significant shifting among routes has occurred (Miller and Crowley, 1989). Elasticities “inflated” by passengers who merely shifted routes are among those reported in the literature and used in drawing generalizations. Nevertheless, in essentially none of the recent observations of service frequency elasticities in excess of +1.0 is route shifting a significant factor, although there are certainly other influences at work. Historical Data The Mass Transportation Commission of the Commonwealth of Massachusetts performed a variety of mass transit service improvement and fare reduction experiments in the early 1960s that provide what is still the most extensive quasi-experimental data set available on individual transit route frequency change impacts. Full coverage of the experiments is provided in the case study “Mass Transportation Demonstration Projects in Massachusetts.” Mid-point arc headway elasticities calculated from individual Massachusetts demonstration project results (Mass Transportation Commission et al., 1964) are presented in Table 9-1 along with other reported 1960s and 1970s headway elasticity findings (Holland, 1974; Mayworm, Lago and McEnroe, 1980). Note that headway elasticities are negative.3 4 The median headway elasticity among those derived from the Massachusetts experiments is -0.4, or -0.6 omitting depressed urban areas. There are indications that due to data limitations, these elasticities may have somewhat understated the long term potential for ridership gains in the study area.4 The other pre-1980 elasticities reported in Table 9-1 (and Table 9-6 “Fare and Service Elasticities”) average -0.5 (expressed as headway elasticities). 3 The measure “headway elasticity” indicates the percentage change in ridership observed or expected in response to a 1 percent change in the headway. The negative sign indicates that the effect operates in the opposite direction from the cause. Thus a headway elasticity of -0.50, for example, indicates that a 1 percent decrease in headway has caused or is expected to cause a 0.50 percent gain in ridership. (See also footnote 2, under “Overview and Summary” — “Traveler Response Summary.”) Service elasticity and headway elasticity are both used to express the degree of transit ridership response to frequency changes. Those calculated by the authors of this Handbook and presumably all other sources have been derived using arc elasticity formulae that give the same elasticity value (except for sign) for both service — expressed in bus trips, miles or hours — and headway. There are several reasons these elasticities may be somewhat understated. For one thing, the Massachusetts experiments were short — 3 to 12 months in duration. Also, the elasticities were calculated on the basis of revenue, "before" ridership data not having been reported. In cases where smaller fares were changed for shorter trips, there may have been more of a ridership increase than revenue increase, because indications are that service improvements attract proportionately more short trips than long trips.

9-7 Table 9-1 Bus Route or Small System Headway Elasticities Observed in the 1960s/70s Route / Service Territory Headway Elasticity Months After Implementation Massachusetts Demonstrationsa Boston-Milford suburban route (new headway approx. hourly) -0.4 10-12 Uxbridge-Worcester suburban route (new headway hourly) -0.2 7-9 Adams-Williamstown city route (new headway approx. hourly) -0.6 1-3 Pittsfield city route (raised from 3 to 8 round trips daily) -0.7 1-3 Pittsfield city route (raised from 10 to 15 round trips daily) -0.6 1-3 Newburyport-Amesbury (depressed area) city route (new headway 30 min. peak/60 min. midday)b -0.4 6-8 Fall River (depressed area) city service (overall 20 percent service increase) nil 4-6 Fitchburg-Leominster city route (new afternoon headway 10 min., to match morning)b,c -0.3 6-8 Boston downtown distributor, Phase 1 (new midday headway 5 min., to match peak)c -0.8 5-7 Boston downtown distributor, Phase 2 (new headway 4 min. base, 8 min. midday)c -0.6 8-10 Boston rapid transit feeder route (new midday headway 5 min., to match peak)c -0.1 4-6 Other Contemporary Findings Detroit city route (new headway 2 min. peak, 3.5 min. midday) -0.2 — Chesapeake, VA, suburban service (new headway 35 to 42 min.) -0.8 — Stevenage, England (peak period/off-peak; new headway 5 min.) -0.4/-0.3 — —Madison, WI, circulator routes (Saturday/Sunday; new headway 20/30 minutes) -0.2/-0.6 Notes: a Mid-point arc elasticity calculated on the basis of revenue. b Includes impact of minor route extension. c Approximate elasticity computed for full service day by using an unweighted average of peak and off-peak (or morning and afternoon) headway improvements. Sources: Massachusetts Demonstrations — Mass Transportation Commission et al. (1964). Massachusetts elasticity calculations — Pratt, Pedersen and Mather (1977). Other Findings — Holland (1974), Mayworm, Lago and McEnroe (1980).

