11  Group decision making

Much decision making in the corporate environment involves groups. Group decisions are often better than individual decisions. Groups are often formed to achieve this better performance.

11.1 Intellective versus judgment tasks

Group decision-making tends to be analysed using two types of tasks.

The first is “intellective tasks” that have a demonstrable solution (often labelled “eureka” type problems). The following example of an intellective task is drawn from Newell et al. (2022):

A good example of one of these problems is the rule induction task used by Laughlin and colleagues (Laughlin, 1999; Laughlin, Vanderstoep, & Hollingshead, 1991) which requires participants to induce a rule involving standard playing cards. The task begins with one rule-following card exposed face-up on the table; participants are then asked to select a new card from the deck in order to test their hypotheses about the rule. For example, the 8 of Diamonds might be face up and the to-be-discovered rule might be ‘Two Diamonds followed by two Clubs’. If a participant selects a card consistent with the rule the card is placed to the right of the first card; if it is inconsistent it is placed underneath the first card. Participants continue these trial-by-trial tests of their hypotheses and attempt to use the feedback to infer the rule.

The second is “judgement tasks”, which involve evaluative or aesthetic judgments with no demonstrable solution. An example of a judgment task is forecasting a future event. At the time of the judgment, the “correct” estimate is unknown and typically cannot even be evaluated after the fact. For example, forecasting a 30% probability that Donald Trump will win the election cannot, on its own, be evaluated as correct or incorrect.

11.2 Statistical versus deliberating groups

To analyse group decisions, we need to distinguish two different types of groups.

Statistical groups involve members giving their inputs individually. Those inputs are aggregated to make the forecast or decision. Voting is a simple example, which works well if the majority is right.

There is strong evidence that statistical groups make better decisions than individuals. A story involving Francis Galton is a classic example. Galton (1907) had people estimate the weight of an ox. The median of the individual predictions was only 0.8% from the actual figure.

Deliberating groups involve individuals providing input during deliberations. Those inputs can affect and be affected by the inputs of other group members. People aim to influence others. People might change their minds.

The process for a deliberative group decision making process is best understood contrasted with that for an individual. The first figure below from Newell et al. (2022) shows a schematic of Brunswik’s Lens Model of judgment for an individual. There is a real world in which the criterion or event to be judged exists. The right-hand side is the mind of the judge. The judge is exposed to a set of cues through which they see the world.

For a deliberating group, there are two extra stages. Individual member judgments are developed and then revised before being combined into the group judgment. This schematic shows that group judgment depends on individual judgment. It also shows how discussion can then affect individual adjustments, possibly by the sharing of cues, before a final group judgment emerges.

Hill’s (1982) review of 50 years of research on group deliberating found that group judgments were typically as accurate as the second-best individual member of the group. Similarly, Gigone and Hastie (1997) stated that group judgments tend to be more accurate than the typical judgment of a group member but less accurate than that of their most accurate member. For example, Laughlin et al. (1991) found that using the intellective task described above, the best individuals generated more correct hypotheses than the groups. Groups and the second-best individuals generated a similar number of correct hypotheses.

Deliberating groups have the potential to arrive at the correct answer even when the majority have an incorrect belief at the beginning. But the converse is also possible. Thorndike (1938) had people answer questions individually before answering those same questions in groups. If the majority of the group knew the correct answer to a problem, the group’s decision was correct 79% of the time. (The incorrect minority were able to derail the group 21% of the time.) If the majority of the group answered a question incorrectly when answering individually, the group converged on the right answer only 44% of the time. The result of this dynamic was that the average group decision was better than majority voting, but only marginally so (66% versus 62%).

11.3 Deliberating group failures

Deliberating groups can generate poor decisions due to both the rational conduct of group members and because of their “biases”. They can also temper some biases seen in individuals.

11.3.1 Rational group failures

The following are “rational” reasons as to why group decision making processes may fail:

Informational signals: It is sensible to take into account what others have said in a group deliberation. If you know Jane is knowledgeable and has good judgement, hearing that she supports a project is evidence that can affect your support. But if she is wrong, she can derail the group. Seeing other people make errors can provide “social proof” to an error.

Consider what would have happened if you stated your opinion first. What if Jane trusts your judgement?

Self-censorship: People tend not to give information contradicting their preferred outcome. In one study of over 500 mock jury trials, the experimenters never once observed someone giving information in this circumstance.

Reputational cascades: People might know what is right (or what they think is right), but they go along with the group or certain members of the group due to concern for their reputation or standing.

11.3.2 Group biases

Many of the failures of individual decision making occur in or are amplified by groups. (Many of the below observations are based on single studies.)

Escalation of commitment: Deliberating groups are more likely to escalate commitment to a course of action despite receiving negative feedback about the outcome. They are also more susceptible to the sunk cost fallacy, the consideration of past costs that should be irrelevant to the decision about future action (Stasser and Dietz-Uhler (2001)).

Representativeness: Groups can amplify the representativeness heuristic, where we judge probability based on resemblance or similarity (Stasser and Dietz-Uhler (2001)).

Overconfidence: People in deliberating groups set narrower confidence intervals than individuals around their estimates. As a result, despite their increased accuracy, groups exhibit a similar level of overprecision to individuals. Groups also overplace their judgements. 98% of subjects believed their group judgements were in the top half of all group judgements with respect to accuracy. (Sniezek and Henry (1989))

Shared information dominates: Shared information has a disproportionate effect on group members. If information is distributed so that key material is unshared (held by only a few group members), this can cause deliberating groups to perform worse. Stasser and Titus (2003) describe a study in which experimental participants were asked to consider candidates for the president of the student government. When individuals had all information, the group chose the best candidate 83% of the time. When information was unshared, and participants’ personal information suggested an inferior candidate was better, the groups rarely chose the best candidate despite having all the information required for a good decision distributed between them.

