The research on minority influence points to a number of conclusions. First, if minorities are to influence majorities, they do so by persuading majority group members. Second, minorities are more persuasive when they act consistently, both in their expressed position and in their agreement with each other. And, third, tasks for which there are demonstrably correct answers present a better opportunity for minority influence than those that do not have a demonstrably correct answer. The mechanism by which minority influence occurs, that is, the factor(s) that mediate the relationship, has not been addressed frequently or definitively. Many minority group experiments are conducted on ad hoc groups, do not code the messages delivered during discussion, or both. Exceptions exist. Meyers, Brashers, and Hanner (2000) examined the argument patterns used by successful and unsuccessful minority and majority subgroups, and found that minority and majority groups’ patterns of argument differed, as did successful and unsuccessful minority factions. Van Swol and Seinfeld (2006) examined the types of information provided by minority and majority group members in discussion, finding that successful minorities discussed more common information than successful majorities. Future research would benefit from examining these minority communication processes in greater depth.
Individual Differences
* * *
There is considerable research on individual differences that can lead to a person being an influential minority member, although it is not usually referred to as minority influence. Instead, it is generally termed opinion leadership. The study of opinion leadership has a long history in the social sciences, beginning with early studies on the role of influential network members (Katz & Lazerfeld, 1955; Lazerfeld, Berelson, & Gaudet, 1948) and continuing through modern diffusion scholarship (Rogers, 2003). Harnessing the power of opinion leaders is an effective means of affecting behavior change within a population, with examples ranging from campaigns to increase safe sex practices (Kelly et al., 1992; Miller, Klotz, & Eckholdt, 1998) and mammogram testing (Earp et al., 2002) to encouraging the implementation of different treatment practices among physicians (Soumerai et al., 1998).
Because of conceptually amorphous definitions of what constitutes opinion leadership, and because many of the most effective means to identify influential network members require an arduous and expensive data collection process (see Valente & Pumpuang, 2007, for a review), the usual methods by which campaign designers go about trying to identify opinion leaders is varied and inefficient. In response to these problems, Boster, Kotowski, Andrews, and Serota (2011) developed self-report measures to assess three dimensions of opinion leadership: the extent to which someone is highly connected within their social network, the extent to which someone possesses unusual persuasive abilities, and the extent to which someone is a subject matter expert on healthy lifestyle issues.
A person high on all three measures (usually designated as being above the 75th percentile) is termed a superdiffuser. Boster et al. (2011) report considerable evidence consistent with the construct validity of these measures (see also Carpenter et al., 2009). Moreover, research has demonstrated that they can be adapted to areas outside of the health domain, such as politics (Serota, Carpenter, Andrews, & Boster, 2009). Finally, a campaign conducted using superdiffusers to diffuse multivitamin usage in a college population provided some preliminary evidence that superdiffusers can be effective diffusing health information, with more students at the experimental campus reporting having heard about people taking multivitamins, and fewer students ceasing to take multivitamins, compared to the control campus (Boster, Carpenter, Andrews, & Mongeau, 2012). Future research would benefit from examining the utility of employing superdiffusers to induce change and the extent to which the use of superdiffusers combines additively or nonadditively with the implementation of a traditional mass media campaign.
Of course, influential members of a network cannot always be relied on to pursue the same goals as the campaign designer. A health superdiffuser’s opinion on a health topic could differ from the stance advocated in the campaign, and such a person could work assiduously to prevent adoption of the campaign recommendations. Evidence of this type of effect has been found in other programs of research. David, Cappella, and Fishbein (2006) conducted an experiment in which adolescents engaged in a chat room discussion regarding an antidrug public service advertisement, hypothesizing that those who were high sensation seekers (and hence more likely to have tried marijuana and have pro-marijuana attitudes) would engage in biased processing, counterargue more, and dominate discussion in a manner that persuades other group members not to accept the campaign message. They found data consistent with their predictions: high sensation seekers spoke more than middle or low sensation seekers, contributed more pro-drug than anti-drug comments, and made more negative comments about people in the discussion group. Furthermore, this behavior resulted in subjects in the chat condition reporting more pro-marijuana beliefs and attitudes, more pro-marijuana normative pressure, and being more likely to believe that they would be ostracized socially if they refused marijuana, compared to the no-chat condition.
Outsiders Versus Insiders
* * *
The question of whether in-group members are persuaded more effectively by a fellow in-group member versus an outsider has been investigated in one form or another for some years. Lewin (1947) found that women who were part of decision groups were substantially more likely to serve organ meats than those who heard a lecture urging them to do so (32% v. 3%). Mackie and colleagues (Mackie, Gastardo-Conaco, & Skelly, 1992; Mackie, Worth, & Asuncion, 1990) have also found evidence that in-group members are more persuasive than out-group members, but the effects of message source combined nonadditively with argument quality. Subjects were persuaded more by a relevant, strong message from an in-group source than a relevant, weak message (and not persuaded at all by messages from an out-group source). When the message topic was irrelevant to the group, however, there was a main effect only for group status.
