Abstract
The first goal of this article is to demonstrate that the dichotomy between informational and normative influence in group decision making is long outdated and should be replaced with a distinction among compliance, comparison, and argumentation influence processes. The second goal of this article is to use this distinction as the basis for a literature review of the impact of various input factors on social influence during small group discussion. The third goal of this article is to present the Simplified Model of Group Social Influence Processes, an interactive input–process–output model relevant to decision-making groups. The article ends with a discussion of the implications of this model for future research and further model development.
The goal of this article is to distinguish between three social influence processes (argumentation, comparison, and compliance) in small decision-making groups while proposing an interactive input–process–output (IPO) model integrating these processes. The formalization of the basic IPO model harkens back to independent proposals by Gouran (1973) in communication and Hackman and Morris (1975) in psychology. In it, a communicator’s relevant cognitions (including beliefs, attitudes, motives, goals, emotions, personality traits), when activated by contextual factors including previous discussion content (input), influence that communicator’s subsequent message (process), affecting a second communicator’s cognitions (output). Research consistent with this model is prevalent in business administration (e.g., Weisband, 1992), communication (e.g., Pavitt, 1994), and psychology (e.g., Kelly & McGrath, 1985). An interactive version of the model includes mutual feedback loops, in which that second person applies his or her cognitions (input) in planning a response (process), impacting the first person’s cognitions (output), message plan (input), message (process), and so on. Such models require some understanding of what occurs during communicative episodes, which is the case in this context.
A large proportion of the research germane to this model is the product of paradigms relevant to (a) the group polarization effect, the tendency for discussion groups to make more extreme decisions than their average member would have made otherwise; and (b) the hidden profile, in which information jointly known by group members as a whole favors one option but that known uniquely by individual members implies a different choice. These paradigms are founded on four assumptions. First, proposal options are known by the members of the group before discussion. Second, group members enter discussion having used their stock of decision-relevant information to formulate a prediscussion preference for a particular option. Third, these members leave discussion with a postdiscussion preference that may have changed as a consequence of discussion. Fourth, the group reaches a decision by, in effect, combining individual postdiscussion preferences. Social influence occurs when individual member prediscussion and postdiscussion preferences differ and/or when the group decision differs from the average of members’ prediscussion preferences. 1 These assumptions often do not reflect occurrences during real-world group decision making, particularly as proposal options are frequently not common knowledge before the fact. Nonetheless, many aspects of the model should still be relevant in those circumstances.
Central to the model are three social influence processes that, although intertwined in practice, are independent in theory. Argumentation processes lead group members to discuss their decision-relevant knowledge and to be influenced by the knowledge voiced by other members. Comparison processes lead group members to voice their decision preferences and to be influenced by hearing one another’s preferences. These latter two terms are borrowed from the two central theories explaining group polarization, persuasive arguments theory (Vinokur & Burnstein, 1974) and social comparison theory (Baron, Dion, Baron, & Miller, 1971). Compliance processes lead group members to voice the decision preferences they believe will gain them social approval in their group. Compliance processes differ from both argumentation and comparison in that there is only lip service to the majority view and no actual preference change.
This article begins with a discussion of the bases for distinguishing among these processes. It continues with an examination of input factors influencing these processes’ strength and direction. Next, it models the influence of these processes in terms of the mutual causal relationship between discussion and group member option preferences and any resulting group decision. It ends with limitations that motivate further work.
The Basis for the Distinction Among Social Influence Processes
Research predating the group polarization and hidden profile paradigms suggested three empirical generalizations concerning group social influence. First, group discussion usually leads to convergence of members’ preferences. Second, group decisions reflect this convergence. Third, when a group minority exists, it normally changes its preferences toward the majority’s. Most subsequent work has been consistent with all three. In fact, the group polarization effect is primarily a statistical artifact resulting from minorities conforming to more extreme majorities (Boster, Mayer, Hunter, & Hale, 1980; Davis, Kerr, Sussmann, & Rissman, 1974; Ohtsubo, Masuchi, & Nakanishi, 2002).
