A Closer Look at Public Transportation Mode Share Trends

A Closer Look at Public Transportation Mode Share Trends



Recent releases of census transportation information, American Housing Survey results, the National Household Travel Survey (NHTS), American Public Transportation Association ridership statistics, and Federal Highway Administration vehicle-miles of travel data provide opportunities for researchers and policy analysts to glean information on travel behavior trends in the United States. Several data sources, specifically the NHTS, shed light on changes in transit use and mode share trends at the national level. This paper looks at transit mode share trends with both field count and survey data results.

The research indicates that unlinked transit trips declined in the early 1990s followed by ridership growth through 2001, at which point ridership began declining again before rising in 2004. It is clear that transit has grown in terms of total trips, and its overall mode share has stabilized. As overall national travel growth has slowed, transit use appears to be fluctuating between positive and negative growth in terms of both absolute trips and transit's share of overall travel. The research also identifies the shortcomings and differences of the various data sources for determining transit use and mode share trends.

KEYWORDS: NHTS, transit ridership, mode share, commuting.


Mode share and transit ridership trends are relevant to a number of policy deliberations. Policies and investments are often designed to increase transit ridership or mode share, and subsequent measures of ridership provide feedback on market response. Arguments are often made linking mode share trends and public funding for transit. This link between funding and trends in transit ridership has been presented in several different forms by both advocates and critics of various initiatives to fund public transportation. During the recent reauthorization of the federal surface transportation program, transit supporters, for example, used increases in transit ridership during the later half of the 1990s as a reason to support increased federal funding in this area (STPP 2002). Opponents, on the other hand, used the continued decline in the mode share of transit from the decennial census as a reason for reducing federal funding for transit (Cox and Utt 2002). Perceptions of transit ridership levels and trends can influence funding levels, research priorities, and investment decisions at all levels of government (Urban Mobility Corp. 2002).

Developing a clear understanding of what is actually occurring regarding transit use trends is highly dependent on what is measured and reported. Critical issues include:

  • Is the focus on absolute ridership or transit mode share?
  • Is the unit of measurement unlinked trips, linked trips, or passenger-miles?
  • Is the data source observation (count data) or respondent-stated (survey data)?
    • For survey data, is the definition of use "actual" or "usual" mode?
    • For survey data, how large is the sample and what biases might exist?
    • For count data, how accurate and comprehensive is the measure?
  • Are the sampling errors and nonsampling errors such that confidence can be placed in the estimates?

Many researchers have studied the trend in transit ridership and mode share, and others have studied factors and policies that influence trends in transit ridership and mode share. Much of the previous literature and policy debates used a single measurement of ridership. One exception is Pisarski (2003), who compares and contrasts the recent trend in transit mode share using information from the National Transit Database versus information from the decennial census. The research presented here goes beyond Pisarski (2003) by providing an overall look at the recent trend in transit's modal share. This paper contributes to the literature and the policy debates by comparing a variety of data sources and measurements of transit's mode share. In addition to presenting a comprehensive perspective of transit use trends, the paper comments on both the quality of the data and the implications of the trends in the context of ongoing policy deliberations regarding transit funding.

Releases of new data, including census transportation information, the American Housing Survey results, the National Household Travel Survey (NHTS), and regular updates to American Public Transportation Association (APTA) ridership statistics and Federal Highway Administration (FHWA) vehicle-miles of travel (VMT), provide opportunities for researchers, policy analysts, and others to glean information regarding travel behavior trends in the United States. This paper reviews various sources of data, including both survey data results and field count data, from which one can develop estimates of public transportation mode share trends. The analysis provides a richer understanding of mode share trends as well as insight into issues associated with the relationships between the various data sources. Most studies of mode share and transit ridership trends are motivated by a desire to understand causal factors underlying ridership (Joint Center for Political Studies 1985; TRB 2001; Millar 1999; Mason 1998).

In the era of the Intermodal Surface Transportation Efficiency Act of 1991, a plethora of studies targeted strategies for enhancing ridership (Taylor and McCullough 1998; Kain and Liu 1999; Stanley 1998; Project for Public Spaces 1999; Urbitran Associates 1999; Taylor and Haas 2002; Norman 2003; Schmidt 2001). All these initiatives benefit from a rich understanding of ridership and mode share trends.