9-8 Differentiation by Service Level A 1980 exploration of the causes of headway elasticity variations utilized a data set produced from essentially the same case studies as those listed in Table 9-1, but designed to give non- Massachusetts sites somewhat more emphasis. Separate calculations were made, where possible, of peak and off-peak headway elasticities. These, along with all-day elasticities, produced 23 separate mid-point arc elasticity values. The resulting headway elasticity averages, stratified by original bus service level, are listed in Table 9-2. The results clearly indicate a greater sensitivity to frequency changes for cases where the prior service was infrequent. The average headway elasticity for all observations was -0.44, or -0.47 including only those seven observations pertaining to all weekday hours, peak and off-peak (Lago, Mayworm and McEnroe, 1981). (For stratification by time period, see “Temporal Ridership Patterns” under “Related Information and Impacts”.) Table 9-2 Bus Route Headway Elasticities Stratified by Original Service Level Original Service Level (Headway) Number of Observations Arc (Mid-point) Elasticity Standard Deviation Less than 10 minutes 7 -0.22 ±0.10 10 to 50 minutes 6 -0.46 ±0.18 Greater than 50 min. 10 -0.58 ±0.19 All observations 23 -0.44 ±0.22 Source: Lago, Mayworm and McEnroe (1981). More Recent Experience Observations of frequency change results and corresponding arc elasticities from the 1980s and 1990s are summarized in Table 9-3, followed by brief descriptions of selected examples. All but two of these elasticities are computed on the basis of service quantity rather than headway, and thus have positive rather than negative signs, but are otherwise comparable to the elasticities in Tables 9-1 and 9-2. The elasticities reported in Table 9-3 (along with post-1980 bus elasticities from Table 9-6 “Fare and Service Elasticities”) average slightly above +0.5 (expressed as a service elasticity).5 The results for systemwide evaluations are of special interest. They tend to reflect change in overall transit usage, without being confounded by route-specific effects, which may reflect shifts from one route to another without a corresponding change in transit mode share. The Santa Clarita and Charlottesville examples are perhaps the most free of confounding route choice effects, although the Santa Clarita example does have service hours enhancements mixed with the frequency improvements. 5 Updated indication that elasticities for individual routes with intermediate to infrequent service tend toward the upper values of the normal range is offered by Dallas Area Rapid Transit (DART) crosstown route service elasticities in the range of +0.9 to +1.0 for both peak and off-peak frequency increases (Hufstedler, 2004). The type of route may be an additional factor in the rider response.

9-9 Table 9-3 Bus Service Elasticities for Frequency Changes Observed in the 1980s/90s Transit System or Route Time Span Headway Change (Minutes) Service Measure Arc Elasticity Notes and Comments Vancouver, WA to Portland, OR 1980 Mixed, e.g., 19-23 to 10-15; AM peak Peak buses +0.33 (all hours) See description below Charlottesville [VA] Transit System 1980-1981 From 60 to 30 in peak periods Vehicle miles +0.33 (all hours) See description below Mt. Pleasant bus route, Toronto, ON Sept.- Nov. 1987 From 10 to 15 in peak periods and 15 to 30 evening Headway -0.47 pk. -0.29 off- peak See description below and case study Tasta to central Stavanger, Norway early 1990s From 30 to 15 Headway -0.26 Headway measure gives negative sign Santa Clarita [CA] Transit (local fixed route system) 1992/93 - 1997/98 Primarily 60 to 30 with service hours enhancements Service (bus) hours +1.14 (all hours) See description below and case study Foothill Transit, L.A., CA (system) 1993-96 Various, plus new weekend service Service hours +1.03 (all hours) Frequency upped on all lines Community Transit (Snohomish County system, WA) 1994-96 Primarily 60 to 30 plus new services as well Service hours Over +1.0 (see notes) Confounding factors include U of W “U-Pass” introduction Santa Monica, CA Big Blue Bus system 1996-98 Various, plus some new service Service hours +0.82 (all hours) See description below Lincoln Blvd. route Santa Monica, CA March - Sept. 1998 20 to 10 (40 to 10 on link to LAX) Service hours +0.97 6AM-6PM; see description below Note: Elasticities are log arc formulation, except Toronto is mid-point arc. Sources: Vancouver, WA — Public Technologies (Sept., 1980); Charlottesville, VA — SG Associates and Transportation Behavior Consultants (1982); Toronto — Miller and Crowley (1989); Norway — Lunden (1993); Santa Clarita — Kilcoyne (1998a and b) and Santa Clarita Transit (1993-1998); Foothill Transit and Community Transit — Stanley, 1998; Santa Monica — Catoe (1998); all elasticity calculations except Toronto by Handbook authors. Individual Examples The more recent U.S. service elasticity experience with a frequency emphasis begins with a Service and Methods Demonstration (SMD) project in effect from late 1979 to mid-1980. To promote transit on the route connecting Vancouver, Washington, with downtown Portland, Oregon, the Vancouver transit operator decreased average headway during peak periods and extended operating hours. Starting in February 1980, the number of morning buses was increased from 10 to 14, decreasing the headway of 19 to 23 minutes to between 10 and 15 minutes. The number of afternoon buses was increased from 6 to 15, and the hours of service were expanded from 6:18 PM to 9:33 PM. Service was extended along two branches to provide a feeder component. Daily ridership increased from 1,400 to over 1,700 (Public Technologies, Sept., 1980). Attributing the ridership increase to the added number of buses, the Handbook authors calculate a resulting log arc service elasticity of +0.33.