Asch’s (1951) well-known findings, in which a significant proportion of participants hearing confederates state decision preferences in obvious contradiction with available visual evidence voiced those heard preferences, were critical in differentiating compliance processes from actual persuasion. Less well known are the results of subsequent participant interviews. Some admitted parroting the majority view although believing it false, exhibiting compliance as defined here. Others believed their observations to be wrong given the confederate judgments, equating to true influence. Using a wide variety of tasks, Crutchfield (1955) replicated Asch’s experimental and interview findings. In Deutsch and Gerard’s (1955) extension, judgments made publicly were more affected by confederate reports than anonymous judgments, revealing what they called normative influence (compliance). In addition, anonymous judgments were often in error when compared with Asch’s highly accurate control conditions. Deutsch and Gerard attributed this latter finding to actual persuasion, which they called informational influence.
However, participants in these and comparable (M. Sherif, 1935) studies were restricted to statements of preferences and barred from discussion of supportive reasons, obviating the possibility of argumentation processes. Group polarization and hidden profile research permitted both preference and argument exchange, which allowed for distinguishing comparison from argumentation processes. Evidence supporting persuasive arguments theory (Ebbesen & Bowers, 1974; Vinokur & Burnstein, 1974) and revealing that information known by one group member alone can influence others (Winquist & Larson, 1998) is consistent with argumentation process accounts. Findings favoring the social comparison explanation for group polarization (Baron et al., 1971) and the common knowledge effect, in which jointly known information holds sway over member preferences in lieu of what is newly learned during discussion (Gigone & Hastie, 1993), are in line with comparison process explanations. Compliance processes are uncommon in these research paradigms, as private and public prediscussion preferences are usually identical.
Although influence via social comparison and argumentation have been considered competitive explanations, evidence in favor of both is sufficiently strong that a general account for group social influence must include both plus the proviso that each member originally prefers the option with the greatest amount of known supportive evidence. Boster and Mayer’s (1984) phase model, in which group members first exchange prediscussion preferences, motivating comparison-based influence, and second, the reasons for them, fueling argumentation-founded influence, is supported in published research (Dennis, 1996) and unpublished reanalysis of my own data (Pavitt, 1994; Pavitt & Aloia, 2009). Winquist and Larson’s (1998) parallel process model distinguished circumstances in which each form of influence takes precedence over the other. These models are too simple, as both processes are intertwined. Comparison processes influence argumentation, as arguments are more effective when attributed to the majority, and argumentation processes affect comparison, as the expression of argument reinforces the perception of a group member’s preference. In addition, pressure toward cognitive consistency can turn what was originally verbal compliance into true preference change (Festinger, 1957).
Factors Influencing Social Influence Processes
Situational Factors
Task type
Group decision tasks differ in the extent to which the decision can be compared with a demonstrably correct answer. Most generally, the more demonstrable the group decision, the more that argumentation processes and less that comparison processes will be engaged (Huang & Wei, 2000; Kaplan, 1989; see Mesmer-Magnus & DeChurch, 2009, for a relevant meta-analysis), as measured by the proportion of statements providing task-relevant information versus individual preferences (Kaplan & Miller, 1987; Kelly, Jackson, & Hutson-Comeaux, 1997; Pavitt & Aloia, 2009). Relatively more demonstrable decisions also result in participants placing more value on informational items in contrast with items reflecting normative judgments (Kaplan, Schaefer, & Zinkiewicz, 1994). Describing a hidden profile task as having a demonstrable answer (i.e., as a problem to be solved rather than a judgment to be made) increased the discussion of diagnostic information (Stasser & Stewart, 1992).
There is no evidence allowing for the specific contribution of task demonstrability to compliance processes at this time. Finally, task should moderate several other factors discussed below.
Time pressure
According to the attentional focus model (Karau & Kelly, 1992), the presence of optimal time for decision making encourages the full discussion of issues and the opportunity for argumentative influence. Time scarcity, in contrast, results in pressures for compliance and influence through comparison processes. Time pressure increases the proportion of task-relevant comments at the expense of relationally oriented talk (Karau & Kelly, 1992; Kelly & McGrath, 1985) and of preference-relevant utterances in lieu of the informational-influence-relevant (Kelly et al., 1997). During hidden profile studies, groups under time pressure will work faster, as measured by the number of informational items mentioned per minute, but fewer items in total will emerge, with these effects accentuated by task complexity (Kelly & Karau, 1999; Kelly & Loving, 2004; Parks & Cowlin, 1995).
Schachter’s (1951; see also Festinger & Thibaut, 1951; Sampson & Brandon, 1964) classic work detailing group response to deviance implies that increased time pressure leads to greater attempts by group majorities to gain compliance from deviants. It is believed that these appeals begin as argumentation processes but end as attempts at verbal compliance. Although there is evidence that these attempts subside with extreme time pressure as the group gives up the effort, it is sensible to hypothesize compliance processes as greatest when deadlines approach.