Count-Based Measures

Figure 1 illustrates the reported transit ridership expressed as annual national total ridership on public transit. These trends are drawn from two data sources: APTA,1 which receives quarterly vehicle boarding counts from members that are factored into a national total; and the National Transit Database (NTD),2 which gathers annual sampled counts of ridership reported to the Federal Transit Administration by agencies receiving federal funds.

The numbers reported show meaningful positive increases in transit ridership of approximately 22% between 1995 and 2001. The 2002 and 2003 APTA data show a reversal of the trend as the economy slowed and related fare and service changes resulted in declines in ridership in those years. Data for 2004 indicate a recovery in ridership to 2001 and 2002 levels. For 2002, NTD data indicate a very slight increase in ridership and for 2003 show a decline similar to that shown by APTA. Both of these sources report measures of persons boarding transit vehicles (called unlinked trips). If a person has to board two or more vehicles to complete a trip to a destination, this is defined as a linked trip. Both data sources are subject to errors associated with farebox data-collection methods and neither contains the full universe of transit operators. However, they represent the best available aggregate count data and reasonable sources for understanding industry trends.

Figure 2 shows the most recent data on overall travel trends as measured in percentage change in VMT. The trend shows a declining growth rate over the past several years. The data, based on FHWA reporting of VMT through 2004, include total urban and national VMT. Urban VMT rates of change go from being higher than national totals, indicating a growing share of total VMT in urban areas, to a situation where total VMT outpaced urban VMT, indicating more rapid growth in non-urban areas. In both cases, the pace of VMT growth has clearly slowed.

Figure 3 displays the relative rates of change for VMT and transit ridership. More rapid rates of change for transit ridership indicate times when transit is likely to gain market share (assuming constant length transit trips, because the figure compares transit trips against vehicle-miles for auto). Based on this indicator, transit was losing market share between 1990 and 1995, gaining share from 1996 through approximately 2000, and subsequently losing share in more recent years.

Figure 4 indicates changes in person-miles of travel (PMT) for auto and transit, using an estimated measure of PMT. Transit PMT is estimated by multiplying trips measured by APTA by an average transit trip length developed yearly from NTD data. Auto VMT are converted into PMT by factoring VMT by vehicle occupancy. Vehicle occupancy uses NHTS data and interpolates between survey years. This enables the development of a measure of mode share that compares person-miles for privately operated vehicles versus public transit. It accounts for the differences in average trip length by mode and thus more accurately reflects travel by each mode. Because unlinked transit trips are significantly shorter than auto trips, the mode share calculation, based on PMT, is markedly lower than the level for trip-based count or survey measures.

Figure 5 indicates a slight increase in PMT-based transit mode share from 1995 through 2001, with the trend reversing and showing a decline in share for transit after 2001. This measure shows the estimated 2003 mode share being at one of the lowest historical levels with approximately 1% of total national PMT being carried by public transit.

An alternative strategy for reporting PMT-based mode share is to use urban, noncommercial vehicle PMT, since this is the more comparable market for most transit services. Transit is not intended as an alternative for commercial/freight traffic or for intercity travel. Based on the share of VMT that is urban, approximately 60%, and factoring out commercial traffic from the measure of VMT, the values for PMT-based mode share increase by about 75% to approximately 1.75%. These adjusted comparisons give a clearer impression of the role that transit plays in urban personal mobility; however, the PMT-based indicator is relevant in the context of discussion of overall transportation investment policy.

Survey-Based Measures of Transit
Mode Share

The previous section's derived mode share estimation is only one way to explore transit mode share trends. Other national survey data also provide insight into transit mode share.

Census Journey-to-Work

Journey-to-work mode share can be calculated from census data long-form information. These data are available for prior censuses and contain a large sample with a high response rate. The census data are based on a question that asks: "How did you usually get to work last week?" Guidance is provided to the respondent relating to multimodal trips where the dominant mode is to be noted as the primary mode, and how to handle multiple work trips, working away from the normal workplace location, etc. A detailed list of transit modes is defined including taxi, ferry, commuter rail, etc. Work at home is a category of response and is typically included in the denominator of the mode share calculations.

For the census, the spring delivery results in the respondent answering with respect to the narrowly defined timeframe and hence does not capture seasonal variation. The greatest sensitivity regarding the application of census data relates to whether or not the "usual trip" language impacts the interpretation of the results in contrast to other measures. A perception exists that transit may be an occasional mode for noncaptive travelers and hence usual mode measures might underrepresent actual everyday average use. This is discussed in more detail below.