9-10 A system that has focused local bus service expansion primarily on frequency and service hours enhancements is Santa Clarita Transit, serving outlying suburbs of Los Angeles. Local service revenue hours were increased by 66 percent and miles by 99 percent in the five years from FY 1992-93 through FY 1997-98. Service improvements, accompanied by limited route adjustments and extensions, featured expanded weekday and Saturday service hours, addition of Sunday service, and effectively a doubling of frequencies on a majority of routes. The affected routes originally had only hourly service with some 30-minute combinations. Local ridership growth, 120 percent, has exceeded the service growth. The corresponding bus hours log arc elasticity is +1.14, and the bus miles elasticity is even higher (Kilcoyne, 1998a and b; Santa Clarita Transit, 1993-1998; elasticity calculations by Handbook authors). Population growth was modest during this period. The case study “Frequency and Service Hours Enhancements in Santa Clarita, California” provides further background and details. Another Los Angeles area system undertaking frequency enhancement is Santa Monica Municipal Bus Lines. In March 1998 the “Big Blue Bus Line” upped the 6:00 AM to 6:00 PM frequency on its Lincoln Boulevard route, which connects the Los Angeles International Airport (LAX) with downtown Santa Monica, from a 20 to a 10 minute headway. Simultaneously, frequency on the relatively new route extension connecting with the LAMTA cross-county Green Line Light Rail was upped from a 40 to a 10 minute headway. Peak and midday route performance was 66.1 boardings per service hour before the changes. This performance statistic was already back up to 64.5 after five months, equivalent to a service elasticity of approximately +0.97. The Lincoln Boulevard route has benefited from diversion of ridership, via the Green Line, from LAMTA bus services connecting Los Angeles with oceanfront communities. It has also been the beneficiary of new travel agency advertisements identifying the Lincoln Boulevard route as a means of getting from LAX to Santa Monica and area attractions. Rather than being an anomaly, the Lincoln Boulevard improvements are part of a Big Blue Bus Line expansion that has increased service by 23 percent systemwide since 1996. Guided by public input, and a goal of system simplicity, this service increase has been about 90 percent frequency enhancements and 10 percent routing adjustments. Boardings per service hour were 65 in 1996 and 63 in 1998, indicating a log arc service elasticity of +0.82. The response to service expansion is thought to have been enhanced by a major image building campaign and benefited from a rebounding local economy (Catoe, 1998; elasticity calculations by Handbook authors). A contrasting experience is offered by Charlottesville, Virginia. There the hourly service frequency was doubled in peak periods, extensive route restructuring was undertaken, and two new routes were added. While daily vehicle miles increased 110 percent, ridership went up a modest 28 percent over a one year period, exhibiting a service miles log arc elasticity of +0.33. The failure of ridership to increase in proportion to service was ascribed to a largely fixed market consisting primarily of transit dependents. However, it was also reported that the 6 to 21 year old buses were unreliable and that for several restructured routes, the original pattern had generated better ridership. Service was returned to an hourly headway and a design close to the original configuration (SG Associates and Transportation Behavior Consultants, 1982; elasticity calculation by Handbook authors). A panel survey of transit riders in suburban Toronto was used to study response to changes in headway on the Mt. Pleasant Road trolleybus route. Peak-period mid-point arc headway elasticities for the route were determined to be higher in this case (-0.47) then off-peak headway elasticities (-0.29) (Miller and Crowley, 1989). This result — higher peak than off-peak elasticities — is not typical. It may reflect the circumstance that the “off-peak” service reduction involved

9-11 For additional estimates of service elasticities, based primarily on time-series data, refer to the “Frequency Changes with Fare Changes” subsection. Sensitivity Indicators It may be concluded that response to bus service frequency improvements tends to be greatest when the prior frequency was less than three buses or so per hour (Pratt and Bevis, 1971), when the route involved serves middle and upper income areas (Holland, 1974), when the travel market involved is predominantly comprised of trips short enough that walking is an option, and when other factors are favorable (see “Underlying Traveler Response Factors” — “Physical, Operating and Economic Environment”). The response to service frequency changes is apparently least when the service modifications primarily affect lower income areas, when the prior service was relatively frequent, and when the travel market served is characterized by long trips. Train Frequency Changes Aside from providing new facilities or lower fares, fixed rail systems are for the most part restricted to scheduling and frequency changes as a form of service improvement. The available quasi-experimental data on passenger response are mostly in the realm of commuter rail operation. Described in terms of the factors identified above as influencing response to bus frequency changes, commuter rail lines typically serve middle and upper income areas. Although they have relatively long time intervals between trains, they also predominantly serve long trips. Thus an average or somewhat above average response to service changes might be expected if there is a correlation between bus and commuter rail service impact. Commuter Rail Demonstrations Listed in Table 9-4 are the ridership impacts of demonstration project service changes in three Northeast applications. Marketing activities were involved in all cases, as were certain off-peak fare incentives in the Boston experiments. Fares were increased in the Philadelphia demonstration (Mass Transportation Commission et al., 1964; Southeastern Pennsylvania Transportation Authority, 1971). The computed service elasticities range from +0.5 to +0.9, which indeed reflect average to above average sensitivity to service levels. (Further detail on the Boston experiences is provided in the case study “Mass Transportation Demonstration Projects in Massachusetts.”) In the Philadelphia demonstration, average trip length increased by 5.8 percent (Southeastern Pennsylvania Transportation Authority, 1971), resulting in a mid-point arc elasticity with respect to passengers miles of +1.6. One interpretation is that long commuter rail trips may be more sensitive to service levels than shorter trips; another is that the longer trips may have involved travel on services with poorer initial frequencies. evening service only. Midday service was unchanged and omitted from the elasticity calculations. Of more import is the shifting among transit routes demonstrated by this study (see both “Mode Shifts and Sources of New Ridership” under “Related Information and Impacts” and the case study “Mt. Pleasant Bus Route Service Reduction in Toronto — Panel Survey” for further information).