Karau and Kelly (1992) also noted that time scarcity will tend to accentuate already existing group structures, moderating the impact of time on communication utilization. Relevant examples will be discussed below.
Group Factors
Prediscussion preference distribution
Boster and Mayer’s (1984) phase model and reanalysis of our own data (Pavitt, 1994; Pavitt & Aloia, 2009) imply that, after preference exchange, discussion of relevant information is less likely when groups are initially unanimous, precluding argumentation processes. In contrast, disagreeing minorities motivate the search for relevant information (Nemeth & Wachtler, 1983; Schulz-Hardt, Frey, Lüthgens, & Moscovici, 2000; C. M. Smith, Tindale, & Dugeni, 1996). Nemeth (1986) argued that the assumption of the majority’s correctness underlies its influence on a minority (comparison processes). In contrast, minority influence occurs when this assumption is dropped, resulting in increased information exchange (argumentation processes). Information exchange is greatest when each group member begins with a differing preference (Brodbeck, Kerschreiter, Mojzisch, Frey, & Schulz-Hardt, 2002; SchulzHardt, Brodbeck, Mojzisch, Kerschreiter, & Frey, 2006). Groups beginning in disagreement provide a greater number of written reasons for their decision than groups starting in unanimity (Propp, 1997).
The presence of larger minorities increases their persuasiveness and unwillingness to give in to majorities (Asch, 1951), and their presence requires more discussion to reach a decision (Wood, Lundgren, Ouellette, Busceme, & Blackstone, 1994). The persuasiveness of large minorities fails when the relative faction size approaches equality (Nemeth, Wachtler, & Endicott, 1977), with the ideal minority size for successful persuasion about half the majority’s. Wood et al. (1994) also noted that minority influence was more effective in more demonstrable tasks (perceptual judgments rather than opinions); this is no surprise given its relevance to argumentative influence.
Spiral of silence theory and research (Noelle-Neumann, 1993) supports the notion that people perceiving themselves as in the minority are less likely to voice their opinions than the majority. Noelle-Neumann interpreted this tendency in terms of compliance processes, but Asch’s (1951) findings imply that comparison processes are also at work.
Group cohesiveness
Although traditionally treated as a unitary concept, research (Back, 1951) implies otherwise. Relational-based cohesiveness, founded on the basis of interpersonal attraction among group members, leads to more equal, relationally positive, and task-irrelevant talk among group members and fewer overt attempts at social influence than task-based cohesiveness resulting from the potential of the group to reach goals unattainable by individual members (Castaño, Watts, & Tekleab, 2013; Tziner, 1982). In meta-analyses (Beale, Cohen, Burke, & McLendon, 2003; Mullen & Copper, 1994), task-based cohesiveness has been consistently associated with better group performance, but findings for relational-based cohesiveness has not, particularly in real-world groups. Groups with task-oriented goals or norms exchange more preferences and information and are more disagreeable than groups with relational-oriented goals or norms (M. L. M. Henningsen, Henningsen, Cruz, & Morrill, 2003; Postmes, Spears, & Cihangir, 2001). Finally, in Schachter’s (1951) study of conformity pressure, attempts at persuading deviants were greatest in groups composed of interested participants discussing a relevant topic than the opposite. It follows that compliance processes are also more prominent in groups with high task-based cohesiveness, and that these processes are accentuated under time pressure.
Leader style
When groups have an assigned or elected leader, that leader’s style can result in an atmosphere more or less conducive to information and preference exchange. Overly directive leaders discouraging open discussion and explicitly favoring a pet proposal result in fewer proposals and less information exchange than nondirective leaders (Flowers, 1977; Larson, Foster-Fishman, & Franz, 1998; Leana, 1985). This tendency often results in social influence consistent with the directive leader’s desires, although Cruz, Henningsen, and Smith’s (1999) groups brought up information disfavored by their directive leaders. It follows that directive leadership is conducive to compliance processes and participative leadership to argumentation processes. Although more uncertain, the levels of free discussion and freedom of expression with participative leadership imply that comparison processes also flourish.