Additional evidence for the persuasiveness of in-group members comes from studies building on self-categorization theory (Turner, 1987, 1991). Haslam, Jetten, O’Brien, and Jacobs (2004) found evidence that information provided by in-group members affected how subjects anticipated a potentially stressful upcoming event (a mental arithmetic task), although information from out-group members did not. When an out-group member described the task as either stressful or challenging, subjects reported equal levels of stress. When an in-group member was the source, however, subjects reported feeling substantially less stress when the source said the task was challenging compared to when they described it as stressful.
Drawing on social comparison theory and self-categorization theory, Haslam et al. proposed that information from in-group sources was viewed as more valid than information provided by out-group members, as the in-group member was more qualified to provide information regarding social reality. In a conceptually similar experiment, Platow et al. (2005) found evidence that social proof in the form of canned laughter was more influential on subject behavior when it was attributed to in-group members versus out-group members. Subjects listening to a stand-up routine were more likely to laugh, smile, laugh for a longer duration, and rate the audio recording more positively if they were led to believe that the people on the prerecorded laugh track were in-group members compared to out-group members. Platow et al. (2006) extended the effect to a physiological dependent variable in a pain tolerance experiment that required subjects to hold their hands in ice water. Subjects who had received assurance from an in-group member (vs. out-group member) that the second experimental trial was easier than the first demonstrated significantly lower physiological arousal during the second trial (as measured by galvanic skin response).
Out-group members are not always ineffective persuaders, however. Wilder (1990) found evidence that differentiating out-group members, such as by providing information to subjects about out-group members’ hometowns, majors, and names, inc
reases their persuasive impact to the same levels as in-group members. So, although in-group members are usually more persuasive than out-group members, there are conditions under which in-group members will be relatively ineffective, and out-group members will be relatively effective.
Groups Persuading Non-group Members
* * *
Whether or not groups persuade non-group members is a question that first entails making a conceptual distinction, albeit one that is very difficult to make in practice. A group member can persuade a non-group member, independent of group affiliation—indeed, without the target even knowing the group membership of the source. Conversely, a group member could also persuade a non-group member, but with the source’s group affiliation highly salient to the target (recognizing that salient target group affiliation is not necessarily the cause of persuasion, if persuasion does occur). Finally, the collective messages or actions of a group can persuade a target (recognizing that the members of the source group can vary in the extent to which they are differentiated as individuals to the target).
It is on this last area that the remainder of this section focuses. A good example is the research on the intra-audience effect, conducted by Hocking, Margreiter, and Hylton (1977) as a test of an idea proposed by Hylton (1971). Students were asked to attend a bar on one of two nights as part of a bogus experiment. Thirty confederates attended the bar on both nights. On the first night, the confederates exhibited little reaction to the band that was playing. The confederates talked among themselves, and did not seem to pay any attention to the band, nor react in one way or another to the songs that were played. The following night the same confederates acted very differently to a performance by the same band—they danced, applauded loudly, and cheered for an encore. The students asked to go to the bar were queried during the next class meeting for their evaluation of the band. As hypothesized, those students who had attended on the night when confederates exhibited positive reactions to the band rated the band far more positively than those students who had attended the bar on the night that the confederates exhibited indifference to the band.
Although the mediator(s) of the effect could not be specified, Hocking et al. (1977) proposed that the nonverbal messages of the confederates led to the differences in band evaluations. Notably, Festinger’s social comparison theory (1950, 1954) could be applied to the case of intra-audience effects as well. Festinger theorized that when people are unsure of the accuracy of their beliefs, they are motivated to bolster them by consulting physical reality or by comparing themselves to others. The latter is more likely to occur when an objective, nonsocial means of comparison is not available. Other people are viewed as an accurate source of information, and informational influence can occur as a result of this comparison process. (It should be noted that research since Festinger’s original formulation has uncovered a variety of other motivations for engaging in social comparison.) In the case of the Hocking et al. (1977) experiment, subjects were asked to form an attitude for a stimulus (a band) for which nonsocial, objective means of determining quality did not exist. Thus, the subjects looked to the opinions of others, and surmised from the enthusiastic nonverbal messages of the confederates that the “correct” belief was a positive endorsement. It is also possible to view the results of Hocking et al. in light of the research by Haslam et al. (2004) and Platow et al. (2005, 2006) discussed previously. Perhaps the students in the bar viewed themselves as being in the same in-group as the other bar goers, making them more likely to be influenced by intra-audience effects due to self-categorization and social comparison.
Directions for the Future
* * *
Four aspects of persuasion involving groups have been examined in the previous sections. Two dealt with persuasion within groups (choice shift, minority influence), one with the relative persuasive impact of minority versus majority members, and one with the ability of groups to influence individuals not affiliated with the group (intra-audience effects). This final section summarizes these corpuses and provides suggestions for future avenues of research.
