Set Your Committees Up for Success

From investment committees weighing major capital expenditures to R&D stage gate reviews, organizations rely on committees to make high-stakes go/no-go decisions. The logic seems sound: Bring together diverse expertise and viewpoints, and one will identify comprehensive solutions to help make high-stakes decisions. Yet, while assembling the “right people” in the room is necessary for solid decision-making, it is only the first step. Even the most qualified experts can fail to share crucial insights. This is because committee dynamics and unwritten norms can discourage open disclosure, leaving critical information and dissent trapped inside individuals’ heads, never making it into the collective discussion. From self-censorship, to interpreting silence as agreement, to the suppression of dissent, groups of experts routinely struggle to surface the information needed during discussions to make a sound decision. The result is a graveyard of spectacular and catastrophic misjudgments, from the Challenger Space Shuttle disaster to the Bay of Pigs invasion. Why do committees often fail at surfacing pertinent information? And how can their fates be reversed? Our research on the Food and Drug Administration’s Advisory Committees reveals that not only is who sits at the table important, but also how the committee process is structured. Simple changes to committee protocols can fundamentally shape the incentives to share information, dramatically improving decision quality and unlocking collective intelligence. Why Committees Fail Despite the many advantages that committees bring, decision-making groups are known to face several challenges in surfacing and processing information. At the core are cognitive, psychological, and social costs to sharing the information that is privately held by individual members of the committee. All these challenges stem from a common root: sharing unique information incurs significant costs to the individual. Cognitive costs. Our human cognitive limits often make the sharing of private information less likely. This is because unique, or “unshared,” data often requires greater cognitive effort to recall and articulate—especially when its relevance to the current discussion is uncertain. The result is that individuals in groups naturally gravitate toward shared and common information as it is easier to retrieve, validate, and discuss. This bias toward the “easily available” can lead teams to privilege redundancy over discovery. Psychological costs. Sharing private information introduces a form of psychological vulnerability that individuals prefer to avoid. Voicing a divergent viewpoint or raising a previously unmentioned concern risks being wrong—or worse, it may be perceived as out of sync with the group. This phenomenon, known as evaluation apprehension, can be especially acute in hierarchical or high-status teams where mistakes carry reputational and career penalties. Individuals may self-censor not because they lack valuable insights, but because the psychological toll of potential judgment or dismissal outweighs the perceived benefits of contributing. Social costs. The social dynamics of group settings may further discourage the introduction of unshared information. Groups tend to reward consensus and cohesion, creating subtle pressures that reinforce conformity. Discussing commonly held knowledge helps build rapport and reinforces a sense of alignment, while injecting new or contradictory data can feel disruptive. Social motivations like mutual enhancement, where members gain credibility by affirming shared understanding, can lead to a cycle where members seek validation over accuracy. In such environments, the very norms that foster harmony may also suppress the diversity of thought necessary for rigorous decision-making. When committee processes fail to actively encourage the surfacing of diverse perspectives, participants naturally default to safer behaviors: discussing what everyone already knows, seeking consensus, and minimizing personal risk by limiting contributions. These failures are not inevitable consequences of group decision-making. Rather, they are predictable outcomes of flawed committee design. As our research on the FDA shows, subtle changes in how committees operate can lead to powerful improvements in information sharing, unlocking the genuine potential of the collective expertise assembled in the committee. A Simple Change in the Voting Protocol As part of its regulatory authority, the FDA convenes Advisory Committees (AdComs) to assist its decision-making. These AdComs are expert panels whose purview includes evaluating new drugs and medical devices, and making recommendations on product approvals before they enter the U.S. market. For example, in December 2020, the Vaccines and Related Biological Products AdCom met to discuss Emergency Use Authorization for the Pfizer-BioNTech Covid-19 Vaccine. Given the consequential nature of the recommendations, a typical AdCom meeting lasts a full day, with the vast majority of time spent on presenting, scrutinizing, and debating data and evidence on the product’s safety and efficacy. Following these extensive discussions, the meeting typically culminates in a vote of the AdCom’s confidence in the product. In 2007, the FDA made a seemingly minor procedural change to the way its AdComs operated. Prior to the change, post-discussion voting was typically conducted in a sequential manner: each committee member would visibly and audibly cast their vote and explain their position in turn, one after the other. After the change, voting was to be simultaneous: members would all vote at the same time, and only after voting closed would they go around to reveal their votes. The FDA’s stated goal with the change was to mitigate the possibility that members were “herding.” When voting happens sequentially, early votes can initiate a bandwagon effect and unduly influence voters later in the voting order. By contrast, simultaneous voting eliminates this possibility, freeing each member to render an independent judgment based on their unique perspective. This procedural change created a powerful quasi-experiment for us to study committee dynamics. We analyzed over 500 verbatim meeting