The Effects of Uber Vouchers on Transportation Choice, Habit Formation, and Academic Outcomes: Evidence from a Randomized Controlled Trial

How to cite

Kofoed, Michael; Neller, Seth; Harris, Matthew. 2026. "The Effects of Uber Vouchers on Transportation Choice, Habit Formation, and Academic Outcomes: Evidence from a Randomized Controlled Trial." AEA RCT Registry. AEARCTR-0017861. https://doi.org/10.1257/rct.17861-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Michael Kofoed* , Seth Neller , Matthew Harris

* Primary investigator

Registry ID
AEARCTR-0017861
DOI
https://doi.org/10.1257/rct.17861-1.0
First registered
Last updated
Status
In development
Trial dates
2026-02-27 – 2028-11-06
Countries
Private
JEL codes
I23, R41
Registry record
http://www.socialscienceregistry.org/trials/17861

Abstract

Uber partnered with the University of Tennessee, Knoxville (UTK) to evaluate the effectiveness of ride sharing services in alleviating campus parking scarcity and the last mile problem (i.e., students still needing a way to get from one mode of transportation to their final destination). Commuting students require access to parking, but on-campus students also demand on-campus parking, even if it is just for private transportation to work or the store. Ride sharing services like Uber can aid in this dilemma by serving as complements to existing public transportation and substitutes for private automobiles (Hall et al. 2018). Our objective is to study how treated students change their transportation preferences, whether they continue to use rideshare after they use all their vouchers (habit formation), and whether access to alternative transportation increases academic outcomes.

We will use a randomized control trial (RCT) to evaluate the effectiveness of ridesharing in improving academic outcomes such as GPA or reenrollment at UTK the following semester. Students are eligible to participate in the trial if they are first-year students who live on campus and do not have access to a parking pass. Eligible students are randomly assigned via lottery to receive either 10-$10 Uber vouchers (treatment group), or only 1-$10 Uber voucher (control group) to use for one semester. Approximately 1,000 students will be part of the treatment group, and up to 5,000 students will be part of the control group. The study will use ride data from Uber, student data from UTK Registrar’s office, and parking permit request data from the Ofiice of Parking and Transportation. Results from this study will be considered at the university level to measure how access to Uber vouchers can form new habits after the vouchers run out, and whether it increases students’ academic outcomes.

Intervention

The allocation of Uber vouchers to first-year, on-campus students is testing how they can alleviate parking pressures, influence student behavior, and improve academic outcomes. Qualified students enroll in the trial by receiving and registering through a link for a lottery to receive 10-$10 Uber vouchers. The control group will randomize 5,000 students to receive 1-$10 voucher. The treatment group will randomize 1,000 students to receive 10-$10 vouchers. For vouchered rides, Uber will provide data on pick-up location, drop-off location, and amount paid. For unvouchered rides, Uber will provide trip data aggregated at the month level by student. The data provided by Uber will be given a randomized subject identification number by the UTK Registrar so that no data will be given to the researchers that can be identified.

Experimental design

In late fall 2025, the UTK Office of Parking and Transportation will send emails to qualified students notifying them of a new partnership with Uber and to watch out for a link to register for the lottery to receive 10-$10 Uber vouchers. In January 2026, a second email will provide a link for students to register.

Funding permitting, we will perhaps repeat this expirement during the Fall 2026 semester. We may increase the amount or expand the pool of potential students to all UTK undergradautes but will update the pre-analysis plan before conducting this expansion.
Once a student clicks the link, it will bring them to a Qualtrics survey that first presents the consent form. If a student consents to participate in the research, the Qualtrics survey will then prompt them to complete the attached survey. Once a student completes the survey, the software will enroll them in the study. The survey should take a student around 20 minutes to complete. Since the student will be consenting to use of FERPA protected data, Qualtrics will prompt them to sign the form with the UTK NetID/password authentication. This authentication was provided to us by OIT.
Once the portal closes, we will randomize 1,000 students into receiving 10-$10 Uber vouchers (treatment group). Students who did not receive the treatment will receive 1-$10 Uber voucher as compensation for the survey (control group); up to 5,000. At the end of the semester, Uber will match names, email addresses, and cell phone numbers (needed to register for an Uber account) to their data that will include two subgroups of data. For vouchered rides, Uber will provide data on pick-up location, drop-off location, amount paid. For unvouchered rides, Uber will provide trip data that has been aggregated to the student by month level (number of rides, average price, fraction of rides that originate and end in Knox County). All other variables included in Uber’s transfer but not required for analysis will be excluded before the dataset is shared with the research team.
Before the data are merged, the research team will use receive three key variables from the survey: student Uber account email, student Uber account cell phone number, and NetID. The research team will then create an anonymous randomized subject id number. The research team will provide a key containing the Uber account email, student Uber account cell phone number, and randomized subject id number to Uber. These variables Uber will match the data with this key and then drop the Uber account email and student Uber account cell phone number.
UTK Registrar’s office will receive the data from Uber and then link student to the data set. The UTK registrar’s office will use the second half of the key created by the research team that contains the NetID and the randomized subject id number to link the Uber data to the data from the UTK administration. These variables include demographic data such as race, gender, high school GPA, ACT/SAT scores, Pell status, parking pass requests, UTK GPA, enrollment, and major. UTK administration will then de-identify the data by dropping all pii including NetID. The researchers will then obtain a data set that only has the randomized subject id number. The research team will destroy the key after providing the needed pieces to Uber and the UTK registrar. The UTK Parking and Transportation Office and the UTK Registrar are not part of the research team but are helping in obtaining and linking the data. The UTK Registrar will serve as the safe harbor for the data linkage for UTK.
The final analytical dataset will contain the randomized subject identification number, the treatment status, data on pick-up location, drop-off location, amount paid (for vouchered rides), number of rides, average price, fraction of rides that originate and end in Knox County for six months previous and every month for a year after voucher delivery (for unvouchered rides), race, gender, high school GPA, ACT/SAT scores, Pell status, parking pass requests, previous semesters UTK GPA, Spring 2026/Fall 2026, UTK GPA, Fall 2026/Spring 2027 enrollment, and major choice. All these variables will be on the student level. After the research team deletes the key, they will not have any access to the personal identifying data.

Design details

Randomization method
Lottery by computer
Randomization unit
Individual
Clusters
6,000 total students
Observations
6,000 total students
Arms
3,000 treated students 3,000 controlled students
Primary outcomes
Number of vouchers used; Number of trips overall; Reenrollment at UTK the next semester; GPA; Requesting a parking past for the next semester.
Secondary outcomes
Different effects of vouchers on different types of students

Cited by

No citation data available for this registration yet.

Keywords