Abstract
The city of Pittsburgh's Department of Mobility and Infrastructure (DOMI) and Move PGH have asked Carnegie Mellon to evaluate the impact of their innovative Universal Basic Mobility project on the geographic and socioeconomic mobility of low-income Pittsburgh residents. Lee Branstetter will lead the evaluation effort, using smartphone apps to measure the impact on geographic mobility and an existing partnership with Allegheny County DHS to leverage the resources of the DHS data warehouse to measure the impact on socioeconomic mobility.
Description
Our CMU team will begin a pilot study in cooperation with the City of Pittsburgh’s Department of Mobility and Infrastructure, SPIN (a national provider of e-scooter transportation services), and Move PGH, a nonprofit. This study will provide a group of low-income citizens in Pittsburgh access to a menu of subsidized transportation options, including standard mass transit (through monthly transit passes), Pittsburgh’s POGOH bike-sharing service, SPIN (which offers access to smartphone-activated electric scooters across the city), and the Zipcar car rental service. Pittsburgh residents will be recruited into a randomized controlled trial in which the treatment group will get 12 months of generously subsidized access to the full menu of transportation options described above. Members of the demographically equivalent control group will receive gift cards in return for downloading our GPS app and responding to surveys over the same period.
Modern technology allows us to track the movement of study participants in by leveraging the GPS capabilities embedded in smartphones. Use of a smartphone app by members of both treatment and control groups will allow for precise measurement of the magnitude and nature of any impact of the treatment on geographic mobility. This information will be supplemented with detailed user-specific information on utilization of the transportation options described above.
We can also utilize the ongoing partnership between Allegheny County Department of Human Services (ACDHS) and CMU to measure the impact of experimentally induced increases in geographic mobility on the short-term and long-term socioeconomic mobility of our study participants. ACDHS is a national leader in efforts to link the various administrative data sets generated by the major social assistance programs, local schools, police departments, and courts into a single integrated data warehouse, in which a unique identifier code links individual county residents’ data records across data sets. The data warehouse provides comprehensive coverage on the recent income, employment, and social services utilization of the majority of low-income residents of Allegheny County.
In principle, participants in the treatment group could have access to total transit resources worth hundreds of dollars per month for up to one year. This could lead to large measured increases in geographic and socioeconomic mobility. By offering a diverse group of low-income citizens a broad menu of options, the project can also begin to elicit low-income consumer preferences for these options, contingent on consumer characteristics (gender, age, etc.), environmental factors (weather, temperature, time of day, etc.), and other variables by observing participants choices.
The results of this research could provide important guidance to public policy efforts to improve the geographic and socioeconomic mobility of low-income urban residents.
Timeline
Because of unexpected delays in obtaining IRB approval and in recruiting participants, we expect to formally begin the field experiment described above in late spring 2023. Initial plans to recruit residents relied on a community-based organization, Manchester Citizen's Corporation, quickly recruiting the full set of 100 participants based on its close connections to the community. Unfortunately, these efforts were not successful (MCC only succeeded in recruiting 2 prospective participants), and we had to transition to completely different recruiting methods that targeted the whole city and relied on other means. These alternative approaches are bearing fruit and, at the time of this writing, we are closing in on our recruitment target.
When the active study begins, we will be monitoring participant usage of the subsidized transportation options and participant mobility. As the 12-month field experiment unfolds, collaborators with access to ACDHS data will begin measuring and analyzing any emerging differences in income, employment, and utilization of social services between the treatment and control groups.
The active study period, during which we are providing Pittsburgh residents with subsidized transportation, is now expected to last from May 2023 to April 2024, which is after this grant expires. We expect to spend May and June of 2024 analyzing and interpreting the data gathered during the active study period.
Note that we have included as a supplemental information document a press release, generated by the office of Mayor William Peduto that unveiled the Universal Basic Mobility project and described some of its basic characteristics.
Strategic Description / RD&T
Deployment Plan
IRB approval. IRB approval was granted in early 2023.
Recruitment of Participants. Recruitment efforts is focusing on low-income citizens who lack regular access to personal vehicles. Participants will be fully informed about the nature of the study and each participant will sign a consent form providing a record of their fully informed consent to the study's terms and conditions, including the collection of GPS data and use of ACDHS data to measure income, employment, and social service utilization. After consenting, participants will be randomly assigned to a treatment group or a control group. The treatment group will therefore have approximately 50 participants. Each participant, whether in the treatment group or the control group, will be required to download a smartphone app that uses the GPS features of the participant's smartphone to measure the participant's mobility across the city. These data will be anonymized, through a procedure approved by the IRB, to protect participant privacy, but they will enable the detailed, granular analysis of differences in the geographic mobility of the treatment group relative to the control group. Participants will also be asked to fill out short online surveys that capture features of their mobility, employment, and job search activities (if any) not captured by GPS data or other information sources. Control group participants will receive gift cards or equivalent online payments to compensate them for their participation in this study.
