Stimulated by the work of Moscovici (1985) on minority influence, social psychologists have shown increasing interest in how minority messages shape a target’s opinion. The typical study on minority influence presented a minority source advocating a single influence appeal to a lone target (Wood, Lundgren, Ouellette, Busceme, & Blackstone, 1994). This research tradition thus focused on the reception-side of social-influence, though in some recent work minority preparation for group discussion has been considered (Levine & Russo, 1995; Van Hiel, in press; Zdaniuk & Levine, 1996), as well as minority influence in freely interacting groups in which participants are subjected to reciprocal influence (Gruenfeld, Thomas-Hunt, & Kim, 1998; McLeod, Baron, Marti, & Yoon, 1997; Schulz-Hardt, Frey, Luthgens, & Moscovici, 2000; Smith, Tindale, & Dugoni, 1996; Tindale, Vollrath, Davis, Nagao, & Hinsz, 1990).

Levine and colleagues (Levine & Russo, 1995; Zdaniuk & Levine, 1996) studied how membership of factions of varying sizes affected acquisition of issue-relevant information by prospective group members in six-person groups. The results revealed a positive, linear relationship between faction size and bias toward supporting information (Zdaniuk & Levine), although Levine and Russo reported more bias in one versus two person minorities, and four versus five person majorities. Van Hiel (in press) found greater information acquisition bias of two-person minorities when task relevance was low, and concluded that minority members show greater bias when they are either not enthusiastic and interested, or insufficiently motivated.

Schulz-Hardt et al. (2000) reported that group members prefer supporting information, which resulted in greater group discussion bias when more group members supported a particular decision alternative. Smith et al. (1996) and Tindale et al. (1990) found that dissent reduced the degree of polarization in majority members, and Gruenfeld et al. (1998) reported that after group interaction, participants in majority factions exhibited greater cognitive complexity than did minority members. Finally, McLeod et al. (1997) found that anonymity increased the expression of minority arguments, but decreased the effectiveness of these arguments in influencing majority opinions.

However, previous researchers who considered minority actors of social influence did not focus on the question of how minorities might raise their effectiveness in shaping majority opinions. Van Hiel (1998) started from the assumption that effective factions take advantage of factors known to maximize social influence. Thus, application of this “looking glass self” metaphor to the fact that effective minorities direct the target’s attention towards informational resources (e.g., Moscovici, 1985), would lead to the expectation of greater information orientation of minorities when they actually try to persuade a target. This expectancy was corroborated by Van Hiel’s (1998) finding that effective minorities, as compared to effective majorities, generated a greater amount and diversity of arguments during group discussion.

In the present study, the source of social influence was also stressed. In particular, the pivotal role of behavioral style was emphasized and participants’ self-reports on the effectiveness of social-influence strategies in persuading the opposing faction were analyzed. Moscovici (1985) asserted that consistent style is a key determinant of minority effectiveness (see also Mugny, 1982). Consistency, defined as holding on to a position over time and modality, is assumed to enhance influence. Rigidity, a variant of consistency in which the influence source is perceived negatively, is thought to reduce influence. Bassili and Provencal (1988) extended this research and studied a broader range of minority characteristics. They administered adjectives that refer to the Big Five personality dimensions and found that minorities are perceived as more assertive, less agreeable, more honest, more competent, and more consistent than majority members. Bassili and Provencal (1988) concluded that the general picture of minorities that emerged from their findings is “one of strength, conviction, integrity, and consistency” (p. 12). These factors have previously been related to minority effectiveness and, bearing in mind the assumption that effective factions take advantage of factors known to maximize social influence, the present authors hypothesized that assertiveness, honesty, competence, and consistency would be considered effective for minorities, whereas agreeableness would be considered as effective for majorities. Flexibility versus rigidity was predicted to be considered effective for minorities.

The central aspects of the present procedure were analogous to those of Levine and Russo (1995) and Zdaniuk and Levine (1996). Participants were invited in groups of 15-17 and individually completed the experiment on a personal computer. They were told that they would have a small group discussion on the controversial issue of liberalization of softdrugs, which would end whenever group members reached unanimous agreement. They were told also that they were assigned to a 4-, 5-, or 6-person discussion group (in reality, always a 6-person group) on the basis of their answers on “the drug questionnaire” which was administered two weeks before the experimental session. This questionnaire featured nine items that approved or disapproved of the legalization of softdrugs (Cronbach’s alpha = .74).