The Choice Shift
The choice shift is a robust phenomenon, replicated frequently. The effect is consistent as well; certain experimental materials produce risky shifts and others produce cautious shifts across experiments. Thus, the focus of scholarship has been on theoretical frameworks to explain the findings. Numerous explanations have emerged over the last 50 years; some grounded in normative influence and some in informational influence. There has been little effort to integrate them, however, despite their pronounced similarities.
An integrated theory would have to address the relationship between informational and normative influence. Both normative and informational influences operate in choice shift experiments, and future theories must recognize this fact to advancement knowledge in this area. Evidence thus far suggests that informational and normative influence combining additively might be inadequate. Furthermore, testing such a model would be contingent on the ability to distinguish clearly between informational and normative influence, a task that has proven elusive to date.
Finally, the external and ecological validity of traditional choice shift experiments must be addressed if the claims of applicability proposed in many of the discussions of the phenomena are to be justified. Most experiments on the topic are conducted using students who are of the same status, have similar backgrounds, are not highly involved in the topic, and are making decisions in the absence of stress and time pressures. The contexts in which the choice shift is often offered as an explanation—the Cuban Missile Crisis, the Bay of Pigs invasion, the escalation of the Vietnam War—did involve those potentially important moderators. If, or how, those factors affect group decision making will not be understood until experiments are conducted that can approximate them convincingly.
The Influence of Minorities
Much like the choice shift, research on the influence of minorities would benefit from theoretical integration. Such a theory would have to account for the balance of normative and informational influence used, with majorities using the former more often than the later and minorities using the later more often than the former. The specific mechanisms by which the majority and minority exert normative and informational influence also remain unclear. Finally, an integrative theory would also have to account for the variety of persuasive outcomes unearthed in minority and majority research, such as direct influence, indirect influence, and delayed influence.
Similar to the choice shift, minority influence studies have many of the same external or ecological validity challenges. For instance, in natural groups, minorities have characteristics that are not always modeled in experimental situations. For example, minorities might know the majority’s position better than the majority knows the minority position (Robinson & Keltner, 1996), giving them an initial persuasive advantage. Similarly, many experimental settings may occur too quickly to let nuances of minority influence unfold. Although majorities have the initial advantage when it comes to normative influence, minorities may be able to increase their ability to influence via normative means by silencing some members of the majority (Noelle-Neumann, 1993), thereby decreasing the majority’s primary advantage.
Other factors change during discussion as well. Minorities espousing consistent positions throughout discussion are more persuasive than those who do not, a result that emerges from majorities viewing minorities’ consistency positively. Because it takes time to demonstrate consistency of opinion, a majority’s understanding of a minority position may be seen as less radical after the majority hears the complete minority explanation. This point is particularly applicable when the majority has preconceived notions of the minority position. In a study examining preferences of English faculty in selecting works for an introductory course syllabus, Robinson and Keltner (1996) found that traditionalists overestimated the amount of change advocated by revisionists, who in fact selected many works from the traditional literary canon.
In a discussion format if the majority impression of the minority changed over time, it would affect the majority’s view of the reasonableness of the minority.
In-Group Versus Out-Group Sources
Research on in-group and out-group sources has generated interesting findings about when either group will be influential. Disparities between experiments in how group identity is induced, however, complicate matters. Many experiments adopt Turner’s (1982) definition of in-groups as “two or more individuals who share a common identification” (p. 15), and what constitutes an induction of common identification varies considerably across experiments. Many experiments use relatively trivial common identifications, such as university affiliation (Mackie et al., 1990, 1992; Wilder, 1990). Although typically effective at inducing perceptions of group identity, such inductions are not isomorphic with the level of in-group identity one would feel with one’s immediate family or a long-standing work group, and the extent to which these differences affect results is not entirely clear. Finally, as with the previously discussed research streams, the area would benefit from theoretical developments that specify when one would expect in-groups to be persuasive and when we would expect out-groups to be persuasive.
Intra-audience Effects
Little research has been conducted on intra-audience effects since the work of Hylton (1971) and Hocking et al. (1977). In addition to the need for more research in the area, such experiments would be aided by a clearer conceptual development of the term “intra-audience effect.” One question stemming from this ambiguity involves the difference between intra-audience effects and emotional contagion (Hatfield, Cacioppo, & Rapson, 1992). One distinction between the two might be the extent to which intra-audience effects are characterized as verbal and nonverbal signals that are interpreted cognitively by the receiver as information about the social environment. Conversely, the emotional contagion explanations focus on imitation, below conscious awareness, and the involvement of uncontrollable processes such as mirror neuron activation (see Iacoboni, 2009, for a review). Another distinction contrasts intra-audience effects based in informational influence versus the normative influence a group might have on a person’s evaluation of an attitude object, perhaps because of a desire to be perceived as a good group member. As with other steams of research discussed in this chapter, whether these distinctions could be measured accurately is just as important to advancement as whether the distinction can be made conceptually.
The SAGE Handbook of Persuasion Page 74