Provision of subsidized transportation. Members of the treatment group will be provided with an extensive menu of transportation options available for their use for each of the 12 months of the active study period. This menu will include up to five 30-minute free rides on the SPIN e-scooter service per day, an unlimited monthly Port Authority transit pass, an unlimited number of free 30-minute rides with the city's POGOH bike sharing service, free membership and up to $50 in free credits with the Zipcar car sharing service. Participants in the treatment group will be able to extend their use of these options past the program limits by paying with their own resources. Each transportation provider other than the Port Authority will share individual user ride data with the CMU team. These data will include origin and destination time stamps and GPS coordinates, as well as the amount of usage. Transportation support provision and data collection will begin as soon as the project starts.
Analysis of participants' transportation choices. The limitations of conventional mass transit, the high cost of personal automobiles, and the limited degree to which either of these options fully satisfy the transportation needs of the urban poor, have been long known by transportation researchers. The widespread diffusion of smartphones has enabled a proliferation of new transportation options that offer some for the flexibility of personal vehicles without the fixed costs of vehicle ownership. These new options are often clustered in wealthier neighborhoods and priced at a point that limits access to these options on the part of the urban poor. However, it may be possible for modest subsidies to dramatically improve access, and through that, the geographic and socioeconomic mobility of poor families in our cities. However, the potential that urban planners and transportation policymakers associate with these newer options is tempered by a lack of extensive data on how much and under what circumstances low-income citizens will choose to utilize them, even when access is substantially subsidized. Some of these options have well known shortcomings, especially in challenging weather conditions. By accessing detailed user-specific data on transportation option usage in the context of our project, we can obtain extremely useful information on the nature of and the intensity of low-income resident usage of these new options. As Pittsburgh’s seasons change, and as a diverse group of low-income citizens is recruited into this study, we may be able to infer important elements of consumers’ preferences for these various options. Ever since the pathbreaking work of Nobel laureate Daniel McFadden in the 1970s, economists have built increasingly rich statistical models of transportation mode choice that help researchers quantify the value riders associate with different transportation options. Application of these models will enable us to explore how low-income citizens value different the transportation options included in our "menu," and how those valuations depend on consumer characteristics and external factors, like weather. This analysis will begin at the inception of the project and continue through its end.
Measurement of effects on mobility, income, employment, and social service utilization. As we have already noted, the GPS capabilities embedded in those smartphones allow for the detailed tracking and measurement of individual mobility patterns. Our research team has already acquired extensive experience creating and deploying iOS and Android-compatible smartphone apps that gather detailed GPS location data several times per hour, in a way that achieves a good balance between mobility measurement and battery drain on users’ phones. Through experience, we have identified robust approaches to mobility measurement that can contend with the noise in GPS data. Working together with CMU’s Institutional Review Board, we have developed data anonymization protocols that allow us to analyze these rich data while also protecting the privacy of our participants. Finally, by implementing the best practices of a randomized controlled trial (RCT), we can experimentally manipulate participants’ access to particular transportation options, as described in this proposal, and then make valid causal inferences about the impact of this additional transportation access on mobility outcomes.
The same smartphone apps that collect high-frequency GPS data can also facilitate the sending and receiving of short surveys that capture aspects of the mobility experience which cannot be measured by GPS data alone. We can use these tools to measure self-reported employment and income. However, we can also utilize the ongoing partnership between Allegheny County Department of Human Services (ACDHS) and CMU to measure the impact of experimentally induced increases in geographic mobility on the short-and long-term socioeconomic mobility of our study participants. ACDHS is a national leader in efforts to link the various administrative data sets generated by the major social assistance programs, local schools, police departments, and courts into a single integrated data warehouse, in which a unique identifier code links individual county residents’ data records across data sets. The data warehouse provides comprehensive coverage on the recent income, employment, and social services utilization of the majority of low-income residents of Allegheny County. Birth and school records link adult members of the treatment group to the academic performance of their children in “reporting” school districts. While CMU researchers will not have direct access to sensitive individual-level records, cleared data analysts inside ACDHS can perform statistical analyses on those records and share the results with CMU researchers. This collaboration gives our team an unusual degree of insight into the short- and long-run effects of additional geographic mobility on a broad range of socioeconomic outcomes. This data analysis will begin at the inception of the project. It will continue through the end of the project.
Expected Outcomes/Impacts
We believe this project will advance the "innovating mobility for all" agenda by contributing to research on multi-modal connections, novel modes of transport, and improved transportation access to disadvantaged neighborhoods. It will break new ground in several ways.
First, it will support the City of Pittsburgh's efforts to implement and evaluate the first significant "universal basic mobility" program in the region and, in some ways, the first program of its kind in the nation. This is a high-profile undertaking that has already attracted attention in the press and on the part of officials at the federal Department of Transportation (including former CMU professor Robert Hampshire) This will build on and strengthen the longstanding relationship between the research community at CMU, Mobility 21, the city government, the nonprofits that are part of this program, and the transportation service providers that are engaged in it.
