This chapter will discuss

Download 214.97 Kb.
Date conversion15.05.2016
Size214.97 Kb.
1   2   3   4   5

Approaches to Predicting a Group's Decision

Comparing Positions 3 and 4 with Position 5 helps explain the structurational viewpoint.

As you can recall, Position 3 involved the persuasive arguments theory. Position 4, in contrast, combined the social comparison and persuasive arguments proposals. In both positions, to look at what members think before group discussion is important. Scientists hypothesize that by looking at these early opinions and arguments, they can predict a group's decision.

To see how Position 5 differs from Positions 3 and 4, we will look at the way in which Positions 3 and 4 would predict a group decision. For example, Jack and Martha are discussing the choice dilemma of whether their town should quickly start recycling plastic garbage or wait a year. Various odds indicate whether the plan would succeed financially, but the idea of starting soon is attractive.

Comparison with Position 3
Persuasive arguments theory, from Position 3, predicts that Jack and Martha will openly discuss the arguments they know. It does not matter if the arguments are for or against the option they prefer. If they have thought of it, they will say it. Therefore, discussion will faithfully represent all of Jack's and Martha's arguments. At the end of discussion, Jack and Martha will have shared the arguments that have come into their minds. Jack will know everything that Martha thought about, and vice versa.

According to persuasive arguments theory, a scientist could interview Jack and Martha before their discussion and use the information to predict their decision. Through such an interview, the researcher would know Jack's and Martha's arguments. In essence, he or she would know beforehand what Jack and Martha would eventually know. The scientist could see that, for example, Jack and Martha together have more arguments for caution than for risk. Therefore, the scientist could predict a cautious decision. If together they had more risky arguments than cautious ones, the scientist could predict a risky decision.

Comparison to Position 4
As you can recall, Position 4 combines the social comparison and persuasive arguments theories. It does not hypothesize that Jack and Martha would tell all the arguments they think of to each other. Instead, it predicts that Jack and Martha will focus on the arguments that support their favorite option. In this way, discussion becomes biased. Jack and Martha's discussion will not include all the information they know. Hence, at the end of their meeting, they will know only the arguments that have come up in discussion.

Let us again say that a scientist has interviewed Jack and Martha before their meeting. If this scientist works under the Position 4 proposal, will he or she be able to predict Jack's and Martha's final decision? Yes. The proposal assumes that this early knowledge can help predict a decision, although doing so will be a little more difficult than it would for someone working only with the persuasive arguments theory.

Hence, under both Positions 3 and 4, a scientist should be able to predict what Jack and Martha will decide by finding out what they are thinking before they talk with each other.

Position 5
Structurational theorists do not look at Jack's and Martha's prediscussional ideas to predict their decision. Do they feel such ideas have no effect on the discussion content and final decision? No. They say that Jack's and Martha's prediscussional arguments and opinions do have an effect, but they view this effect in their own way.

Previously we saw the idea that discussion comprises the arguments that Jack and Martha have thought of before they talk with each other. Position 3 claimed that the discussion faithfully includes the ideas; Position 4 hypothesized that the inclusion was biased. Nevertheless, both theories saw discussion as a way to compile ideas. Structurational theorists do not believe this. They see group discussion as a strategic process. Jack and Martha use communication strategies to support some viewpoints and dismiss others. They use tactics that either reinforce or refute arguments about plastic recycling.

Before Jack and Martha discuss their choice dilemma, a scientist could measure the arguments that both of them can make about their group's options. The scientist could not, however, use the same method to measure the argumentative strategies that Jack and Martha use when they talk together. Therefore, a structurational theorist who wants to predict Jack's and Martha's decision will not focus on their prediscussional ideas but on the content of their discussion.

Research into Structurational Theory

How do the structurational and persuasive arguments approaches compare? Meyers (1989a, 1989b) expressly designed research to examine these two approaches to group polarization. In her study, participants initially made individual decisions about three choice dilemmas. They then wrote down all the arguments they could think of in support of both risky and cautious answers to each dilemma. Next, the participants met in groups, discussed the dilemmas, and came to group decisions. Finally, each person made one last, individual decision about each dilemma.