American Community Survey (ACS)

As the planned replacement for the decennial long form, this annual smaller sample survey is similarly structured and has been in the pretest application stages before planned ongoing implementation starting in 2005. The commuting questions in this survey follow the census long-form language by querying about the most frequent mode in the reference week. Respondents are continually surveyed (unlike the census). Work-at-home respondents are included in the denominator. The available ACS results from the sample counties are counties that, with respect to transit mode share, are more transit-intensive. (The U.S. Department of Transportation (DOT) has evaluated the census long form mode share results for these same counties in comparison to national average mode shares to determine why the ACS has shown a somewhat higher mode share.)

American Housing Survey

This survey 3 is conducted by the Census Bureau for the Department of Housing and Urban Development. It collects data on the nation's housing: apartments, single-family homes, mobile homes, and vacant housing units; and household characteristics, income, housing and neighborhood quality, housing costs, equipment and fuels, size of housing unit, journey to work, and recent moves. National data are collected in odd numbered years, and data for each of 47 selected metropolitan areas are collected about every 6 years. The national sample covers an average of 55,000 housing units. Each metropolitan area sample covers 4,100 or more housing units. The mode question is identical to that asked in the census long form, in the ACS, and in the 2001 NHTS person file.

Omnibus Household Survey

The Omnibus Household Survey,4 conducted by the Bureau of Transportation Statistics, is a major data-collection exercise to assess customer satisfaction in fulfillment of the DOT Performance Plan. The survey asks supplementary questions every other month to address five DOT strategic goals: safety, mobility, economic growth, the human and natural environment, and national security. It asks general questions about satisfaction with the transportation system and public interactions with DOT agencies. Data for the survey come from telephone interviews of approximately 1,000 randomly selected households and are weighted to allow inferences about the non-institutionalized population aged 18 years or older currently living in the United States. The mode question, asked every other month, is stated as:

On a typical day in September, to get to work did you:

01) Walk

02) Drive or ride in a personal vehicle, not in a company car

03) Drive or ride in a carpool or vanpool

04) Use public transit

05) Drive or ride in a company car

06) Bicycle to work

07) Use a combination of modes

97) Other

Figure 6 shows the trends for these different national travel surveys. Each survey uses somewhat different sampling methods, definitions of terms, and reference time periods. Information about sample sizes, errors, and the specific questions is readily available from the respective survey agencies. These data suggest that over a longer period of time (e.g., comparing 1990 and 2000 data), the transit mode share has declined for census and household survey data sources. Survey information from the more recent years paints a somewhat less clear picture. Of particular interest is the NHTS. This source indicates a mode share of 1.59% of person-trips on transit. Differences in survey questions, mode classifications, and samples require modifications to the data to make meaningful comparisons to the prior years' data. Adjustments for sample and definition differences result in a mode share of 1.76%, closer to the 1.81% the 1995 survey. Thus, this data source suggests a very slight decline in overall mode share for transit in the past six years. This is discussed in more detail below.


The survey methodology for carrying out the NHTS is refined with each application in order to provide the best possible data while still trying to preserve comparability over time. In comparing the 2001 NHTS with the 1995 Nationwide Personal Transportation survey (NPTS) transit mode share calculations, there were several subtle differences that needed to be accounted for to enhance the comparability of the estimates.

Use of an add-on sample.The 2001 national sample NHTS numbers produced a transit mode share of 1.561%. This, however, is not directly comparable to the 1995 number for a number of reasons. The 1995 NPTS included the add-on samples (add-on samples are larger samples above and beyond the original national sample purchased by specific geographies to address local needs), which, while factored to remain representative of national totals, nonetheless, produced a slightly different transit mode share. When the 2004 release of the NHTS database with the add-ons included became available, the transit mode share was again calculated and produced a slightly higher 1.591% share.

Adjustment for higher walking trip reporting. The 2001 NHTS was designed to try to do a better job of gathering information about walking trips. This included an additional probing question to specifically solicit information on walk travel. The result was a significant increase in reported walking trips presumed to be well beyond actual changes in the walk mode and a result of the change in survey design. This increase in total trips about which information was gathered had the effect of slightly depressing the transit mode share because the total trip denominator was now a larger number. To quantify the impact of this, it was assumed that the walk trip rate remained the same as in 1995 for purposes of estimating total trips. With this adjusted measure of total trips, the transit mode share would have been approximately 0.04% higher.