11.3.3 Reductions in bias

Deliberating groups can temper some biases. Here are two examples.

Availability: Groups tend to rely less on the availability heuristic (we discussed this heuristic on page 2.5). The availability heuristic is tempered possibly because the group members have different memories. Across the group the available memories may be more realistic. However, groups can also be subject to availability cascades. An idea held by one person can spread through the group, eventually producing a widespread belief.

Hindsight bias: Groups tend to have reduced hindsight bias, with groups more likely to recall accurately their original prediction than individuals. (Stahlberg et al. (1995))

For further reading on group biases, see Sunstein and Hastie (2014) and Sunstein and Hastie (2015).

11.4 Groupthink

Newell et al. (2022) offer the following analysis of the concept of groupthink:

Janis (1972) coined the term ‘groupthink’ to describe the type of decision making that occurs in groups that are highly cohesive, insular, and have directed leadership. Through the detailed analysis of a number of historical fiascos (the Bay of Pigs, President Johnson’s decision to escalate the war in Vietnam, President Truman’s decision to do the same in North Korea) and a comparison with major decisions that were successful (the implementation of the Marshall Plan after World War II), Janis identified the characteristics of groupthink-affected decision making: ‘groupthink-dominated groups were characterized by strong pressures towards uniformity, which inclined their members to avoid raising controversial issues, questioning weak arguments, or calling a halt to soft-headed thinking’ …

More specifically, Janis (1972) proposed a model of groupthink with five antecedent conditions, eight symptoms of groupthink, and eight symptoms of defective decision making that result once groupthink has taken grip. The antecedents are those mentioned earlier: cohesiveness, insularity, directed leadership, along with two others – a lack of procedures for search and appraisal of information, and low confidence in the ability to find an alternative solution to the one favoured by the leader. The symptoms of groupthink are:

  • the illusion of invulnerability creating excessive optimism and encouraging extreme risk taking;
  • collective efforts to rationalize in order to discount warnings that might lead members to reconsider their assumptions;
  • an unquestioned belief in the inherent morality of the group, inclining members to ignore the ethical or moral consequences of decisions;
  • stereotyped views of rivals and enemies as too evil to warrant genuine attempts to negotiate;
  • direct pressure on any members that express strong arguments against any of the group stereotypes;
  • self-censorship of doubts or counterarguments that a member might have in order to create an illusion of unanimity within the group; and
  • the emergence of self-appointed ‘mindguards’ who protect the group from adverse information that might shatter shared complacency about the effectiveness and morality of the group’s decisions.

The groupthink concept has had a huge impact in both the academic literature … and popular culture (a Google search produced an incredible 7,630,000 hits for ‘groupthink’ (October 2021)). It is easy to understand the appeal of such a seductive concept. The eight symptoms of groupthink seem applicable to a vast variety of decision contexts, and appear to provide a useful framework for thinking about how defective decision-making arises. Indeed, two of us were so struck by the similarities between the groupthink symptoms and the characteristics of the decision-making leading up to the invasion of Iraq in 2003 that we wrote to the Psychologist magazine stating that, ‘at the time of writing (March 2003) it may just be the heads of state and government who are the victims, but if war results many more heads may be lost to groupthink’ (Newell & Lagnado, 2003, p. 176).

The groupthink model has also inspired research on biased information seeking in groups. In an illustrative study, Schulz-Hardt et al. (2000) found that the tendency to seek information that supports rather than conflicts with a chosen alternative was exacerbated in groups comprised of like-minded individuals. Specifically, the more group members who had chosen the same alternative prior to a group discussion, the more strongly the group preferred information supporting that alternative. It is easy to see how this pattern fits with predictions from the groupthink model. If groupthink leads to the suppression of debate and counterargument then it is quite plausible that members will fall prey to a self-confirming information-search strategy …

However, the seductiveness and applicability of the groupthink concept may also be its weakness. Aldag and Fuller (1993; Fuller and Aldag, 1998) have questioned whether groupthink is merely a convenient and over-used label and argued that direct empirical support for the groupthink model is almost non-existent. In a characteristic quote they stated,

Our contention is that even the most passionately presented and optimistically interpreted findings on groupthink suggest that the phenomenon is, at best, irrelevant. Artificially gathering a sampling of decision-relevant factors into a reified phenomenon has only resulted in the loss of valuable information.

(Fuller and Aldag, 1998, p. 177)

The lost information that Fuller and Aldag refer to is the advances that they suggest could have been made in understanding deficiencies in group decision making if so much research had not been constrained to fit within the groupthink framework. Fuller and Aldag go as far as suggesting that some researchers have unwittingly acted as virtual ‘mindguards’ of the groupthink concept. Ironically, the groupthink concept has induced groupthink amongst groupthink researchers!

The truth about groupthink probably lies somewhere in between the colourful language used by Fuller and Aldag (1998) and the rhetoric of its proponents. There is little doubt that the concept has proved to be very useful in focusing attention on the potential flaws of group decision making …, but the time to move on to testing other potential models of group functioning … and to break free from the constraints of groupthink theorizing is almost certainly long overdue.