9-12 Table 9-4 Commuter Rail Demonstration Project Impacts and Overall Service Elasticities Location Railroad Demonstration Phase Increase in Service Increase in Ridership Implied Arc Elasticity Philadelphia Reading Final 9.2% 8.6% +0.9 Boston Boston & Maine 2 77% 37.5% +0.6 Boston New Haven 2 26% 11.5% +0.5 Note: Mid-point arc elasticities; calculated disregarding effects of fare changes and marketing. Sources: Philadelphia Demonstration — Southeastern Pennsylvania Transportation Authority (1971); Massachusetts Demonstrations — Mass Transportation Commission et al. (1964); elasticity calculations — Pratt, Pedersen and Mather (1977). The longer commuter rail lines in Boston were likewise associated with greater traveler response to headway changes than the shorter lines.6 In Boston, it was also specifically observed that the ridership response was greater for the lines with the poorer pre-demonstration service levels (Mayworm, Lago and McEnroe, 1980). Table 9-5 presents a summary by original service level of commuter-rail elasticities estimated from the five-corridor demonstration in the Boston area in 1962-64 (Lago, Mayworm and McEnroe, 1981). Table 9-5 Individual Commuter Rail Service Elasticities from the Boston Area Demonstration Original Service Level (Headway) Number of Observations Arc (Mid-point) Elasticity Standard Deviation 10 to 50 minutes 11 -0.41 ±0.13 Greater than 50 minutes 4 -0.76 ±0.10 All observations 15 -0.50 ±0.20 Source: Lago, Mayworm and McEnroe (1981). Results obtained in the New York City area, although not directly comparable, appear to be consistent with the primary Philadelphia and Boston findings (Tri-State, 1966). Overall, these data tend to suggest that commuter rail patronage responses to frequency changes are in the same general realm as bus ridership responses on routes with similar demographics and original service frequencies. Rail Rapid Transit (Metro) In contrast to commuter rail, time series based estimates by London Transport indicate that rail rapid transit, in the instance of the London Underground, has a lower sensitivity to frequency changes than bus. As presented under “Frequency Changes with Fare Changes,” the 6 These limited observations are not in direct conflict with the apparently greater sensitivity of short versus long bus trips to headway changes. Very short trips via bus are an alternate to the walk mode and this is not the case with any normal length commuter rail trip.

9-13 Underground exhibits a miles operated service elasticity of +0.08, just under half that for London buses (London Transport, 1993). This general relationship is as would be expected, given the much higher overall service levels typical of rail rapid transit. This being only one observation, however, it provides insufficient evidence to safely generalize that rail rapid transit service elasticities will average on the order of half those for bus frequency changes, even though a comparable conclusion can reasonably be reached for fare elasticities. Service Hours Changes Service hours changes are quite distinct from frequency changes, but their effect is not often identified separately. For example, a significant part of improvements undertaken by Santa Clarita Transit in California from 1992 to 1998 consisted of service hours expansions; later weekday and Saturday operating hours and addition of Sunday service. Yet frequency enhancements were a larger part of the added bus hours of service. The one impact assessment conclusion that can be reasonably drawn is that both types of actions must have contributed substantially to the outstanding ridership response, reflected in a service elasticity of +1.14 (see “Bus Frequency Changes” and the case study “Frequency and Service Hours Enhancements in Santa Clarita, California”). Extended evening service may, on peak-period-only commuter routes, consist of as little as one trip added after the evening rush hour to serve stragglers. A classic example was documented in the early days of bus service to the new town of Reston, Virginia. A bus was added in 1970 to pick up late passengers in downtown Washington between 7:00 and 7:26 PM. Ridership on the bus varied between 15 and 20 passengers per trip, but more than 80 new riders were attracted to the system. These riders needed the assurance that they would not be stranded at their workplace by a late meeting or other delay (Furniss, 1977).7 7 Two newer instances have been reported of general ridership increase upon introducing evening or weekend service. Whatcom Transportation Authority obtained a significant increase in a time of static ridership by introducing a single evening route connecting Western Washington University with other major generators in its Bellingham service area (Elmore-Yalch, 1998). In Dallas a group of suburban shuttles exhibited a discernible weekday ridership increase, to 4,400 weekday riders in total, in response to introduction of Saturday service that carried 1,400 riders (Hufstedler, 2004). One successful NJT example was bus Route 59, connecting Newark and Elizabeth and extending to the smaller city of Plainfield through wealthier suburbs. Saturday service hours were expanded and Sunday service, discontinued over two decades before, was restored with hourly headways between 8:00 AM and 6:00 PM. After two years the route was attracting some 1,100 boardings on a typical Sunday, compared to 5,700 on weekdays and 3,100 on Saturdays. About 45 passengers per Sunday one-way trip were served at a farebox recovery ratio of 46.8 percent. Another successful example involved commuter rail service on the Main/Bergen County line. Two trips were added on Saturday, two more were extended, and six round trips were added to Sunday service. The annual ridership for this additional weekend service was 73,473 after two years, with a farebox recovery ratio of 52 percent (Michael Baker et al., 1997). Additional perspective is provided by a package of suburban transit service enhancements initiated in 1994-95 by New Jersey Transit (NJT). Out of 40 projects, including 15 involving expansion or introduction of evening and weekend service, 23 were retained after the trial period. The success rate for evening and weekend service enhancements was well above the 40-project average.