How did the two approaches compare? Which was the most effective at predicting the outcomes?

Effectiveness of Persuasive Arguments Approach
Meyers found several results that were inconsistent with persuasive arguments theory. First, this theory would hypothesize that participants' arguments during group discussion were similar to the ones they had written on their lists. This was not what Meyers actually found, however. She discovered that many arguments from the participants' lists were omitted from group discussion and that group discussion included many new arguments.

Second, persuasive arguments theory would hypothesize that a researcher could have used the participants' lists of prediscussional arguments to predict how much each group would polarize after discussing the choice dilemmas. This was not the case, however. Last, the theory would predict that the arguments that persuaded each participant most during group discussion were the ones that were new to that participant. The hypothesis is that "novel" arguments are persuasive. Meyers measured the novelty of an argument by counting the number of times participants wrote it on their lists. Through her calculations, Meyers found this hypothesis false. Arguments tended to be slightly more persuasive if they were less novel.

Effectiveness of Structurational Theory
How well can structurational theory explain Meyers's findings? It can account for some, but not all, of her results. There are inconsistencies. For instance, structurational theory would predict that many arguments from the participants' lists would not show up in group discussions. The theory cannot explain, however, any better than persuasive arguments theory, why the participants brought up new arguments.

Another inconsistency that Meyers found relates to the way structurational theory predicts a group's decision. As you can recall, the theory makes certain claims about how someone could predict the amount a group will polarize after it discusses a choice dilemma. It asserts that a scientist can predict this amount by looking at the number of arguments the group discusses, both for and against risk and caution. Meyers found evidence to support this idea in some tests, but not in all.

Overall Results
The results of Meyers's research imply that persuasive arguments theory alone cannot account for the group polarization effect. Of course, we have already shown this. We discussed the problems and failures of persuasive arguments theory earlier in this chapter.

As Meyers herself said (1989a, p. 376; 1989b, p. 129), however, neither do her findings clearly support the structurational view. Thus, it is unclear whether we need a theory such as the structurational approach to explain the group polarization effect. We may not require an explanation that gives group discussion such a large role in the social influence process.


We have looked at Meyers's results in terms of the persuasive arguments theory and structurational theory. We have seen that neither completely explained her findings. What about our combined proposal from Position 4? How well can it explain Meyers's results? Can it explain the findings that persuasive arguments theory alone cannot?

For example, Position 4 can explain why some arguments in participants' lists did not come up during discussion. These arguments were not "reasons for" the participants' prdiscussional choice. Position 4 cannot explain why some arguments occurred during discussion that were not on the participants' lists, but neither can any of the other positions we have discussed.

In addition, researchers have performed some studies on how group members "utilize" information during discussion. This work can provide an explanation of why novel arguments are less persuasive that is consistent with Position 4. We will be discussing this research in Chapter 12, "Decision Theory."

What is clear is that we know of no single theory about group polarization that is absolutely successful. No one theory explains everything about how social influence works when groups discuss choice dilemmas.

Figure 7.5 diagrams the social influence process from the standpoint of structurational theory.











"CORRECT" ------>













Table 7.1 summarizes the five positions we have proposed to explain social influence. It also includes a list of the theories relevant to the group polarization effect and indicates the position each exemplifies.

Table 7.1

Summary of the positions toward social influence



Relevant Theory


No social influence. Group decisions combine individual prediscussional opinions.

Social decision schemes


Prediscussional opinion is influenced by exposure to other member's opinions.

Social comparison


Prediscussional opinion is influenced by arguments raised during discussion.

Persuasive arguments


Prediscussional opinion is influenced by both arguments and exposure to opinions.

Persuasive arguments/ Social comparison combination


Postdiscussional opinion is formed through group discussion. Pre- and postdiscussional opinions may be unrelated.


Universal Theory of Social Influence

We have covered many theories and research studies that apply to a specific type of decision, the choice dilemma. As we have shown, the group polarization effect was an exciting and surprising discovery for researchers who looked at choice dilemmas. We have revealed part of the large volume of work that now surrounds the topic.