Definition of transit.The 2001 NHTS had a slightly different set of submodes that were classified as public transit. In 1995, intercity bus and courtesy bus were included in the calculation of transit mode share. The 2001 survey disaggregated the data to allow a closer estimation of what is typically referred to as public transportation. If the 2001 transit mode definition is adjusted to be most comparable to the 1995 data, transit mode share in 2001 shows an increase of approximately 0.065%.

Children under age five.The 1995 and prior surveys excluded trips by children under five years of age. This population segment travels only modest amounts and disproportionately less by transit. If the 2001 data are adjusted by removing children under five to be most comparable to 1995, the mode share for public transportation increased approximately 0.029%.

Collectively, these adjustments produce the mode share calculation to be used in comparison with the 1995 NPTS findings (summarized in table 1). It is important to understand that these adjustments are made only to increase the comparability between the 1995 and the 2001 survey numbers. In absolute terms, the 2001 NHTS directly calculated mode share number appears to be the more accurate reflection of actual transit share. The particular interest in exploring this issue in greater detail is to both allow a more comparable data trend analysis and specifically to explore the relative change in mode share between 1995 and 2001, as indicated in NHTS/NPTS data, versus the changes perceived and calculated by looking at field data on ridership changes and calculated mode shares. This will be discussed in more detail later.

Table 2 presents a variety of different survey-based measures of transit mode share. These are for various points in time, various survey methods, and various trip purposes. Caution should be used when comparing these data items; however, the collective message can provide guidance to analysts regarding mode share trends.

Table 3 shows the transit share from the Omnibus Household Survey. These numbers should be used with caution for two reasons. The sample is small for measuring transit share, and the transit category may exclude transit used as part of a trip on a combination of modes. Thus, while these data may be interesting, we did not include them in figure 6 nor do we comment on them in detail.


One of the challenges in comparing mode shares measured across data sources is understanding the comparability of questions that inquire as to usual mode from those that seek information about a specific trip (actual mode). NHTS is unique in that both questions are asked of respondents, thus providing an opportunity to reflect on the differences. As indicated in table 2, NHTS data on actual work trip mode share is noticeably different, with actual mode share on transit being more than 1% lower than the usual mode measures. This indicates that individuals who indicate a usual mode of transit are less likely to use transit as an actual mode on a given day.

Usual mode questions typically refer to the conditions for the prior week. Thus, the respondent is answering in the context of a specific period of time that may have included multiple work trips and multiple modes. Presumably, someone who travels on a given mode more than half the time would indicate that as the usual mode. It is not uncommon, for example, for a transit traveler to commute by transit four days per week and then take an auto on Fridays to facilitate an evening event or early work departure. Similarly, a regular auto traveler may choose to use transit on a given day due to auto unavailability or other factors. The usual mode range of categories also includes a "work at home" choice; thus, this deflates the shares for the other categories slightly as this category is now included in the denominator in the share calculation. For actual mode questions, work at home is not a choice; thus the shares for all other options are proportionally slightly larger.

Table 4 presents an analysis of the usual and actual travel mode for work trips from the 1995 and 2001 surveys. This table confirms an interesting phenomenon. Auto usual mode travelers are far more likely to be strongly loyal to the auto mode, with very modest use of transit for their actual trips (0.1% transit use for actual trips in 2001), whereas usual transit mode travelers used auto modes for 18.4% of their actual trips in 2001. These data confirm the behavior that is required to produce the differences between the usual and actual mode shares observed in NHTS data. To further verify this phenomenon, we analyzed 1995 data, also presented in table 4.

One can apply some algebraic calculations to derive the required mode loyalty for auto and transit travelers for the reported differences in usual and actual mode relationships to be valid. For these conditions to be true required less than 4% of usual mode auto travelers to use transit on a given trip. As the actual transit use by usual auto travelers declines there is an opportunity for greater auto use by transit travelers. An equation can be defined to describe the conditions that would be required to produce any given combination of transit usual mode and actual mode shares.