9-14 Frequency Changes with Fare Changes Frequency Versus Fare Sensitivities Results of urban transit frequency changes implemented in connection with fare changes suggest that either type of change may have the greater impact depending on circumstances. Statistical analysis covering two years of fare and service changes in greater Dallas revealed greater sensitivity to fares than service in the center city, and the converse in the suburbs, for both suburban express and local services (Allen, 1991). The added ridership attracted by an experimental bus frequency increase of approximately 25 percent in Fitchburg, Massachusetts, was effectively nullified by a 25 percent fare increase (Mass Transportation Commission et al., 1964). Additional background and findings on the Dallas fare and service changes are provided in the case study “Fare and Frequency Changes in Metropolitan Dallas.” The direct comparisons between observed fare and service elasticities shown in Table 9-6 for Dallas, San Diego and London, developed using time series data, are thought to reflect primarily service frequency adjustments as contrasted to routing and coverage changes. An exception is “San Diego...(all bus routes),” which is shown for comparison. Table 9-6 Fare and Service Elasticities for Dallas, San Diego and London Fare Elasticity Service Elasticity Service Measure Dallas (1985-1987) urban bus (DTS) suburban express bus suburban local -0.35 -0.26 -0.25 + 0.32 + 0.38 + 0.36 bus revenue miles San Diego (1972-1975) (all bus routes) established bus routes (-0.51) -0.67 (+0.85) + 0.65 bus miles London (1971-1990) bus Underground (Metro) -0.35 -0.17 + 0.18 + 0.08 operated miles Sources: Allen (1991); Goodman, Green and Beesley (1977); London Transport (1993). When results for frequency changes with fare changes are taken in conjunction with other frequency, fare and service change results, additional conclusions may be inferred. Ridership appears likely to be more sensitive to fare changes than frequency changes where frequency levels are high. Conversely, response to service changes is almost always greater than to fare changes of similar magnitude where service levels are low and especially when new routing, coverage or express service is involved. (See Chapter 10, “Bus Routing and Coverage,” under “Response by Type of Service and Strategy” — “Service Changes with Fare Changes,” and also Chapter 4, “Busways, BRT and Express Bus” — “Underlying Traveler Response Factors” — “Service Coverage and Frequency.”)

9-15 Mutually Reinforcing Fare and Frequency Changes Fare increases together with service reductions obviously lead to ridership loss at the same time as they offer cost savings potential. In the District of Columbia, institution of a 25¢ Metrorail to bus transfer charge and an increase of approximately 70 percent in elderly and disabled and other reduced fares, along with service reductions, led to a bus ridership decline of 11 percent on weekdays and 14 percent on weekends averaged over the first 2 full months. Corresponding bus revenues were up by 6 percent on weekdays, but down 3 percent on weekends. Two-thirds of the bus ridership loss was attributed to the service reductions, which included route eliminations and consolidations in addition to frequency reductions (Washington Metropolitan Area Transit Authority, 1995). Dallas fare increases approaching 50 percent, coupled in the same year with a 16 percent decrease in service, mostly frequency reductions, were accompanied by a 16.5 percent ridership loss and a 20 percent revenue gain. Part of the loss was attributed to a local economic downturn. Three previous years of service increases, initiated with a 29 percent base fare reduction, had afforded almost a 50 percent ridership gain (Allen, 1991). (See also the case study “Fare and Frequency Changes in Metropolitan Dallas.”) Commuter Rail Indications are that the typical commuter railroad patron is much more influenced by service frequency than by fares, although findings are not entirely consistent. The first phase of 1960’s era Boston & Maine demonstrations included both fare decreases (28 percent) and service increases (77 percent). Overall Phase 1 patronage rose 27 percent, but the increase on two individual lines which received only fare reductions was a mere three percent. Although most fares were raised in Phase 2, ridership continued upward. The experience on Boston area lines of the New Haven Railroad was comparable (Mass Transportation Commission et al., 1964), and an 11 percent fare increase as part of the Philadelphia area Reading Company demonstration similarly failed to erase positive patronage response to service frequency improvement (Southeastern Pennsylvania Transportation Authority, 1971). On the other hand, cross-sectional model adjustment based on time series data from Maryland’s MARC Brunswick line suggested that 1993-94 log arc fare elasticity may have been on the order of -0.70, 25 percent higher than the modeled sensitivity to a frequency improvement focused outside the peak. This is a corridor with highly competitive travel options. Evidence from the other two MARC lines was inconclusive (Parsons Brinckerhoff et al., 1994, unpublished worksheets). Combined Service Frequencies Some transit service improvement actions involve deployment of buses to serve a given street or closely defined corridor in an operating mode differing from the pre-existing or alternative service. Overlaying express bus routes on existing local routes is an example. In such cases it cannot strictly be said that the frequency of service has been changed in proportion to the new bus runs, and some riders may not benefit from the new service. Other riders, however, obtain increased options with additional amenities such as express speed.

