Back to History

Fields Changed

Registration

Field Before After
Trial Status in_development on_going
Abstract Exclusion can damage group member's health and work performance. The central objective of the study is to understand the patterns of exclusion and whether the information on the likelihood of being excluded helps or harms the victims. In our laboratory study, we invite a set of subjects about whom we have information in multiple dimensions, including age, gender, body type, face shape, eye and hair color, willingness to trust and reciprocate, degree of altruism, etc. Subjects are assigned to groups of six. Using the distributions of the attributes, we create fictional person cards and ask the subjects to rank them according to the expected likelihood that they will be excluded by their group of six. Subjects receive a positive payoff for each person card that they rank on the same position as ranked by the group in total. The subject whose attributes are closest to the person card that is ranked with the highest likelihood to be excluded by the group is selected as the group's victim and receives a reduced payoff. In the baseline treatment, the subjects provide an initial ranking, before having a group discussion via computer chat. The information on the initial ranking is not provided by the experimenter. After the discussion, the final rankings are submitted and the game ends with payoffs and no repetition. In the revealed treatment, the initial ranking decision phase is followed by an information phase, in which the potential victim - given the initial rankings - is revealed by the experimenter. The remaining part of the game is the same as in the baseline treatment. Exclusion can damage group member's health and work performance. The central objective of the study is to understand the patterns of exclusion and whether the information on the likelihood of being excluded helps or harms the victims. In our laboratory study, we invite a set of subjects about whom we have information in multiple dimensions, including age, gender, body type, face shape, eye and hair color, willingness to trust and reciprocate, degree of altruism, etc. Subjects are assigned to groups of six. Using the distributions of the attributes, we create fictional person cards and ask the subjects to rank them according to the expected likelihood that they will be excluded by their group of six. Subjects receive a positive payoff for each person card that they rank on the same position as ranked by the group in total. The subject whose attributes are closest to the person card that is ranked with the highest likelihood to be excluded by the group is selected as the group's victim and receives a reduced payoff. In the baseline treatment, the subjects provide an initial ranking, before having a group discussion via computer chat. The information on the initial ranking is not provided by the experimenter. After the discussion, the final rankings are submitted and the game ends with payoffs and no repetition. In the revealed treatment, the initial ranking decision phase is followed by an information phase, in which the potential victim - given the initial rankings - is revealed by the experimenter. The remaining part of the game is the same as in the baseline treatment. Due to technical issues, the study is run online and not in the laboratory, but with the same set of subjects randomized across treatments. Due to the online setup, we no longer need the six-subject groups and cannot provide the group discussion. Instead, rankings and re-rankings are performed by the subjects in the baseline treatment without any interaction with others. The baseline treatment is subdivided into one subgroup with a neutral request to reevaluate their ranking and one subgroup with an ethically framed request. In the treatment group, after the initial ranking, we report a number of person cards that were either among the top five most likely victims ("mimic treatment") or explicitly not among the top five most likely victims ("attention treatment") of a previous session. Subjects then receive a neutral request to reevaluate their ranking. In contrast to the original plan (see above), we do not actually choose victims who have a payoff loss. The advantage of the new online setup is that we can evaluate each individual's decisions as an independent observation. The downside of the new online setup is that some of the subjects drop out before completing the experiment.
Last Published January 31, 2024 01:33 PM May 31, 2024 08:51 AM
Intervention Start Date March 04, 2024 March 18, 2024
Intervention End Date May 31, 2024 September 30, 2024
Experimental Design (Public) In our laboratory study, we invite a set of subjects about whom we have information in multiple dimensions, including age, gender, body type, face shape, eye and hair color, willingness to trust and reciprocate, degree of altruism, etc. Subjects are assigned to groups of six. Using the distributions of the attributes, we create fictional person cards and ask the subjects to rank them according to the expected likelihood that they will be excluded by their group of six. Subjects receive a positive payoff for each person card that they rank on the same position as ranked by the group in total. The subject whose attributes are closest to the person card that is ranked with the highest likelihood to be excluded by the group is selected as the group's victim and receives a reduced payoff. In the baseline treatment, the subjects provide an initial ranking, before having a group discussion via computer chat. The information on the initial ranking is not provided by the experimenter. After the discussion, the final rankings are submitted and the game ends with payoffs and no repetition. In the revealed treatment, the initial ranking decision phase is followed by an information phase, in which the potential victim - given the initial rankings - is revealed by the experimenter. The remaining part of the game is the same as in the baseline treatment. In our laboratory study, we invite a set of subjects about whom we have information in multiple dimensions, including age, gender, body type, face shape, eye and hair color, willingness to trust and reciprocate, degree of altruism, etc. Subjects are assigned to groups of six. Using the distributions of the attributes, we create fictional person cards and ask the subjects to rank them according to the expected likelihood that they will be excluded by their group of six. Subjects receive a positive payoff for each person card that they rank on the same position as ranked by the group in total. The subject whose attributes are closest to the person card that is ranked with the highest likelihood to be excluded by the group is selected as the group's victim and receives a reduced payoff. In the baseline treatment, the subjects provide an initial ranking, before having a group discussion via computer chat. The information on the initial ranking is not provided by the experimenter. After the discussion, the final rankings are submitted and the game ends with payoffs and no repetition. In the revealed treatment, the initial ranking decision phase is followed by an information phase, in which the potential victim - given the initial rankings - is revealed by the experimenter. The remaining part of the game is the same as in the baseline treatment. Due to technical issues, the study is run online and not in the laboratory, but with the same set of subjects randomized across treatments. Due to the online setup, we no longer need the six-subject groups and cannot provide the group discussion. Instead, rankings and re-rankings are performed by the subjects in the baseline treatment without any interaction with others. The baseline treatment is subdivided into one subgroup with a neutral request to reevaluate their ranking and one subgroup with an ethically framed request. In the treatment group, after the initial ranking, we report a number of person cards that were either among the top five most likely victims ("mimic treatment") or explicitly not among the top five most likely victims ("attention treatment") of a previous session. Subjects then receive a neutral request to reevaluate their ranking. In contrast to the original plan (see above), we do not actually choose victims who have a payoff loss. The advantage of the new online setup is that we can evaluate each individual's decisions as an independent observation. The downside of the new online setup is that some of the subjects drop out before completing the experiment.
Randomization Method subjects are recruited with a homogeneous recruiting tool they are randomized via computer program to different treatments and to different groups of six within each treatment subjects are recruited with a homogeneous recruiting tool they are randomized via computer program to different treatments and to different subgroups within each treatment
Planned Number of Clusters 120 to 150 individuals in 20 to 25 groups of six in each treatment (i.e. 240 to 300 individuals in 40 to 50 groups in total) 120 to 150 individuals in each treatment (i.e. 240 to 300 individuals in total)
Planned Number of Observations 120 to 150 individuals in 20 to 25 groups of six in each treatment (i.e. 240 to 300 individuals in 40 to 50 groups in total) 120 to 150 individuals in each treatment (i.e. 240 to 300 individuals in total)
Sample size (or number of clusters) by treatment arms 120 to 150 individuals in 20 to 25 groups of six in each treatment (i.e. 240 to 300 individuals in 40 to 50 groups in total) 120 to 150 individuals in each treatment (i.e. 240 to 300 individuals in total)
Back to top