These data, while logical when analyzed, are contrary to some perceptions of the impacts of reliance on the usual mode question in many travel surveys. The usual mode question has troubled some policy analysts because of the lack of certainty it creates in the data, and because there is the expectation that over time the loyalty to any given mode is lessened as more choices are available. It is more common—due to higher auto ownership/availability, more working spouses and flexible work arrangements, more prevalent alternatives to driving (work at home, transit, car/vanpool)—to presume that travel arrangements may be becoming more diverse with individuals choosing different modes in response to specific activity plans for the day. Thus, there have been some concerns that the usual mode measure might be underestimating transit use as well as use of other modes like walk, bike, and shared ride. However, for transit, the data do not bear out this perception. In fact, the usual mode question strategy appears to overstate the actual share of workers commuting on transit in any given day.

As the data in table 4 suggest, actual travel day behavior can vary significantly from usual mode. In general, for individuals whose usual mode is transit, less than 70% of them used transit on the actual day. Usual transit users frequently share a ride, walk, or use a single-occupant auto. The 2001 survey suggests that usual transit users were slightly more loyal to transit on the actual day than was the case in 1995. In the case of auto, the data suggest that of those with auto travel as a usual mode, over 97 percent used auto on the specific travel day.

These shares are consistent with those that would make the reported usual mode and actual mode transit shares mathematically correct. Usual mode auto travelers seldom use transit for their actual trip, with travel on transit being only a fraction of 1% of actual trips. Usual transit travelers have the least loyalty to their mode of all the usual categories, whereas drive alone auto usual mode individuals have the greatest mode loyalty.

Between 1995 and 2001, the differences between usual mode and actual mode grew significantly indicating that non-usual transit users were less likely to use transit. This increased loyalty to all modes perhaps runs counter to perceptions of a commuting force that is using a range of travel options. A host of factors, including the share of households with more than one vehicle and the tight time constraints for workers in the strong economy period leading up to 2001, might be supporting the trend to greater mode loyalty. It is important to remember that a significant share of the auto commuters do not have walk or shared ride options available to them and may not have transit access at one or both ends of their work trip, thus they use autos 100% of the time.

It is difficult to draw many conclusions from these data on the degree of captivity of transit travelers or the degree of options open to auto travelers. However, it does make both mathematical and logical sense when reviewed in the context of the observed travel behavior of the public. While occasional use of transit may be growing for work travel as the work force grows, there is no evidence that the share of occasional transit use by usual auto travelers is growing. This analysis is restricted to work travel and should not be generalized to other trip purposes.


One of our objectives in compiling various measures of transit use was to take into account that individual measures of transit use or mode share are each subject to various limitations associated with the way they are collected. As noted, each source has limitations including sample sizes and response rates as well as the inherent sensitivity to response bias that is of particular concern to transit researchers.

Transit, by its very nature, suffers from response bias in many travel surveys. Transit travelers are more inclined to have language or literacy problems, be reluctant to disclose sensitive information, be less likely to have telephones available for phone surveys, or otherwise be at risk of being underrepresented in survey data-collection efforts. Nonetheless, within each data source the quality of the data appears to be improving over time and longitudinal comparisons provide insight into overall aggregate national trends. While one would appropriately use greater caution when working with smaller subsets of these databases, the application of ridership and mode share trend data for national transit policy is supported by the collective set of available data. There is no evidence or speculation that response bias has changed over time; thus, no basis exists for discounting the value of the longitudinal survey data as an indicator of trends.

Within the transit community there are a number of initiatives that are likely to improve the data on ridership levels over time. APTA and the Federal Transit Administration are collaborating to produce a single national estimate of annual transit ridership that will reduce the confusion of having multiple national count measures. NTD data are continuing to improve with more agencies using farebox count data rather than sampling as the basis for ridership estimates, and swipe or scan passes improve the count information. The emerging implementation of Automated Passenger Counters offers the prospect of more reliable ridership and trip length counts. The initiative to implement monthly NTD data collection also offers the prospect of greater quality control for NTD data.

Survey design and sampling methods continue to improve; however, resource constraints and the intractability of literacy, language, and disclosure fears impeding response rates will continue to challenge survey methods to provide highly confident measures of transit use, particularly for sensitive subgroups. Larger samples offer the greatest chance of improving confidence in survey data, particularly for geographic or other more narrowly defined market segments. Careful sample design and more aggressive nonresponse followup can increase the response rates and minimize bias.

