The influence of anonymity on joint problem-solving

How to cite

Mickeler, Maren; Zunino, Diego; Hadengue, Marine; Kretschmer, Tobias. 2022. "The influence of anonymity on joint problem-solving." AEA RCT Registry. AEARCTR-0010037. https://doi.org/10.1257/rct.10037-1.1

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

Maren Mickeler* , Diego Zunino , Marine Hadengue , Tobias Kretschmer

* Primary investigator

Registry ID
AEARCTR-0010037
DOI
https://doi.org/10.1257/rct.10037-1.1
First registered
Last updated
Status
Completed
Trial dates
2022-09-12 – 2022-09-18
Countries
France (Paris, Provence-Alpes-Côte d’Azur & Hauts-de-France)
Registry record
http://www.socialscienceregistry.org/trials/10037

Abstract

Organizations increasingly rely on digital platforms to post predefined challenges or problems, asking for members' submissions. Individuals can contribute directly when they post original ideas and solutions or indirectly when they give feedback to others' submissions. In general, a solution’s overall quality depends on individuals’ direct and indirect contributions. While research so far has primarily focused on the antecedents and outcomes of the former, very little is known about the drivers and consequences of individuals’ indirect contributions. We test whether granting individuals anonymity stimulates their feedback-giving behavior with regard to contributions by others. To do so, we run a field experiment with business school students on a digital platform that allows direct and indirect contributions of peers in the context of a hackathon. Our treatment takes place at the platform level and involves granting individuals anonymity in providing feedback to the solutions of others on the platform. We measure the quantity and valence of the feedback provided by participants, whether they incorporated the feedback received, and the quality of participants' final submissions.

Intervention

We run a field experiment on digital platforms that allow direct and indirect contributions among peers during a hackathon. Specifically, participants must engage in a creative process to define and solve a specific problem within the context of higher education. Over the course of five days, participants learn about the problem, are asked to develop solutions and provide feedback to other participants. Our treatment takes place at the platform level and involves granting individuals anonymity in providing feedback to others on the platform. We randomly assign (i) students to teams and (ii) teams to experimental conditions. We set up two versions of a digital platform that match each other in all functions but one, which is the provision of participants’ identity to peers. In the control version of our platform, participants interact with their peers using their full, real names. In the treatment version, participants interact with their peers using anonymous pseudonyms, which are generated by randomly combining an adjective with an animal noun, in a similar fashion to shared documents on collaboration platforms such as Google docs (e.g., “happy zebra” or “colorful dog”).

Experimental design

The experiment is based on a fully randomized between-subjects design. It is scheduled to take place in the context of a hackathon that runs at a major business school in France. Participants are 739 students, who have just joined the school at one of its three sites. Students work in groups of five on the task of developing a creative solution for higher education via a digital platform. Overall, the hackathon runs for five days. Day 1 exposes students to the topic, after which they work on identifying problems and receiving feedback from both coaches and peers on day 2. On day 3 students work on developing solutions and receiving feedback from both coaches and peers, after which they finalize their solution and vote on the best contributions on day 4. Finally, a subset of selected students presents their solutions to a panel of coaches and a jury of experts on day 5.

Design details

Randomization method
We assign participants to our treatment and control groups using a computerized random number generator.
Randomization unit
We use a two stage randomization where we first assign (i) students to teams and (ii) teams to experimental conditions.
Clusters
We plan to have 70 teams in our control and 78 teams in our treatment group.
Observations
Overall, we plan to have 739 participants (348 individuals in our control and 391 in our treatment group).
Arms
70 teams (348 individuals) in our control, 78 teams (391 individuals) in our treatment.
Primary outcomes
First, we measure feedback giving as the number of comments each student provides to others on the platform. We measure both the number of comments posted as well as the number of participants’ quick reactions to peers’ contributions. Second, we measure the tone of the feedback given by running sentiment analysis algorithms on comments provided, as well as the number of participants’ negative reactions. Third, we measure whether groups incorporate the feedback received by comparing the initial solutions uploaded on the platform at the beginning of the experiment with the final solutions submitted at the very end.
Secondary outcomes
We measure the quality of solutions posted on the platform.

Cited by

No citation data available for this registration yet.

Keywords