9-16 Express Service Options In situations where the provision of new or expanded express bus service has resulted in increased overall frequency of service from residential areas to the central business district (CBD), ridership increases have exhibited service elasticities on the order of +0.9. These findings suggest that where express service is appropriate, a combination of increased service and express runs may attract additional patronage — possibly half again as much — as would a similar bus trip increase applied to local service alone. Further detail on frequency changes with express service is contained in Chapter 4, “Busways, BRT and Express Bus.” Transfer Versus No Transfer When differing services are coordinated to provide a useful combined frequency, some passengers appear governed in the choice of their transit trip by the departure and arrival times, and others appear governed by the other characteristics of the service offered. In rural England, a study was made of local transit travel under circumstances of combined frequency. Riders were offered hourly service alternating between a through trip and a trip requiring one transfer. If departure/arrival time governed, 50 percent of the riders would be expected to use the transfer service. If other trip characteristics governed, none of the riders would be expected to use it. In actual practice 24 percent elected to use the service requiring the transfer (Tebb, 1977). Regularized Schedule Minimizing Passenger Wait Times With the right kind of systematic, easy to remember and well-advertised bus schedule, effects similar to those in rail might be possible to engender (Pratt and Bevis, 1971). Hard information on actual response to provision of easily remembered departure times is extremely scarce, although anecdotal evidence is reported of appreciable gains in ridership when schedules have been reorganized to give simple “clockface” timings, for example, where buses always arrive at 10 minutes, 30 minutes and 50 minutes after each hour (Webster and Bly, 1980). In Oslo, Norway, surveyed riders were found willing to accept longer journey times to avoid transfers. Regular riders indicated willingness to accept 8-10 minutes more journey time or to pay NOK 2.25 (about $0.33 at the time) in order to avoid switching to a waiting vehicle. In cases where a 5-minute wait for the next connection was required, passengers were found willing to accept a 14 minute increase in journey time or to pay NOK 4.00 to avoid the transfer (Stangeby, 1993). A number of travel demand analyses have shown that while the average wait for local, often irregularly scheduled bus service can be adequately described for travel estimation purposes as one-half the headway, the average wait for commuter rail service cannot (Parsons Brinckerhoff et al., 1994). The wait for commuter trains is apparently perceived by the potential commuter as being some lesser amount. Readily available schedules and long-term dependability of service, allowing one to minimize wait at the station, are presumably major factors in this favorable perception of commuter rail scheduling. (This phenomenon is further discussed in Chapter 8, “Commuter Rail.” See the “Underlying Traveler Response Factors” section in particular.)