transcripts before and after the voting protocol change using natural language processing techniques and linguistic analysis. What we discovered offers valuable lessons for any organization that relies on committees to make important decisions. Better Discussions, More Dissenting Voices, and Better Decisions After the implementation of simultaneous voting, FDA AdComs exhibited a substantial decline in unanimous voting outcomes: the proportion of unanimous committee recommendations decreased from 45% to 27%. This shift appears to reflect the FDA’s intended goal of reducing “herding” behavior and promoting greater independence among members. Our analysis reveals a more nuanced explanation. In fact, the decrease in unanimous voting outcomes was primarily driven by concurrent changes in discussion content and dynamics. By applying a word embedding model to analyze the meeting transcripts, we see that under the new voting procedure, committees tended to discuss a greater breadth of topics. In other words, the experts were more likely to share their private information, and the committee discussion tended to scrutinize the problem more comprehensively. Because committee members now explored the issue from multiple, fresh angles, they unearthed previously hidden insights. Those fresh perspectives and the surfacing of unshared data translated directly into a wider distribution of viewpoints and, therefore, vote outcomes. Why did discussions change so markedly? By removing the possibility of viewing their peers’ leanings before voting, the revised protocol realigned individual incentives towards engaging in information sharing and fact finding. Without the possibility to glean other members’ votes before having to cast their own, the experts now had a stronger motivation to probe, share their own privately-held data, and test assumptions throughout the discussion. We further investigated what kinds of behaviors drove this expansion in discussion breadth and found clear evidence that experts actively changed their interaction patterns. Members began to “pull” more information from one another by purposefully directing a higher proportion of questions to peers. They also redistributed speaking time more evenly amongst themselves, and they adopted language that felt authentic, positive, and free of status-laden cues. These behavioral shifts and changes in linguistic patterns ultimately helped reduce the psychological and social costs typically associated with offering unshared or dissenting information, encouraging a richer and more open exchange of ideas. Importantly, these changes also translated into better decisions. After the change in voting protocols, drugs recommended for approval by the AdComs were significantly less likely to be associated with post-market boxed warnings or full withdrawals due to safety or efficacy concerns. In other words, the AdComs made more sound recommendations as to which products should be allowed in the market. In sum, a simple procedural change altered the incentives for individuals to overcome the “costs” and expend the effort needed for solid discussions, triggering a cascade of behavioral adaptations. By removing early visibility of peer votes, experts compensated by proactively uncovering one another’s views during discussion. This led to greater information revelation which, more than voting independence alone, appears to underlie the improvement in decision-making performance. Towards Better Group Decision-Making Organizations seeking to improve the quality of their collective judgments must look beyond selecting expert participants. They need deliberate mechanisms that motivate members to surface and assimilate diverse insights, challenge assumptions, and collaborate constructively. Our research offers three key insights to structuring group meetings: Focus on committee process, not just outcomes. Effective committee decision-making depends not only on who is in the room, but on whether the meeting structure encourages members to contribute their knowledge fully. A key insight from our research is that how meetings are run (especially the incentives embedded in procedures like voting protocols) shapes how deeply members engage, share private information, and seek out diverse perspectives. Too often, organizations focus on the decision outcome while overlooking the mechanics that produce it. This leads to focus on achieving consensus and moving on. But sustainable decision quality hinges on creating protocols that lower barriers to contribution and encourage learning. Address status and communication barriers head-on. In high-performing committees, we found more balanced speaking time and greater use of language that promoted fact-finding and sharing of viewpoints. These features are not accidental: they reflect an environment where psychological safety is actively cultivated. Leaders play a key role here. They must monitor and manage participation patterns, encourage genuine engagement, and positively reinforce contributions that challenge the consensus or introduce new information. When members feel safe to speak and are invited to question each other, they are more likely to surface unshared insights critical to robust decisions. Structure pre-vote discussions deliberately, and adopt simultaneous voting. The richest moments of value creation in group decisions happen before the vote, when diverse viewpoints collide, overlap, and cohere into a shared understanding. Encourage committee members to direct questions to one another. One surprisingly powerful tactic is to implement simultaneous voting using electronic buttons, which prevents members from seeing how others vote in real-time. Unlike the common practice of going around the room round-robin, simultaneous voting reduces social influence and “herding” effects. When participants know they must form an independent judgment, they engage more actively during discussion. Our research shows this simple procedural change can significantly enhance the quality of collective decisions. You cannot guarantee perfect group decisions. But, you can stack the deck by deliberately engineering processes that incentivize candor, curiosity, and critical challenge.

Set Your Committees Up for Success

Summary.

Partner Center