Scientists also, however, have moved beyond polarization itself. They have attempted to extend the theories we have discussed to completely different group phenomena. Researchers wish to apply these proposals to more general phenomena as they attempt to find a universal theory of social influence in groups. One example of a theory to explain social influence in general is the valence model.

Description of the Model

Hoffman (1979) proposed the valence model. It attempts to account for all group decision-making behavior and is based on Lewin's conception of forces in a "life-space" (discussed in Chapter 3). The concept is that forces propel a group either away from or toward alternative courses of action, which exist within the group's life-space. Each course of action has a valence, which can be either negative or positive. If an alternative generates a negative valence, it repels the group from its region of the life-space. If it sends out a positive valence, it attracts the group toward it.

A group will choose the course of action that has the greatest positive valence. This means that the choice has a high level of group acceptability. This will occur, however, only as long as the alternative has surpassed a minimum threshold of group acceptability. If no course of action reaches this threshold, the group will fail to reach a decision. If two or more alternatives meet the group's standards, two results could occur. The group could find the solution with the greatest valence most attractive, or the group could progress into conflict concerning the different choices.

The Structural Perspective

Hoffman and his associates held viewpoints that were consistent with the structural perspective. They believed that group discussion can (1) create valences and (2) reflect the relative amounts of valences. First, discussion can create the valences of the courses of actions. For example, Harry belongs to a fishing group. The group comes together and does not really have an opinion about morning fishing or evening fishing. The members begin to talk, and soon the alternative of evening fishing becomes less attractive. They convince one another that morning expeditions are better. Thus, the group has given the alternative of evening fishing a negative valence, and it has given morning fishing a positive valence.

Second, group discussion reflects the relative amounts of valence among alternative courses of action. How did Harry and his group reach their decision? It may be that evening fishing had a negative valence for some members before they came to the group. Perhaps they had poor fishing results in the evening. For whatever reason, the negative valence of evening expeditions may have influenced the group as the members talked about when they would fish.

Analyzing Valence in Groups

Hoffman believed that he could measure valence by coding communicative acts in a group according to which alternatives they describe and whether the communication supports or rejects the alternative. For example, a group is deciding the sports activity for which they should get tickets. Mary says that she wants to go to a baseball game. Jerry replies that he thinks baseball is boring. Cindy says she likes basketball. Three communicative acts have occurred in this group. We can code Act 1 as being about the alternative of seeing a baseball game and that it supports the alternative. Act 2 also refers to baseball and rejects it. Act 3 describes the alternative of basketball and supports it.

Hoffman and his co-workers performed a series of studies to investigate their ideas. They analyzed decision-making group interaction and used this analysis to evaluate their model.

Studies of the Model

Hoffman's first experiment consisted of 19 groups that considered alternative courses of action. The researchers analyzed the group discussion to discover the valences of the options. They measured the valences by subtracting the number of statements opposing the courses of action from the number of statements supporting them. For example, in the group described above, suppose Cindy also says that she does not want to go to a baseball game. In that case, the baseball game would have a valence of -1. Mary's statement supported it, but both Cindy's and Jerry's statements opposed it. One positive statement minus two negative ones equals a valence of -1.

The researchers found that the valences ranged from -20 to +96. They further discovered that the groups did not accept alternatives that had valences of less than 15. Thus, the experimenters tentatively accepted 15 as the adoption threshold value. Other results supported this conclusion. In 6 of the 19 groups, only one alternative reached this threshold level, and the group adopted it. In 11 of the remaining groups, two or more courses of action reached threshold. How did these groups choose which option to adopt? Ten of the 11 chose the alternative with the most valence. The eleventh combined features of two alternatives that achieved threshold. The remaining 2 groups of the 19 behaved contrary to prediction.

In this and later studies, Hoffman divided group discussion into sections to analyze how alternatives gained and lost valence over time. Hoffman concluded that group discussion involves a three-stage process. In the first stage, the group weeds out alternatives that are clearly unacceptable and forgets them. Groups spend little time on these unacceptable courses of action. They usually ignore alternatives after their valence drops below -8. Thus, Hoffman accepted -8 as the rejection threshold value.