Various initiatives underway such as the American Community Survey, which is planned to provide year-round samples and more experienced professional data-gathering staff, may improve data quality for work trip commute questions. Beyond data-gathering initiatives already discussed, there is certainly room for a richer understanding of how transit is used as a component of a multimodal trip and in developing a better understanding of the relationship between linked and unlinked trips at the national level.

A greater understanding of the alternative mode for transit trips would provide more insight into the transportation impacts of transit trips. However, simply understanding the data differences in existing data sources and presenting the variations is an important first step in using transit mode share data for various policy deliberations. At the national level, the collective body of existing data provides a sound basis for having stronger insight into mode share trends. While the data are not perfect, their shortcomings should not dissuade its use for policy deliberations nor discredit the messages that can be gleaned from a multiple source review of mode share trends.


The following observations can be discerned from the body of data on transit use and mode share:

  • The evidence on transit use trends across sources is consistent with declines in unlinked trips in the early 1990s followed by strong ridership growth through 2001, at which point ridership began declining.
  • No body of data exists on the industrywide changes in the relationship between linked and unlinked transit trip making (the ratio of unlinked to linked trips). However, the evolution of more transfer-friendly fare media—such as all day passes and the expansion of rail systems that can produce higher total boardings (as some one seat bus trips now become a feeder bus and rail trip)—may be increasing the ratio of unlinked to linked trips. For example, fare structure changes in New York City, where an all day pass was instituted, contributed to the growth in ridership because individuals no longer had to pay for boarding each subsequent vehicle for multiple vehicle trips. However, the trend in public transit PMT is clear and tracks with the trend in trips, as the average trip length has remained relatively constant according to NTD measures.
  • All the data sources appear to confirm the decline in mode share for both work and nonwork trips through 1995.
  • All the survey data sources appear to confirm the stable to slight upward trend in work trip mode share from 1995 to 2001 (unfortunately count data do not provide trip distribution information to confirm these trends). The census data in 1990 and 2000 bridged the trough in transit mode share and do not reflect the turn in trend in the mid-1990s.

The most challenging discontinuity among the various data items is that the NHTS overall mode share trend from 1995 to 2001 does not appear to support the calculations of the ridership count data sources. The PMT-based measures of mode share showed an increase in the share of trips on transit between 1995 and 2001. Had that been confirmed by the NHTS, the NHTS mode share number would have been approximately 1.95% rather than 1.76%. It is not possible at this point to explain the differences in share. All the data sources, including the count data, are subject to a variety of uncertainties. For example, the significant differences between NTD and APTA data for a given property and for the country as a whole are uncomfortably large (Chu 2004).

Count data are likely to become more reliable and higher due to electronic fareboxes and automated ticket vending and, thus, part of the recent trends in ridership growth may not be actual increases in transit use. Others have speculated that the shift to pass-based fare systems has resulted in unlinked trip numbers increasing faster than linked trips or passenger-mile numbers. Each measure of transit use has a slightly different definition and trip linking and trip length are not robustly determined. Among the possible explanations for the NPTS/NHTS trend from 1995 to 2001, which is inconsistent with industry data, are that the 1995 NPTS overstated transit use or that the NHTS survey method resulted in a noticeable undercount of transit ridership.5In spite of this inability to completely rationalize the various data sources, some clear conclusions can be drawn:

  • Regardless of various refinements that may be identified over time, it is clear that transit has grown in total trip terms and has stabilized its overall mode share or perhaps changed modestly through 2001. The work trip share appears to have grown slightly in the late 1990s, but the duration of the growing mode share may have been quite limited. Also, national aggregate ridership count data do not include trip purpose data nor do they enable measurement of linked trips, thus complicating interpretation.
  • It is equally clear that transit will need to reverse course from the most recent trends and continue to post meaningful year-over-year ridership gains if it is to play a larger role in meeting overall urban travel needs. While there is a heightened sensitivity to transit mode share as it fluctuates between growth and decline, the pace of change has moderated from the long-term historic trend of significant declines. It is also evident that the absolute level of transit use at the national aggregate level is modest in terms of trips and especially modest in terms of PMT.
  • Nationally, transit mode share is not changing rapidly. One should also note, however, that the particular context in each community may deviate substantially from this national trend. The national mode share for transit does not provide a full picture of the contribution of transit to peak-period peak-direction travel in critical corridors in many of the larger urban areas in the United States, nor does it reflect the importance of transit to those who are dependent on or choose to use public transit services. Nonetheless, the overall role of transit as seen from the various measures of mode share is a relevant consideration in public policy and investment programming decisions.
  • This paper does not speak to other trends that are occurring within the population of transit users. There is an acknowledged shift toward a greater share of transit being on the rail mode, there are more long trips as express and rail services penetrate ever more distant suburbs, and there is a growing trend for transit service to access major attractors such as airports, sports stadiums, and major retail complexes. Aggregate measures of ridership and mode share do not fully capture the nature of the mobility role that transit plays in various urban areas.

