The Value of Social Information in the Workplace: An Online Experiment

We will run an online experiment with participants recruited from Prolific, who will be randomly assigned to receive social information about performance in a real-effort task.

Participants are asked to perform a real-effort task: count the number of zeros in a row of zeros and ones. Prior to the task, they are informed that they can choose to receive information about how past participants have performed in the task (i.e., the average performance and the number of rows each participant solved). Their WTP for this information is elicited under two different scenarios in an incentive-compatible manner. In one scenario, the information is provided before they perform the task, and in another scenario, the same information is only provided after they have performed the task. One of the two scenarios will be randomly selected and based on their WTP response for the scenario-that-counts, workers are assigned to receive or not receive social information either ex-ante or ex-post.

2024-05-01

2024-05-07

Performance in the real-effort task, willingness-to-pay for social information that is provided either ex-ante or ex-post, and cognitive uncertainty regarding effort choice

Participants are recruited on Prolific as "workers" to perform a real-effort task. The task involves counting the number of zeros in a row of randomly-ordered zeros and ones, as adapted from Abeler et al. (2011). They are told that they will be paid a piece-rate for performing the task and the work period would last up to 5 minutes. Based on their willingness-to-pay for social information, they will be randomly assigned to receive information about performance in the real-effort task. Participants have to complete a second round of the same task and answer a series of questions. There is no deception at any point in the study.

In the experiment, workers perform an effort-intensive task which involves counting the number of zeros in a row of randomly-ordered zeros and ones, as adapted from Abeler et al. (2011). After giving their consent to participate in the experiment, workers review the instructions about the task, and go through a practice round for up to 45 seconds. Workers are told that they will be paid a piece-rate for performing the task on top of their participation payment of $4. The work period would last up to 5 minutes, in that participants have the option to end the task early if they wish.

Before workers perform the task, they are first asked to indicate how much effort (in terms of the number of rows solved) they intend to put in and then state their cognitive uncertainty about whether their intended effort is optimal. This elicitation follows the protocol of Enke and Graeber (2023), where subjects use a slider to calibrate the statement from a scale of 0% (completely uncertain) to 100% (completely certain) in steps of 5%: `"You indicated that you intend to solve X rows within the 5 minutes. How certain are you that your optimal effort falls within the range of solving between (X - 2) and (X + 2) rows?" Additionally, we ask workers to evaluate how they expect to perform in the task relative to other participants.

Next, we informed workers that a previous version of the study had been conducted and that we could provide them with information about how these past participants (the ``reference population'') had performed. This information contains the average performance, along with a detailed breakdown of the performance of each participant. We employed an incentive-compatible Becker-DeGroot-Marschak (BDM) procedure to determine workers' WTP for social information under two different scenarios. In one scenario, the information is provided before they perform the task, and in another scenario, the same information is only provided after they have performed the task. The WTP elicitation follows the structure implemented in Butera et al. (2022): ``Under scenario X, would you like to receive information about how past participants performed in the task?'' After making their choice, workers were asked to state how much of their bonus of 50 cents they would be willing to give up to ensure that their preferred choice of information would be implemented. In particular, the question asked, ``If the information is to be provided before you begin (after you have completed) the task, you indicated that you prefer / prefer not to receive information about how past participants performed. How much of your 50 cents bonus would you be willing to give up to ensure that you receive / not receive this information?'' In sum, the WTP responses range between -50 to 50 cents and the workers' decisions are used to determine their assignment to the receipt of social information either ex-ante or ex-post based on the scenario-that-counts.

Workers proceed to perform the task for up to 5 minutes, and upon completion of the task, we asked a few questions to elicit their experience during the task (e.g., how difficult they found the task, how stressed they were while performing the task, etc.) and elicit their cognitive uncertainty again. Thereafter, workers are informed that they will perform the same task again for the same duration in a second round. We ask them the same set of questions as in the first round, and elicit their cognitive uncertainty regarding their effort choices in the second round. That is, the second round is structured the same way as the first.

Finally, workers fill out an exit survey aimed at understanding why they made the choice to receive or not receive social information, both ex-ante and ex-post, as well as what makes them cognitively uncertain (or completely certain) about their effort choices. For those assigned to social information, we additionally ask them how they interpreted the information and whether it affected how they approached the task. Finally, workers are informed about their total payment and this concludes the experiment.

Randomization by a computer. The program is implemented using oTree.

Individual

No

We aim to recruit 700 participants on Prolific. Participants must take the study using a laptop or personal computer. We restrict to participants who (i) reside in the US, (ii) have completed at least 100 prior studies, and (iii) have an approval rating of at least 95 percent.

Same as above.

Participants are randomized to either receive or not receive social information, that is to be provided ex-ante or ex-post. This will be determined by their WTP for social information using the incentive-compatible BDM mechanism.