We investigate how worker productivity and behavior in dynamic contests respond to changes in the ambient environment in which these contests are played. We conduct lab experiments among members of the general population.
Intervention
University students are invited to an experimental lab session that lasts for a maximum of 1 hour. They are informed that these sessions will be held on campus, in a room in the School of Design and Environment. They are told that they will "perform different tasks on the computer to evaluate task performance in a lab setting, and answer a survey with questions about demographic characteristics and health behaviors." Invitees are informed of the earnings potential and are also told that "the entire experiment will take a maximum of 1 hour and you can leave once you have completed the experiment." Among those students who are interested in participating, they sign up to one session among a list of alternative sessions, without knowledge of the predetermined ambient lab conditions they will encounter... [remainder omitted here but entered in the "hidden" version of the pre-registry, to be be made public subsequently].
Experimental design
Subjects earn a participation fee of $10. (All dollars are SGD.)
Each lab session has two parts. In the first part, participants answer demographic (including health behavior) questions and do a piece-rate activity that lasts 10 minutes. The activity consists of reversing alphanumeric strings and participants earn $0.20 for every text they correctly reverse.
The second part consists of a two-player contest. Each participant is told that s/he is being paired with another participant with similar performance (based on Part 1).
Here are the contest rules for each of three variants of the contest (Descamps et al., 2018). Participants are informed of these rules and are quizzed on them. Players who pass the contest-understanding quiz are paid an additional $3.
Baseline variant [Omitted here but entered in the "hidden" version of the pre-registered Experimental Design, to be made public subsequently.]
Strategic variant [Omitted here but entered in the "hidden" version of the pre-registered Experimental Design, to be made public subsequently.]
Psychology variant [Omitted here but entered in the "hidden" version of the pre-registered Experimental Design, to be made public subsequently.]
Design details
Randomization method
Each experimental lab session was randomly assigned, with uniform probability, to a predetermined ambient lab condition.
Randomization unit
Our randomization unit is the experimental lab session.
Clusters
We will conduct experimental lab sessions in two waves:
1. December 2021: We will run two "pilot" sessions (December 14) and 18 experimental sessions (three sessions in each of the following days: December 16, 17, 18, 20, 21, 22), totaling 20 sessions. For each session, up to 18 college students will sign up. Among the students who signup and show up to their session on the stipulated day and time, we will admit students in pairs, as the key component of the experiment is a two-player contest. So, for example, if 14 students sign up to a given session and 13 of these show up, we will send the last student to arrive (the 13th student) home with an additional $10 participation fee (totaling $20), or invite him/her to wait for a subsequent session if there is another session taking place later that day.
2. February 2022: Pending room availability and our ability to recruit enough participants and bring them into campus during the December 2021 holidays, we will recruit another ~200-250 subjects to participate in the same experiments. Besides the "polar" ambient lab conditions that we will experiment with in December 2021, we may include intermediate conditions as well.
Observations
For the lab sessions to take place in December 2021 and February 2022, we plan on recruiting 500 to 650 participants. The two pilot sessions taking place on December 14, 2021 will involve 36 subjects (18 times 2). If the pilot sessions run without problems, we will pool the observations with the subsequent sessions. If (unanticipated) problems occur during the pilot sessions, we will detail the problems we encountered when conducting these sessions and may decide to discard their observations, in which case we will justify our decision to the reader.
Arms
A treatment arm is given by a combination of contest variant (baseline, strategic, psychology) by ambient lab condition (with two "polar" ambient lab conditions planned for December 2021). We will sign up (up to) 18 students per session and run 3 sessions per arm, totaling 54 participants per arm. Further details are provided in the uploaded schedule [omitted here but uploaded in the "hidden" version of the pre-registry, to be be made public subsequently].
Primary outcomes
We investigate how environmental quality affects (i) the number of tasks completed by round and (ii) transition probabilities in best-of-3 contests, for example, the probability that a player wins a second round after having won vs. lost a first round to a similarly able player.
Secondary outcomes
We also collect responses to questions measuring the participant's understanding of the rules of the contest being played, and their subjective valuations on the ambient lab conditions that they encounter. Besides measuring task completion during a player-pairing activity followed by a best-of-3 contest, we allow subjects to stop investing from an endowment and measure this choice. Further details are provided in the uploaded pamphlet [omitted here but uploaded in the "hidden" version of the pre-registry, to be be made public subsequently].
Cited by
No citation data available for this registration yet.