Second, it will provide among the first estimates of low-income urban rider preferences across a broad range of emerging transportation options. We will observe consumer choices and relate them to consumer characteristics and weather conditions, providing badly needed insight to urban planners and transportation resources concerning how much and under what conditions low-income populations will actually utilize these new options. Today, the ratio of speculation on this question to hard evidence is far too high. Our project will remedy this significant problem, generating parameters of a discrete choice model that can inform multiple streams of future research.
Third, the project will create, to our knowledge, the first estimates of the impact of access to a multi-modal menu of transportation options on the geographic mobility of the recipients based on a randomized controlled trial. It will not rely on user logs, but rather on GPS data that systematically measures the mobility of a well-defined treatment group and compares that to the mobility of a well-defined and demographically equivalent control group. Calculated metrics of mobility will include measures of distance traveled, number of unique locations visited, and "radius of gyration." In doing so, we will also create and refine a smartphone app and associated data infrastructure for collection and processing of GPS data that could become a tool widely used by other transportation researchers.
Fourth, the project will generate, to our knowledge, the first estimates of the impact of access to a multi-modal menu of transportation options on the socioeconomic mobility of the recipients based on a randomized controlled trial. It will not rely on self-reported employment outcomes, but rather on the records maintained in the state unemployment insurance (UI) database. We will also be able to measure the causal impact of access to our transportation supports on the utilization of social services, the academic performance of recipients' children, and a broad range of other outcomes, and we can estimate these effects in the short-run (that is, during and immediately after the active study period during which participants in the treatment group receive monthly transportation supports) and in the long-run (that is, months or years after the active study period ends).
The analyses described above may lead directly to significant policy implications and science-based policy recommendations. If our results demonstrate that the provision of the transportation supports described above have economically and statistically significant positive effects on the geographic and socioeconomic mobility of the participants relative to the control group, it could provide strong evidence in favor of a nationwide effort to improve the access of the urban poor to these transportation options. If the effects are large enough, public expenditure in support of additional transportation access could pay for itself through higher tax revenue (obtained from the higher incomes of those who benefit) and lower expenditures on social assistance programs. In addition, our estimation of consumer preferences for different options could guide policymakers in terms of selecting which transportation options are most valuable to low-income urban residents, enabling policymakers to obtain the maximum increase in geographic and socioeconomic mobility by targeting the most useful new transportation options.
Expected Outputs
TRID
Individuals Involved
| Email |
Name |
Affiliation |
Role |
Position |
| sagarbaviskar@cmu.edu |
Baviskar, Sagar |
Carnegie Mellon Heinz College |
Other |
Student - PhD |
| branstet@cmu.edu |
Branstetter, Lee |
Carnegie Mellon Heinz College |
PI |
Faculty - Tenured |
| schizeck@andrew.cmu.edu |
Chizeck, Seth |
Carnegie Mellon Heinz College |
Other |
Student - PhD |
| cdrayton@andrew.cmu.edu |
Drayton, Cameron |
Carnegie Mellon Heinz College |
Other |
Student - Masters |
| beibeili@andrew.cmu.edu |
Li, Beibei |
Carnegie Mellon Heinz College |
Co-PI |
Faculty - Tenured |
Budget
Amount of UTC Funds Awarded
$100000.00
Total Project Budget (from all funding sources)
$299743.00
Documents
| Type |
Name |
Uploaded |
| Data Management Plan |
Data_Management_Plan_Mobility_21_Branstetter_Pittsburgh_UBM.docx |
March 10, 2022, 7:52 p.m. |
| Progress Report |
391_Progress_Report_2022-09-30 |
Sept. 29, 2022, 6 a.m. |
| Progress Report |
391_Progress_Report_2023-03-30 |
March 29, 2023, 10:40 a.m. |
| Publication |
Peer-to-Peer Transportation Platforms, Consumer Mobility, and Urban Consumption Patterns |
March 30, 2023, 5:42 a.m. |
| Publication |
Learning individual behavior using sensor data: The case of gps traces and taxi drivers |
March 30, 2023, 5:43 a.m. |
| Publication |
Driving Low-Income Mothers to Greater Success: The Impact of Ridehailing on Income and Employment |
April 10, 2023, 8:52 p.m. |
| Publication |
Can Ridesharing Help the Disadvantaged Get Moving? |
April 10, 2023, 8:53 p.m. |
| Final Report |
Final_Report_-_Branstetter_391.pdf |
Aug. 3, 2023, 6:21 a.m. |
Match Sources
No match sources!
Partners
| Name |
Type |
| City of Pittsburgh |
Deployment & Equity Partner Deployment & Equity Partner |
| MOVE PGH |
Deployment & Equity Partner Deployment & Equity Partner |
| SPIN |
Deployment Partner Deployment Partner |