In the beginning of the experimental session, the participants received the following instruction (written in Dutch): “In the first phase of the experiment, you have to prepare yourself for the forthcoming group discussion. You will read some articles that favor or oppose the liberalization of softdrugs.” In the competitive orientation condition, participants were told, “After the preparation phase, your group first has to decide whether a pro- or contra-legalization pamphlet will be written. Everyone will have to agree with this initial decision. Afterwards, your group will compile the best arguments in favor of the position advocated by the pamphlet.” In the cooperative orientation condition, participants were told, “After the preparation phase, your group has to write a pamphlet on the liberalization of softdrugs. Afterwards, your group will compile the best arguments that show the positive and negative sides of the legalization issue. The pamphlet thus has to reflect a balanced mix of pro and contra arguments.” Participants were then informed about their group. In the minority condition, each participant learned that one person agreed and four people disagreed with his/her opinion. In the majority condition, each participant learned that three people agreed and two people disagreed with his/her opinion. In order to enhance the salience of the manipulation, participants were asked to write down the composition of their group on a separate sheet. They subsequently read the articles on liberalization of softdrugs which were presented on a computer menu. By pressing the key “C” (contra) or “P” (pro) participants were able to choose one of the three articles that argued either against, or in favor of, liberalization. The article was then depicted on the screen and by pressing the key “ESC” participants were able to return to the menu whenever they wanted to do so. Participants were allotted 10 minutes to read this information, which they were told “is certainly not enough to read all of it. You therefore have to choose carefully what information you want to read.” They were also given two sheets of paper (a “pro and contra sheet”) on which they could write down the most interesting facts. Finally, participants filled in a paper and pencil questionnaire that contained the main dependent variables.

The items were administered on 7-point scales anchored by (1) certainly not effective and (7) certainly effective. Participants first rated the effectiveness of each item for their own faction, and next for the opposing faction.

The questionnaire designed by Bassili and Provencal (1988) was used, although various adaptations were made to improve its psychometric features. That is, by using this adaption we tried to address four problems with respect to the original questionnaire. First, items referring to assertiveness as well as agreeableness loaded on Bassili and Provencal’s first factor and a separate factor-analysis on these items had to be conducted in order to differentiate between the two dimensions. Second, not more than three adjectives (“consistent”, “predictable”, and “persistent”) loaded on the consistency dimension. Third, some items on the agreeableness dimension clearly referred to flexibility (“flexible”, “accepting”, and “tolerant”). Fourth, some items loaded on other than the intended dimension (e.g. “cultured” and “frank” loaded on the honesty dimension – whereas “fair” loaded on the competence dimension).

In order to construct the present questionnaire, the authors selected the most typical items of the Bassili and Provencal (1988) scales and added some new adjectives that closely matched the content of these scales. Selection of these “new” adjectives was based on the Dutch AB5C circumplex personality model (De Raad, Hendriks, & Hofstee, 1992). Items measuring rigidity and flexibility based on the study by Mugny (1982) were also included. In addition, the questionnaire featured a short description of the items (for example, the adjective “intelligent” was described as “wise, sensible, has insight”). The reliabilities of each of these scales for the own-faction and the other-faction ratings were as follows: consistency (“consistent”, “persistent”, “tenacious”, “firm”, “unshakable”) .72 and .75; assertiveness (“strong”, “independent”, “assertive”,“courageous”, “certain”) .75 and .79; competence (“intelligent”, “capable”, “systematic”, “efficient”, “competent”) .78 and .72; honesty (“honest”, “truthful”, “sincere”, “fair”, “unbiased”) .73, and .69; agreeableness (“friendly”, “helpful”, “cooperative”, “social”, “empathic”) .74 and .68; rigidity (“rigid”, “authoritarian”, “stubborn”, “obstinate”, “arrogant”) .79 and .74; and flexibility (“tolerant”, “open”, “indulgent”, “flexible”) .60 and .65.

