Leadership Turnover and the Stability of Cooperation Standards in Teams

Participants take part in a repeated group decision task. Groups consist of three contributors and one leader. Contributors repeatedly choose costly effort in a team production setting. The leader observes noisy information about contributors’ behavior and can assign rewards or sanctions subject to limits. The study varies whether the same leader remains assigned to the group in the second part of the task or whether a different leader is assigned.

2026-05-21

2026-12-31

Pilot:
• Implementation feasibility of the oTree experiment
• Correct group, role, phase, and leader assignment
• Correct payoff calculation and data export
• Session duration, completion rate, and comprehension check performance

Main study:
• Group level change in average contributor effort around the phase boundary

Pilot outcomes are feasibility outcomes and are not used to estimate the main treatment effect.

For the main study, average contributor effort is the mean effort of the three contributors in a group and round. The primary outcome is constructed at the group level as average contributor effort in rounds 11 to 13 minus average contributor effort in rounds 8 to 10.

Pilot data are excluded from the confirmatory main analysis.

• Full post turnover change in average contributor effort
• Immediate round 11 change in average contributor effort
• Within group dispersion of contributor effort
• Reward and sanctioning intensity
• Reward and sanction targeting by performance signal
• Contributor payoffs
• Post game survey measures on perceived effort standards, perceived reward and sanction thresholds, predictability, fairness, strictness, and clarity of consequence rules

• Full post turnover change: average contributor effort in rounds 11 to 20 minus average contributor effort in rounds 6 to 10
• Immediate round 11 change: average contributor effort in round 11 minus average contributor effort in rounds 8 to 10
• Within group dispersion: standard deviation of contributor effort within a group and round, averaged over the relevant window
• Reward and sanction intensity: total reward points and total sanction points assigned by the leader per group and round
• Reward and sanction targeting: mean reward and sanction points assigned to contributors by observed signal bracket
• Contributor payoffs: average contributor payoff at the group round level, reported before and after rewards and sanctions where applicable
• Survey measures: post game self reports used for descriptive mechanism checks and heterogeneity analyses, not as causal mediators

The project is a laboratory experiment with a pilot stage followed by a planned main study. Participants interact in groups of four. Three participants repeatedly choose costly effort in a team production task, while one participant acts as leader. The interaction lasts 20 rounds. The study varies whether the same leader remains assigned after round 10 or whether the group receives a different leader. The leader observes noisy bracket signals about individual effort and the team average effort, and can assign rewards or sanctions subject to per person and total consequence limits. Pilot data collected before the main experiment were used only for implementation updates and feasibility checks, including timing checks, interface checks, payoff and consequence calibration, and data export checks, and will not be included in the confirmatory main analysis.

Not available

Computer randomization implemented in oTree during laboratory sessions.

There are two levels of randomization. First, participants are randomly assigned to groups of four and to roles within groups, with three contributors and one leader per group. Second, treatment assignment is at the group level: each contributor group is assigned to either the turnover condition or the control condition.

Yes

Planned main study: at least 80 contributor groups, with at least 40 groups assigned to control and at least 40 groups assigned to turnover. If additional chair funding and laboratory capacity are available before confirmatory outcome analysis begins, the sample may be increased up to a prespecified maximum of 160 contributor groups, with equal allocation across treatment arms. The final sample size within this range will be determined by budget and laboratory capacity, not by estimated treatment effects from the main study. This range is based on pilot informed implementation, budget, and power calibration, including sensitivity analyses over lower, pilot based, higher, and high variance scenarios. Pilot groups are used for feasibility and calibration checks only, are excluded from the confirmatory main sample, and will not be used in the confirmatory main analysis.

Planned main study: at least 320 participants, with four participants per contributor group and at least 80 contributor groups in total. If additional chair funding and laboratory capacity are available before confirmatory outcome analysis begins, the sample may be increased up to a prespecified maximum of 640 participants, corresponding to 160 contributor groups. The final sample size within this range will be determined by budget and laboratory capacity, not by estimated treatment effects from the main study. Pilot participants are excluded from the confirmatory main sample and will not be used in the confirmatory main analysis.

Planned main study: at least 40 contributor groups assigned to control and at least 40 contributor groups assigned to turnover, corresponding to at least 160 participants per treatment arm under complete groups of four. If additional chair funding and laboratory capacity are available before confirmatory outcome analysis begins, the sample may be increased up to a prespecified maximum of 80 contributor groups per treatment arm, corresponding to 320 participants per treatment arm under complete groups of four. The final sample size within this range will be determined by budget and laboratory capacity, not by estimated treatment effects from the main study. Pilot groups are excluded from the confirmatory main sample and will not be used in the confirmatory main analysis.

Power calculations are based on the group level primary outcome, defined as the change in average contributor effort from rounds 8 to 10 to rounds 11 to 13. The unit is effort points on the 0 to 100 effort scale. Effects of 4, 6, and 8 effort points correspond to 4, 6, and 8 percentage points of the feasible effort range and are interpreted as absolute effects in either direction. The pilot based standard deviation of the group level pre to post change is 5.92 effort points, based on 4 complete pilot groups. Because this pilot based estimate is imprecise and the final design differs from the pilot implementation, power is evaluated by simulation for 40 to 80 groups per condition and with sensitivity to lower, pilot based, higher, and high variance scenarios. At the planned minimum of 40 groups per condition, the design is mainly powered for moderate to large effects under the pilot based variance scenario. Increasing the sample within the prespecified range improves power for smaller effects and for higher variance scenarios. Pilot data are used only to calibrate implementation and variance inputs and are excluded from the confirmatory main analysis.