9-17 It is notable that many successful restructurings of small city bus service and midday commuter service have employed “clockface” scheduling as one aspect of the overall design (Dueker and Stoner, 1972; Dueker and Stoner, 1971; Mass Transportation Commission et al., 1964; Tri-State, 1966). The case study “A Combined Program of Improvements with Fare Changes in Iowa City,” in Chapter 10, “Bus Routing and Coverage,” describes an example. A documented case involving Omnitrans in Riverside, California, entailed both route and schedule restructuring. The restructuring was accomplished in the Fall of 1995 within the constraint that total bus service hours not be increased by more than 4 percent. Ridership increased by 20.4 percent over the prior year. Route restructuring focused on enhancing direct travel. The schedule restructuring emphasized consistency and ease of transfer, in addition to providing increased frequency on heavily traveled routes within the service hours constraint. All schedules were standardized to be on 15-, 30- or 60-minute on-the-hour headways (Stanley, 1998). Minimizing Transfer Times Transfer centers are a popular means of facilitating suburban and smaller city transit service as well as making transfers between routes more convenient. While transfer centers can make it easier to institute scheduling enhancements such as coordinated transfers, they are often created for other reasons. In a survey conducted by the Institute of Transportation Engineers of 10 transit transfer centers throughout the United States, only 3 indicated that increasing ridership was a primary objective of the facility. Common objectives were to provide a rest area for operators, enhance the public ’s image of transit, provide a civic facility, aid downtown development or revitalization, provide riders with protection from weather and a better waiting environment, reduce the potential for accidents, and enhance passenger convenience. Half the centers reported that they had no impact on transit ridership, while the other half had positive ridership impacts (Hocking, 1990). To serve fringe areas in a timed-transfer system, a trunk line generally operates with a regular service frequency throughout the day and connects with local timed-transfer lines at a transit center located in the suburban community. This technique eliminates the need to dedicate transit equipment of each suburban route to the costly run between the suburban center and downtown. In smaller cities all routes may be local timed-transfer routes focused on a downtown center and perhaps one or two other activity nodes. The timed-transfer especially benefits passengers who must use more than one bus line to complete their trips. While the presence of a transfer center may make it easier to operate coordinated transfer schedules, also known as timed-transfers, it is the interplay between route design and scheduling that is crucial. The timed-transfer concept utilizes timed connections at a point where routes are focused in order to minimize the wait time and irregularity involved in the transfer between lines. The connecting transit routes must be designed within route running time parameters that facilitate timed-transfer scheduling. Route length, traffic conditions and passenger activity determine run time, and run time determines ability to make a complete bus trip and still maintain timed-transfer meets and bus layover time requirements. 8 For an example and discussion of a strong favorable response to doubling an evening peak period “clockface” feeder bus schedule to match the “clockface” schedule of the rail line served, for benefit of riders returning home, see Chapter 7, “Light Rail Transit,” under “Related Information and Impacts” — “Mode of Access and Egress to Rail Service” — “Feeder Service Effects.” 8

9-18 Timed-Transfer Findings In Portland, Oregon’s Westside community, two transit centers were used as part of a network redesign. A timed-transfer system was successfully implemented in the summer of 1979. Departure times from the transit centers were consistent throughout each day. A high degree of service reliability could be maintained, and schedule efficiency was improved. Ridership in both the peak and off-peak periods increased significantly. By the spring of 1980, daily ridership had increased 40 percent to 13,808. The new service influenced travel patterns. Local trips and non- work trips accounted for the largest increases. In certain areas, local trips increased by 138 percent, and non-work trips increased by 68 percent. Travel to downtown Portland increased by 12 percent. However, it is important to note that the 1979 gasoline shortage occurred during the changes (Kyte, Stanley and Gleason, 1982; Charles River Associates, 1997). In other studies of timed-transfer networks, direct ridership impacts were less apparent. The Urban Mass Transit Administration, in 1983, reviewed the design and cost effectiveness of timed- transfer networks in Ann Arbor, Michigan, and Boulder, Colorado. Large increases in unlinked trips (bus boardings) for the systems were found. However, the study could not determine the extent to which the increases were caused by actual new ridership as compared to the increased transfer boardings inherent in certain timed-transfer designs (Newman, Bebendorf and McNally, 1983). In a study of the Tidewater region in Norfolk, Virginia, improvements in the perceptions of riders were found to be the principal impact of the implementation of a timed-transfer system. From 1989 to 1991, an elaborate multiple hub system was put in place to reduce the required operating subsidies. The resulting service had between two and six routes meeting at a location. Between 40 and 45 percent of bus trips involved a transfer. Of surveyed riders, the majority felt service quality was improved with the implementation of the timed-transfer system, 77 percent felt schedules had improved, and 71 percent experienced decreased travel times. Over two-thirds thought the reliability of service increased. A decrease in ridership was attributed to several factors unrelated to timed transfer, including fluctuation in the resident military population, so it was difficult to determine the ridership response to the timed-transfer system (Charles River Associates, 1997; Rosenbloom, 1998). Transit Reliability Changes A service improvement even more fundamental than schedule enhancement is the achievement of reliability, so that whatever schedules are established are adhered to. Unreliable transit service may result from either environmental factors alone, or in combination with inherent factors. Environmental factors include fluctuating traffic conditions, traffic signals, variations in boarding/alighting demand and availability of drivers and vehicles. Inherent factors aggravate initial deviations from scheduled headways. Platooning, for example, results when late vehicles encounter increased passenger loads at subsequent stops, producing additional delay, while following or early vehicles encounter decreased loads, causing them to be further ahead. Dependable service avoids the reductions in effective frequency that accrue from missed runs, platooning of vehicles, and other unplanned deviations from schedules (Abkowitz, 1978). Attitudinal studies of commuters in Baltimore and Philadelphia early on found “arrival at intended time” to be perceived as the second most important travel attribute for work trips. Only “arrival without accident” was judged by respondents to be more important out of over 35