In a review, Peterson (1997) distinguished between process directiveness (the extent to which leaders encourage discussion of alternative proposals) and outcome directiveness (the extent to which leaders advocate a favored position). The studies cited above conflated these two, with directive leaders low on the former and high on the latter and nondirective leaders the opposite. Peterson presented experimental and real-world evidence that distinctions in process leadership alone were responsible for differences in observer ratings and/or member perceptions of process and decision quality, although outcome directiveness was associated with persuasiveness in an experimental simulation.
Leader style is moderated by the influence of time, as a group’s tendency to accept directive leadership is accentuated in circumstances of time scarcity due to the increase in attentional focus on task completion (Karau & Kelly, 1992).
Personal Factors
Status
Status can be viewed in terms of an assigned position giving a person legitimate power within a formal organization (Altman, Valenzi, & Hodgetts, 1985) or as a by-product associated with characteristics such as sex, race, age, and occupation (Webster & Driskell, 1978). In either case, it generally operates within discussion groups both directly, with higher conformity to those of high status, and indirectly, with high-status individuals more likely to contribute to discussion and those contributions more likely to be accepted by other group members (Bass & Wurster, 1953a, 1953b; Larson, Christensen, Abbott, & Franz, 1996; Saunders, Robey, & Vaverek, 1994; Weisband, Schneider, & Connolly, 1995).
Research relevant to expectation states theory (Berger, Fisek, Norman, & Zelditch, 1977) demonstrates that status differences (i.e., military rank) are associated with influence in dyads. Sex and race have historically worked as predicted, with men more talkative and influential than women in mixed-sex groups except when the topic is female-sex-typed (Lockheed, 1985; Propp, 1995; Smith-Lovin, Skvoretz, & Hudson, 1986), and Caucasian Americans more verbally active and less prone to social influence than African Americans in mixed-race groups (E. G. Cohen, 1972; Katz & Benjamin, 1960). Whether these tendencies have diminished over time is unknown. Differences in age (Freese & Cohen, 1973) and educational standing (e.g., 4-year college student vs. junior college student vs. high school student; Moore, 1968) have behaved likewise.
Assigned leadership alone does not confer status and influence (Carter, Haythorn, Shriver, & Lanzetta, 1951; Mortensen, 1966), particularly if its basis is task-irrelevant (Berger et al., 1977), or if group members have knowledge about the task and one another’s abilities (Webster & Driskell, 1978). Although having their greatest impact on compliance processes, high-status members are more likely to offer information of which only they are aware, inconsistently with the common knowledge effect and relevant to argumentation influence (Larson, Christensen, Franz, & Abbott, 1998; Savadori, Van Swol, & Sniezek, 2001). Finally, status should have less impact in more demonstrable decision situations, as group members will be more likely to ignore a high-status member’s lead and utilize objective criteria (Kirchler & Davis, 1986).
Demonstrated expertise
Expectation states research suggests that perceived expertise leads nonexperts to conform to the expert’s opinion (Berger et al., 1977; Hemphill, 1961; Marak, 1964). High participation in discussion can lead to credibility and influence judgments when group members evaluate that participation as valuable (Gintner & Lindskold, 1975; Riecken, 1958). Demonstrated competence can also counteract cultural status markers, with high participating females influential in otherwise male groups if their participation is judged as in the group’s best interest (Ridgeway, 1982), if they show competence (Bradley, 1980; Schneider & Cook, 1995), or if group members are aware of their relevant past experience (Wittenbaum, 1998). The influence of race (E. G. Cohen & Roper, 1972) and age (Freese & Cohen, 1973) can also be superseded by perceptions of relevant expertise.
Relevant knowledge decreases the effects of status on participation and influence (Dubrovsky, Kiesler, & Sethna, 1991; Hollingshead, 1996). Those with relevant expertise are less susceptible to the common knowledge effect than other group members (Larson, Foster-Fishman, & Franz, 1998; Wittenbaum, 1998, 2000), with their increased discussion of uniquely known information implying greater argumentative influence. Knowledge of specialized information means expertise in that domain and increases its likelihood of disclosure (Fraidin, 2004; Stasser, Stewart, & Wittenbaum, 1995). Relevant expertise has a greater effect on discussion-based influence than sheer talk time (Bottger, 1984). Relevant expertise should mean more influence in more demonstrable tasks, as experts can indicate the better option. In Franz and Larson (2002), experts identified as such mentioned more information than nonexperts, particularly when the task was described as having an objectively correct answer rather than not.
