Women-only Option in a Peer-to-Peer Marketplace: the Case of Long-distance Carpooling

How to cite

Berasategui, Ignacio; Astier, Nicolas; Lambin, Xavier. 2024. "Women-only Option in a Peer-to-Peer Marketplace: the Case of Long-distance Carpooling." AEA RCT Registry. AEARCTR-0013241. https://doi.org/10.1257/rct.13241-3.0

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

Ignacio Berasategui* , Nicolas Astier , Xavier Lambin

* Primary investigator

Registry ID
AEARCTR-0013241
DOI
https://doi.org/10.1257/rct.13241-3.0
First registered
Last updated
Status
In development
Trial dates
2024-03-25 – 2024-12-31
Countries
Private
JEL codes
J16, L91
Registry record
http://www.socialscienceregistry.org/trials/13241

Abstract

Women-only options in digital marketplaces have emerged as a response to an amplified need for safety in the interaction between men and women. These designs remove the uncertainty about the identity of the user's counterparts, and by doing so, they limit the possibility of undesired events. As such, women-only options aim to improve female users' experience and encourage their participation. In one of the largest digital platforms for long-distance carpooling, this project will study the effects of rolling out a women-only option on the participation and usage decisions of female drivers and passengers. The study is planned to consist of three independent interventions that randomize (i) the possibility of choosing the women-only feature for female drivers, (ii) the awareness of the availability of the women-only feature, and (iii) the display of the women-only badge to female passengers.

The first intervention aims to answer three questions. What are the characteristics of women drivers who value the women-only mode the most? How does the availability of this feature change other trip characteristics (asked seat price, automatic booking)? Approximately, how does the introduction of the women-only mode affect the participation decision of female passengers in the presence/absence of alternative transportation means (i.e. buses)?

The second intervention analyzes the intensive and extensive margins of female user participation: how many more trips are posted/booked when the user is aware of the mode? How many more users become active on the platform?

The third intervention aims to quantify the value that female passengers have for the women-only feature. As such, it answers the question: how much is a female passenger willing to pay for a women-only trip? How are the participation and welfare of female users affected when the women-only feature is available?

Experimental design

Intervention 1:

A share of women drivers who have started posting a trip will be randomly assigned to be exposed, as they enter the characteristics of their trip, to the possibility of selecting the “women-only” option. A limited number of origin-destination pairs in Brazil and France will be purposedly left entirely non-treated in order to elicit preliminary evidence on market-level outcomes on passengers’ side (motivating evidence to assess the relevance of intervention 3).

Intervention 2:

Registered women users will be randomly assigned to three groups. One group will receive an email both encouraging them to use the platform and, critically, prompting them to try the women-only feature (treatment group). A second group will receive an email only encouraging them to use the platform (control group). The users will not receive any email (pure control).

Intervention 3:

Women passengers looking for a ride will, as they receive results from their search, be randomly assigned to either see or not the women-only option that was chosen by female drivers (if any). In addition, some random variations in ride prices will be introduced in order to be able to retrieve, from the estimation of a discrete choice model, female passengers’ willingness to pay for the women-only option.

(see PAP for more information)

Design details

Randomization method
Randomization done by computer in all interventions.
Randomization unit
Individual-level randomization in all interventions. In interventions 1 and 2, the driver is the randomization unit. In intervention 3, the passenger is the unit of randomization. However, in intervention 1 we choose to exclude from treatment 30 routes in France, and 30 routes in Brazil, to compare outcomes at the route-level.
Clusters
There is no clustering.
Observations
Given the nature of the business model and the intervention, the sample size fundamentally depends on the number of postings and searches by users. These volumes are difficult to anticipate and predict, even at a regional and country scale. However, we expect ongoing discussions with the platform to provide more specific estimates.
Arms
Total sample size is subject to previous precisions. However, proportions by treatment arms are described below. Intervention 1: 30 routes in France and 30 routes in Brazil will be excluded from treatment. In the rest of the routes and countries, half of the female drivers are assigned to the treatment group (women-only feature in the publication flow), and half will be assigned to the control group (no women-only feature in the publication flow). This intervention focuses on female drivers. Intervention 2: one-third of the users will be assigned to the treatment group (encouragement email with the women-only feature), one-third to the control group (encouragement email without the women-only feature), and one-third to the pure control group (no encouragement email). This intervention focuses on female users who have posted trips as drivers at some point in time. Intervention 3: Half of the users will be assigned to the treatment group (women-only badge is visible), and half will be assigned to the control group (women-only badge is not visible). This intervention focuses on the choices of female passengers.
Primary outcomes
Intervention 1: - Female drivers’ decision to select the women-only feature. - Number of female passengers’ requests to travel via carpooling. - Number of female passengers traveling via carpooling and bus. - Main characteristics of posted trips (e.g., prices, and booking mode). - Female driver’ decision to post after being treated and frequency of the postings. Intervention 2: - Female drivers’ decision to post a trip and frequency of the postings. - Female passengers’ activity (bookings, searches). Intervention 3: - Female passengers’ search and request choices (for each passenger, specific search, specific carpooling trip or bus choice, and main features of the choice including time before departure). - Female passengers’ final mean of transportation (for each passenger, specific carpooling trip or bus choice, and main features of the choice including time before departure).

Cited by

No citation data available for this registration yet.

Keywords