9-19 attributes listed. Similar surveys in Boston and Chicago placed “arrival at intended time” above travel time, waiting time and cost measures. For non-work trips reliability was judged not as important, although it still ranked eighth on the list (Golob et al., 1970; Paine et al., 1967). Effects on Wait Time Increased reliability results in actual transit vehicle arrival times occurring in a tighter distribution around the scheduled time. The range of actual vehicle arrival times at the beginning and end of a trip, and at transfer points, determines the wait time, the overall travel time and the likelihood of missed connections and late arrivals that a rider faces. Maintenance of on-time service has a positive effect on riders and ridership because patrons experience less waiting, decreased travel time, fewer missed connections, more on-time arrivals at their destinations, and reduced uncertainty overall. Waiting times, even for a frequent service, are affected more substantially by service irregularities than the average headway achieved would indicate. Passengers of frequent services arrive more or less continually at the transit stop. Consequently, a larger number of passengers are adversely affected by long unscheduled gaps between buses and trains than are benefited by corresponding short gaps. Table 9-7 lists the percentage of passenger wait time in excess of the optimum Table 9-7 Reliability Impacts on Wait Time for Individual New York City Bus Routes NYCTA Bus Route Waiting Time Index Wait in Excess of Optimum (%) B46 +72% M7 72 B35 61 M4 54 BX41 47 M3 45 M16 52 M2 39 Q32 47 M34 47 M11 30 BX55 26 BX28 23 M79 22 BX30 0.58 0.58 0.62 0.65 0.68 0.69 0.66 0.72 0.68 0.68 0.77 0.79 0.81 0.82 0.95 5 Note: The Waiting Time Index is the minimum average wait (assuming passengers arrive without reference to the schedule), divided by the actual average wait (calculated using the same assumption). The Wait in Excess of Optimum is the actual average wait less the minimum average wait, divided by the minimum average wait, and expressed as a percentage. Source: N.Y. State Office of the Inspector General for the MTA as graphed in Henderson, Kwong and Atkins (1991), with excess wait calculations by the Handbook authors.

9-20 Schedule reliability is in fact demonstrated to save regular commuters even more time than the assumption of random passenger arrivals at the transit stop would indicate. A study of ten bus stops in London found that where bus arrival times were consistent, passenger waiting times tended to be less than that expected based on random arrivals. Passengers were benefiting by setting their arrival time to coincide with bus arrival times. Where service was inconsistent, waiting times more nearly approximated times based on random arrivals (Jolliffe and Hutchinson, 1975). Table 9-8 lists transit and passenger statistics for the bus stops with the most reliable and least reliable service of the 10 examined. Table 9-8 Observed London Bus Headway Reliability and Passenger Wait Times Scheduled Headway Observed Headway Standard Deviation Waiting Time for Random Arrivals Observed Waiting Time Stop with most reliable service 23.0 2.2 12.9 5.8 Stop with least reliable service 20.3 23.9 23.5 10.7 14.0 13.1 Source: Abkowitz et al. (1978). Other work suggests an even greater effect if vehicle-miles are lost from an otherwise perfectly reliable service. On high frequency services, if 10 percent of the buses are cut randomly, average passenger waiting time will increase by 20 percent. For services with long headways, an even larger effect is predicted. Since passengers tend to schedule their arrival especially for infrequent services, a missing bus means waiting an entire extra headway interval (Webster and Bly, 1980). Effects on Ridership In general, the effects on ridership of lack of reliability will be even more pronounced than the increase in waiting time alone indicates. This effect is attributable to the uncertainty about if and when the next vehicle will arrive and consequent anxiety and annoyance to passengers. London Transport has estimated that elasticities with respect to “unplanned” service cuts (i.e., lost vehicle-miles) are some 33 percent larger than with respect to scheduled service cuts (Webster and Bly, 1980). Periodic equipment failures during initial operation of the BART rapid rail system in San Francisco led to public perceptions of undependability and are thought to have inhibited ridership in the early years (Peat, Marwick, Mitchell, 1975). Virginia Railway Express (VRE) commuter rail service encountered severe reliability problems caused by track congestion and related delays after a July 1996 freight train derailment affecting both VRE lines. The aftermath of the derailment caused chronic delays for weeks, along with individual train cancellations. Riders were alienated despite a liberal ticket refund policy. At the same time a new set of commuting options was coming available with the opening of a Metrorail station and carpool lane extensions. In the months following the incident, VRE experienced a 32 percent decrease in ridership. For the year, although VRE had originally projected growth in achievable with full schedule adherence, for 15 New York City Transit Authority bus routes. The passenger wait time is calculated on the basis of actual bus arrivals assuming random passenger arrivals (Henderson, Kwong and Atkins, 1991).

Next: Underlying Traveler Response Factors »
Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 9, Transit Scheduling and Frequency Get This Book
×
 Traveler Response to Transportation System Changes Handbook, Third Edition: Chapter 9, Transit Scheduling and Frequency
Buy Paperback | $20.00
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

TRB’s Transit Cooperative Research Program (TCRP) Report 95: Chapter 9 – Transit Scheduling and Frequency examines scheduling changes made to conventional bus and rail transit, including changes in the frequency of service, hours of service, structuring of schedules, and schedule reliability.

The Traveler Response to Transportation System Changes Handbook consists of these Chapter 1 introductory materials and 15 stand-alone published topic area chapters. Each topic area chapter provides traveler response findings including supportive information and interpretation, and also includes case studies and a bibliography consisting of the references utilized as sources.

READ FREE ONLINE

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!