I wish to distinguish pure status effects, affecting compliance processes as members wish to display their leader’s preference, from pure expertise effects, influencing argumentation processes given the data the expert presents to group members and comparison processes as group members may trust the expert’s preference. Status effects are accentuated by the absence of, and expertise effects by the presence of, task demonstrability. Nonetheless, status and expertise are real-world conflated, as high-status people talk more and so often present more relevant information to the group, and expertise provides status.
Talkativeness
Although intertwined with status and expertise, overall talkativeness is a consequential individual difference on its own (Bonito & Hollingshead, 1997). Callaway, Marriott, and Esser (1985) determined that groups with members above the sample median in dominance made more agreement and disagreement statements, reported more attempted and successful influence attempts, and made more accurate judgments than groups below the sample median. The first option advocated in discussion usually ends up the group’s choice (McGuire, Kiesler, & Siegel, 1987), and individual member preferences shift toward the first advocate when that advocate is self-selected rather than preassigned (Weisband, 1992). The latter finding is probably a function of talkativeness, as the most talkative member is likely to be the first to voice an option preference. There is some evidence that group members who enter discussion with the same task-relevant knowledge as other members are more talkative (Sargis & Larson, 2002), judged as more influential, and are less susceptible to social influence than those who shared the least (Kameda, Ohtsubo, & Takezawa, 1997). This tendency may not hold when group members are aware that those not sharing knowledge items with the other members are equally competent (Larson, Sargis, Elstein, & Schwartz, 2002). Differences in group size will moderate the effect of talkativeness on group process. Researchers (reviewed in Bonito & Hollingshead, 1997) are in agreement that a group’s most talkative member will hog about 40% of the total talk time in groups from 3 to 10 members, with the others accorded ever smaller proportions of the floor as group size increases. The implication is that particularly talkative members are relatively more influential in larger groups than in smaller. This relationship will be further moderated by time scarcity; based on Isenberg (1981), Karau and Kelly (1992) hypothesized that the resulting pressure for task completion might increase the tendency for the talkative to dominate the floor.
It follows that the presence of talkative members would influence argumentation processes. Although past research offers insufficient rationale, it is likely that pure talkativeness also affects comparison processes. Talkativeness should not be associated with compliance processes independently of member status.
Needs for cognition and approval
A. R. Cohen (1957) defined need for cognition (NFC) as the desire to understand ambiguous situations, while noting high NFCs to spontaneously consider persuasive arguments whereas low NFCs required outside prompting. Petty and Cacioppo (1986) reinterpreted NFC as the tendency to engage in effortful cognitive activity. Cacioppo, Petty, Feinstein, and Jarvis (1996) concluded that NFC is associated with recall of recently observed information and consistency of attitudes with that information. High NFCs making individual decisions are less susceptible to framing biases, implying a more rational approach to their problem, and undertake more thorough information searches than lows, with this tendency moderated by situational cues (Bailey, 1997; Levin, Huneke, & Jasper, 2000; S. M. Smith & Levin, 1996). In the group context, Petty and Cacioppo summarized an unpublished conference paper by themselves and Kasmer in which high NFCs were less prone to social loafing while brainstorming (replicated by B. N. Smith, Kerr, Markus, & Stasson, 2001, in a perceptual test). High NFCs engaged in two-person mock juries were more persuasive and perceived by their partners as providing more and higher quality arguments favoring their viewpoint than lows (Shestowsky, Wegener, & Fabrigar, 1998), and as offering more information during group discussion common among group members, although surprisingly not more information unique to them (D. D. Henningsen & Henningsen, 2004).
Need for approval (NFA), the characteristic tendency to seek positive regard from other people, probably influences compliance tendencies. Crowne and Marlowe (1964) observed those high in NFA as more likely to state agreement with the voiced group majority in Asch-type studies, particularly if a confederate reporting the correct perception is verbally criticized by the other confederates and then complies with them. Although likely a case of compliance rather than comparison processes, there were no postexperiment interviews to gauge participant motivations. Other compliance-relevant evidence includes Baseheart’s (1971) observations that high NFAs were only persuaded more than low NFAs when messages implied others’ agreement or disagreement, and Martin and Greenstein’s (1983) discovery that high but not low NFAs were influenced by others’ status. D. D. Henningsen and Henningsen (2004) noted NFA as positively associated with the repetition of shared information and negatively associated with the repetition of unshared information, what would be expected from people wishing to seek others’ approval.