While there are opportunities for improvements in measuring transit use and mode share, and these improvements may be important to individual agency service design and planning activities, the strategy of synthesizing multiple measures of mode share provides a knowledge base that is useful in informing national-level policy deliberations and in identifying data needs and differences.


Chu, X. 2004. Ridership Accuracy and Transit Formula Grants, Transportation Research Board Pre-Print Number 05-1537.

Cox, W. and R.D. Utt. 2002. Census Shows Commuters are Rejecting Transit, The Heritage Foundation, Executive Memorandum, Sept. 5.

Joint Center for Political Studies. 1985. Demographic Change and Recent Worktrip Transit Trends, Working Paper No. 2. Urban Mass Transportation Administration, Washington, DC.

Kain, J.F. and Z. Liu. 1999. Secrets of Success: Assessing the Large Increases in Transit Ridership Achieved by Houston and San Diego Transit Providers. Transportation Research, Part A Policy and Practice33(7–8):601–624.

Mason, J.W. 1998. The Buses Don't Stop Here Anymore: Sick Transit and How to Fix It. The American Prospect56–62, March.

Millar, W.W. 1999. Overview of Public Transit in America. Business Briefing: Global Mass Transit Systems44–46.

Norman, J. 2003. Statewide Study of Public Transportation to Attract Non-Traditional Transit Riders in California. California Department of Transportation.

Pisarski, A.E. 2003. Some Thoughts on the Census: Transit Statistical Match-Up. Transportation Quarterly57(3):11–16.

Project for Public Spaces, Inc. 1999. The Role of Transit Amenities and Vehicle Characteristics in Building Transit Ridership, TCRP Report 46.

Schmidt, S. 2001. Incentive Effects of Expanding Federal Mass Transit Formula Grants. Journal of Policy Analysis and Management20(2):239–261.

Stanley, R. 1998. Continuing Examination of Successful Transit Ridership Initiatives. TCRP Research Results Digest29.

Surface Transportation Policy Project (STPP). 2002. Transit Growing Faster than Driving. Decoding Transportation Policy & Practice3, May 29.

Taylor, B. and P. Haas. 2002. Increasing Transit Ridership: Lessons from the Most Successful Transit Systems in the 1990s. Mineta Transportation Institute, San Jose State University.

Taylor, B.D. and W.S. McCullough. 1998. Lost Riders. Access 13:27–31.

TransManagement. 2004. Counting Transit so that Transit Counts, prepared for the American Public Transportation Association. May. Available at http://www.apta.com/research/info/online/documents/counting_transit.pdf.

Transportation Research Board (TRB). 2001. Making Transit Work: Insights from Western Europe, Canada, and the United States,Special Report 257. Washington, DC.

Urban Mobility Corp. 2002. Mass Transit Debate Continues. Innovation Briefs13:6.

Urbitran Associates, Inc. 1999. Guidelines for Enhancing Suburban Mobility Using Public Transportation, TCRP Report 55.


1. See http://www.apta.com/research/stats/ridershp/#A3.

2. See http://www.ntdprogram.com.

3. See http://www.census.gov/hhes/www/ahs.html.

4. See http://www.bts.gov/omnibus_surveys/household_survey/.

5. Additional perspective on this issue may be gained by reading "Counting Transit so that Transit Counts" (TransManagement 2004).


Corresponding author: S. Polzin, Center for Urban Transportation Research, University of South Florida, 4202 East Fowler Avenue, CUT100, Tampa, FL 33620-5375. E-mail: polzin@cutr.usf.edu

X. Chu , Center for Urban Transportation Research, University of South Florida, 4202 East Fowler Avenue, CUT100, Tampa, FL 33620-5375. E-mail: xchu@cutr.usf.edu