A 7-point scale aimed at measuring the perceived likelihood that the group would adopt the participant’s position was used as a manipulation check (see also Zdaniuk & Levine, 1996). As one would expect, minority members agreed to a lesser extent with the idea that the group would adopt their position (M = 3.34) than did majority members (M = 4.48; F(1, 66) = 19.69, p < .001). Behavioral styles were analyzed in a 2 (actor; minority member versus majority member) × 2 (orientation; competitive versus cooperative) × 2 (target; own faction versus other faction rating) mixed factorial ANOVA. Minority and majority members’ mean scores on the various behavioral styles are presented in Table 1, along with the F tests on the actor-target interaction effect. These actor-target interaction effects are very interesting because they test whether differences between minorities and majorities generalize across own-faction and other-faction ratings. For example, suppose that minorities would be considered effective when they exhibit high consistency, one would expect a high own-faction rating and a low other-faction rating for minority members – whereas an opposite pattern of results would be expected for majority members.

Participants in the study by Bassili and Provencal (1988) viewed a videotape and were instructed to form general impressions of the persons who acted as a minority or a majority in a four-person group. Despite the fact that the behavior of both factions was controlled for, the general picture of minorities that emerged from their findings is “one of strength, conviction, integrity, and consistency” (p. 12). The present results show that a minority’s strength and firmness are more than only a matter of perception, but extend to the minority members’ self-perceptions of how to persuade other group members. This finding has two important consequences. First, the convergence between the observer’s perceptions and the actor’s intentions seems to indicate that there exists an implicit norm of how one “has to behave as a minority member” and “what is expected of a minority member”. This finding illustrates not only that the perception of consistency and strength is “in the eye of the beholder”, but also that minorities that want to exert influence actually feel they should behave this way. Although the present research does not put us in the position to either confirm or reject minority influence theories because another methodology has been used, the convergence with previous studies further strengthens the theoretical position that behavioral style is an important tool for exerting social influence.

Second, the effects found in classic social-influence studies probably underestimate minorities’ influence capacity in interactive settings. Minorities have two routes by which they can exert influence: not only do they cause the perception of strength in a target, but also they are convinced that firm and consistent behavior is effective. Analogously, effective minorities are said to focus a target on informational resources (e.g., Moscovici, 1985), and actually generate a greater amount and diversity of arguments in interactive settings (Van Hiel, 1998).

Of course, the present study does not illustrate the relationship between ratings of effectiveness and measures of actual behavior. It is possible that group members are wrong in their assumptions about the effectiveness of certain behavioral styles and, of course, actual effectiveness is of greater interest than expected effectiveness. It is also possible that minorities might not be able to behave in – what they would consider themselves – an effective way because of various situational constraints. For example, minority members might behave in a less assertive and consistent way during group interaction because the expression of minority opinions is risky and creates social conflict and fear of rejection (Kruglanski & Webster, 1991). Future researchers should therefore contrast behavioral measures with participants’ subjective theories.

Previous researchers identified rigidity as a counterproductive and extreme version of consistency and argued that effective minorities show some level of flexibility. The present results confirm that in order to exert influence, minorities especially should not exhibit rigidity, but quite unexpectedly, flexibility was judged a more effective strategy for majorities. Moreover, opposed to the idea that rigidity and flexibility are two poles of one dimension, only a modest correlation was found between these two styles (r = -.26 and -.30 for self and other ratings respectively), whereas high correlations (especially for the self-ratings) between flexibility and agreeableness (r = .67 and .37 for self and other ratings respectively) were obtained. The present study thus indicates that flexibility shares greater communality with agreeableness than with rigidity.

Finally, previous researchers illustrated clear differences between the stereotypical roles and the interaction styles of males and females. Males commonly are thought to be informative, dominant, task-oriented, assertive, innovative, objective, and competitive. Females are thought to be passive, noncompetitive, subjective, supportive, and inner-oriented (see e.g., Wood & Karten, 1986). Hence, it does seem possible that traditional gender role expectations could have an effect on women’s viewpoint on how to behave as minority group members. Because of the large number of female participants (87%) in the present sample, one thus has to be cautious about generalizing the present results to males.