In summary, high NFCs are more likely to engage in argumentation processes whereas low NFCs, in the words of Cacioppo et al. (1996), are prone to “rely on others (e.g. celebrities and experts) . . . or social comparison processes” (p. 198), along with compliance. Those high in NFA would be more likely than those low to comply and shy away from argumentation processes. I would hesitate to offer a comparison process hypothesis for NFA at this time.
Table 1 summarizes the circumstances under which the aforementioned input factors have their greatest effect on argumentation, comparison, and compliance processes during group discussion. As indicated, there are instances in which factors are proposed to have no or, at this time, unknown influence on given processes.
Conditions in Which Input Factors Have the Greatest Impact on Argumentation, Comparison, and Compliance Processes.
Note. ? refers to High as being a reasonable hypothesis with no present support.
The Simplified Model of Group Social Influence Processes
The Simplified Model of Social Influence Processes deserves that title due to a set of seven simplifying assumptions under which it is operative. Four of these assumptions were proposed at the beginning of this article: (a) That group members enter discussion aware of all proposals, (b) that group members enter discussion having formulated a prediscussion preference for one of the proposals, (c) that group members leave discussion with a possibly changed postdiscussion preference, and (d) that the group combines its members’ postdiscussion preferences into a decision. Fifth, group discussion consists solely of items of information and statements of individual preferences, ignoring procedural, relational, and irrelevant utterances. Sixth, the factors discussed in the literature review govern the tendency for statements to contribute information or signal preference. Seventh, most of these factors remain consistent through discussion, with exceptions including changed distributions of individual member preferences, the intervention of time pressure, and the imposition of a high-status member’s preference.
Discussion begins with the exchange of prediscussion preferences (Boster et al., 1980; Dennis, 1996). If unanimity is discovered, it is decisive, and, although discussion of options may continue, for our purposes, the decision process ends at that point. Absence of unanimity motivates the discussion of information items, activating argumentation processes. Any obvious majority triggers comparison and compliance processes, with its relative strength dependent on the size of the majority. If a high-status member declares an option preference, compliance processes are activated.
As individual group members generally contribute items of information consistent with their prediscussion preference (Pavitt, 1994), the proportion of members supporting various options can be gauged by the option for which each member argues (Lemus, Seibold, Flanagan, & Metzger, 2004; McPhee, Poole, & Seibold, 1981; Meyers & Brashers, 1998). Mention of information items, either known before discussion or newly thought of, adds both knowledge supporting the relevant option and the perception that the communicator favors that option. If in support, the majority viewpoint will be buttressed by argumentation and comparison processes in tandem. Minority members may hesitate to talk, resulting in a self-fulfilling prophecy reinforcing the majority viewpoint.
Argumentation processes can affect member support for various options. If new items supporting the minority appear, argumentation processes favor them and comparison processes weaken. If not superseded by comparison and compliance processes, sufficient information favoring an originally dispreferred option can change prediscussion preferences. The majority switching favor from one option to another will change what is benefited by comparison processes. Analogously, if a majority forms in a previously evenly split group, comparison processes come into play. Explicit statements of preference can occur at any later time, either strengthening or weakening comparison processes depending on whether for the majority or minority option. As argument credibility is affected by the side it supports, argumentation processes can be affected in turn.
With neither perceived majority nor pressure to comply with a high-status member, argumentation processes prevail. High-status member interjections short-circuit argumentation processes and, if group members voice support for the leader’s preferred proposal, the (perhaps illusionary) appearance of a majority can strengthen comparison processes. Finally, time pressure increases the importance of compliance and comparison processes relative to argumentation.
The move to final decision is reflected in discussion content. Decision options lacking significant group favor drop by the wayside, leaving those options with the most support (Hoffman, 1979; Scheidel & Crowell, 1964). The group usually chooses the option with the strongest voiced support across the entire group (Brodbeck et al., 2002; Hoffman, 1979; Hollingshead, 1996; Kelly & Karau, 1999; Kim, 1997; Meyers, Brashers, & Hanner, 2000; Winquist & Larson, 1998), particularly when favored by the majority of individual comments (Lemus et al., 2004; McPhee et al., 1981; Meyers & Brashers, 1998). Although the option with majority approval at the outset of discussion is normally chosen, the following circumstances are conducive to minority influence: a task with a highly demonstrable solution argued for by a minority about half the size of the majority and including members who are particularly talkative, high status, or expert in the relevant topic. Minority influence is also encouraged by consistency in the arguments they offer (Gebhardt & Meyers, 1995; Meyers et al., 2000).
