Please fill out this short user survey of only 3 questions in order to help us improve the site. We appreciate your feedback!
Diverse Committees
Last registered on February 10, 2020


Trial Information
General Information
Diverse Committees
Initial registration date
February 10, 2020
Last updated
February 10, 2020 10:49 AM EST
Primary Investigator
Uni Essex
Other Primary Investigator(s)
PI Affiliation
University of Essex
PI Affiliation
Kings College London
Additional Trial Information
In development
Start date
End date
Secondary IDs
A popular argument in favour of diversity in committees is that different groups might have access to different information. Hence, by omitting some groups this information might be lost to the committee.

We study committee decision making when there is possible correlation between the information held by members of the same group. If people fail to properly account for this correlation, then decision making in homogeneous (non-diverse) committees might be substantially worse. We provide theoretical predictions for four scenarios differing by whether information of members of the same group is correlated or not and whether members of different groups have identical preferences over outcomes or not and then test these using a laboratory experiment.
External Link(s)
Registration Citation
Hughes, Niall, Friederike Mengel and Zia Ul Hassan Khan. 2020. "Diverse Committees." AEA RCT Registry. February 10. https://doi.org/10.1257/rct.5402-1.0.
Experimental Details
Intervention Start Date
Intervention End Date
Primary Outcomes
Primary Outcomes (end points)
(i) Committee decisions (dummy indicating whether committee decision was correct); (ii) individual guesses (dummy indicating whether they were correct)
Primary Outcomes (explanation)
Our primary question is whether heterogenous committees make better decisions than homogenous committes measured by the variables indicated above.
Secondary Outcomes
Secondary Outcomes (end points)
Expected payoffs of committee members of different types.
Secondary Outcomes (explanation)
When preferences are different, then “correct” decisions (in the sense of matching the state) might not be the same as (ex ante) “preferred” decisions. Hence for this case we will also look at the expected payoffs of committee members.
Experimental Design
Experimental Design
Participants in our experiment are randomly assigned to either of two groups, S or T. They answer a pre-experimental questionnaire, play 12 rounds of the game described below and then answer a post-experimental questionnaire.
In each round of the experiment participants will (i) receive a signal, (ii) discuss (via a chat window) with their committee members and (iii) vote in committees of three participants about which of two urns (yellow or purple) was selected at the beginning of the round.
Treatments differ according to whether the signal received at stage (i) is correlated among members of the same group or not. They also differ by the payments received when voting correctly. In some treatments both groups prefer the purple state. In some treatments preferences are not aligned with one group preferring the purple state and one group the yellow state.
Experimental Design Details

Randomization Method
participants register for sessions without knowing what session is about. assignment of sessions to treatments is done in the office by computer.
Randomization Unit
experimental session
Was the treatment clustered?
Experiment Characteristics
Sample size: planned number of clusters
32 sessions.
Sample size: planned number of observations
864 individuals.
Sample size (or number of clusters) by treatment arms
8 sessions per treatment.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
As there are no prior studies with effect sizes on which to base our choice of sample size we proceeded as follows. We conducted two sessions (54 participants) of the treatment with correlated information and identical preferences across the two groups. In these two sessions we detected an effect size of 0.107 for the committee level variable and 0.083 for the individual level variable. We then conducted a power analysis using these effect sizes. Our chosen sample of 216 participants per treatment across 8 clusters allows us to detect a committee level effect of 0.107 with 78.47% power (at the 5% level) and the individual level effect of 0.083 with 88.45% power again at the 5% level. As show up for experimental sessions can be unpredictable, we cannot commit to the exact sample size of 216 participants per treatment. Instead, we propose to proceed running sessions until the threshold of 216 participants has been reached. If we achieve the same show up as in the initial sessions this would mean running 8 sessions (clusters) per treatment.
IRB Name
University of Essex Social Sciences Ethics Subcommittee
IRB Approval Date
IRB Approval Number
Analysis Plan

There are documents in this trial unavailable to the public. Use the button below to request access to this information.

Request Information
Post Trial Information
Study Withdrawal
Is the intervention completed?
Is data collection complete?
Data Publication
Data Publication
Is public data available?
Program Files
Program Files
Reports, Papers & Other Materials
Relevant Paper(s)