This first stage ends when one alternative reaches threshold. During the second stage, the group compares the alternative that reached the threshold with the other courses of action. Members are usually favorable toward the alternative, which typically causes the valence of that option to rise sharply at about this time. In Hoffman's groups, the first alternative to reach threshold became the group's choice 80 percent of the time.

In the last stage, the group appears to decide in favor of this threshold alternative and continues to speak favorably about it, which further increases its valence. In contrast, the group ignores the other alternatives. This continued discussion seems to function in two ways. First, it helps to increase the attractiveness of the choice and is thus a cause of the group's commitment to the action. Second, the discussion expresses this commitment. (Refer to B. A. Fisher's model of group interaction in Chapter 8, "Group Process," for a parallel conception.)

Criticisms of the Model

The structurational theorists (McPhee, Poole, and Seibold, 1981) accepted Hoffman's basic premise that group decision making relates to interaction in two ways. Interaction helps accomplish group decisions, and interaction reflects those decisions. However, McPhee et al. criticized the valence model on two points.

A New Threshold Level

First, they pointed out that Hoffman's choice of a threshold level was arbitrary. They concluded that the best way to discover the threshold was to find the exact valence that best differentiated the alternatives the group chose versus those it rejected. In other words, the threshold should be set at a point where the probability is greatest that the group will choose an alternative with that or greater valence. At the same time, the probability should also be greatest that the group will reject an alternative if it has a valence below the threshold mark.

The research of McPhee et al. revealed that Hoffman had overestimated the proper threshold. They studied ten groups, whose members discussed a total of 110 alternative courses of action. McPhee et al. measured the valences of each option. The groups gave 10 options valences close to +6 but just below that value. The groups eventually chose only one of those 10. The groups also, however, gave 10 alternatives valence values close to +6 but just above that mark. Of these 10, the groups chose 7. Thus, +6 is the best estimate of the threshold level for valences, not Hoffman's estimate of 15.

A New Method of Measurement

Second, McPhee and his associates objected to Hoffman's claim concerning how best to measure a valence. Hoffman measured valences by calculating the difference between the number of positive and negative statements that groups made about a course of action. Hoffman performed his measurements by looking at the statements that came from the group as a whole. This implies that he looked at all statements, no matter who said them.

McPhee et al. questioned this methodology on the grounds that one group member might make many positive comments about a course of action and that all other members might each make only one negative statement about the same alternative. For example, Mary might make five positive statements about baseball, and Cindy and Jerry give only one negative opinion each. Would the group then go to a baseball game simply because Mary made more comments? Probably not. The group still contained a majority that did not want to go to a baseball game.

McPhee et al. decided that there was a better way to predict group decisions. They believed a model should account for the distribution of valence among group members. In other words, the group should choose the option that the greatest number of members have positive valence for. In our example, Mary's comments indicated positive valence for baseball, but Cindy's and Jerry's implied negative valence. The valence distribution for baseball is, then, negative, and baseball is unlikely to be the group's choice. If in contrast Cindy and Jerry had made mostly positive comments for basketball whereas Mary had mostly spoken negatively, then basketball would have a positive valence distribution.

There is a problem with the valence distribution approach. Let us imagine that two out of the members had spoken positively for both basketball and soccer. Which one would be the group's choice? McPhee et al. decided that the option with the largest total valence would be the group's choice.

McPhee et al. compared their model, consisting of valence distributions across members, along with total valence as a "tie-breaker" as just described, with Hoffman's model, which was based on total valence alone. Based on an analysis of their ten groups, they found their model to predict group decisions more accurately than Hoffman's. Several implications flow from this finding. One is that one member cannot sway an indifferent majority. One person may strenuously support an alternative, but if he or she faces an indifferent majority, the group is not likely to choose that course of action. A second implication is that the group tries to follow unanimous decisions. If almost everyone in a group supports an alternative but one person rejects it, the group is less likely to choose that course of action than an alternative that the group unanimously accepts.

1   2   3   4   5

The database is protected by copyright © 2016
send message

    Main page