Figures 1 and 2 are intended to diagram aspects of the Simplified Model as implied by differing perspectives. Figure 1 takes a macro-level view of the process by which the overall implications of the model would be played out in group discussion across the entire length of decision making. Beginning with original preference exchange, this figure highlights the interplay of information and preference exchange and distribution of preferences among members as they impact one another during the rest of the discussion, culminating in the final group decision. Figure 2 zeroes in on the specific factors impacting the model-relevant thoughts and comments of an individual group member from speaking turn to speaking turn. In this figure, output factors affect a group member’s interpretation of the previous group discussion contribution whereas input factors influence a member’s decisions concerning when and what to contribute. Solid arrows displayed under the Process labels represent actual information and preference statements; intermittent lines under the output and input labels indicate the analogous psychological portions of the respective argumentation and comparison/compliance processes. Crisscross arrows between the processes signify the psychological influence each process has on the other. Variables listed at the top affect processing of information items, those at the bottom preference items, and those in the middle both. The diagrammed process repeats itself for every discussant during each discussion contribution by any member. To maintain simplicity in both figures, I have not included the direction (i.e., positive vs. negative) of the relationships, which can be found in Table 1 and the following list of theoretical propositions.
Interactions among discussion exchange and preference distribution for group across discussion.
Interactions among discussion exchange and input variables for individual members across speaking turns.
Propositions
The Simplified Model of Social Influence Processes implies a set of theoretical propositions relevant to the preponderance of information statements indicative of argumentation processes and preference statements representing comparison and/or compliance processes during discussion. I distinguish between propositions in which the causal factor is an input variable versus a process variable. Within these two categories, I further divide the set into two types: static propositions, which include factors that are present at the beginning of group discussion and remain fairly constant throughout that discussion, and dynamic propositions, which concern factors that undergo change as discussion continues. I have sequenced the propositions consistently with the order in which the relevant factors are described in this article. The term pure is used when referring to the impacts of talkativeness, status, and expertise on preference and information statements independently of one another, ignoring the fact that the three are inseparable in practice.
Input Variable Propositions
Process Variable Propositions
Limitations and Suggestions for Future Research
There are a number of factors not included herein that undoubtedly affect group social influence processes. Some of these factors can be called errors of commission, or unrealistic simplifications made in the presentation. One such error is treating all input variables as equally important when they most certainly are not. Although some are included, a second is underestimating the presence of interactions among input variables. A third is assuming that all possible options are known to all members before discussion. A fourth is supposing that members have had the chance to form prediscussion preferences. Regarding the latter, evidence is mixed concerning how argumentation and comparison processes differ between groups with and without this opportunity. Mojzisch and Schulz-Hardt (2010) and Reimer, Reimer, and Hinsz (2007) noted more information exchange and Reimer et al. noted less preference exchange in the former condition, but Pavitt and Aloia (2009) discovered the opposite, with the inconsistency in results likely due to task differences. Fifth, the presumption in this presentation that preference change is a result of group discussion rather than individual thought processes is unproven (Hewes, 1986). Sixth, most of the reported findings are from experimental groups performing free discussion and may not generalize to organizational groups using some form of formal discussion agenda as a guide.
Other factors ignored here could be labeled as errors of omission, or additional input variables deserving attention. First, culture is likely a relevant input variable (Bonito & Hollingshead, 1997), with collectivists probably more influenced by comparison and compliance factors and less by argumentation items than individualists. Second, the strength of arguments on top of sheer amount may determine the influence of majorities on minorities (Limon & Boster, 2001). Third, individual differences in open-mindedness as a consequence of topic involvement might impact individual persuasion (C. Sherif, Sherif, & Nebergall, 1965). A fourth error of omission is the specific role played by various forms of argumentation in influence processes. There is significant research linking argument-relevant discussion content to social influence (e.g., Gebhardt & Meyers, 1995; Lemus et al., 2004; Meyers et al., 2000), but it is impossible from this work to discern the extent to which argumentation versus comparison processes are involved, as Lemus et al. (2004) noted when discussing their findings.
Future research is required to correct many of these commission/omission errors. In addition, we really do not know much about either the process by which communication patterns and individual member preferences change over the course of discussion or how these changes interact with one another. The examination of these dynamic patterns would require methods differing from the present-day norm, in which preferences are measured before and after discussion and related to communication content for the discussion as a whole. For one example, manipulations can occur part way through a discussion (e.g., a high-status member suddenly voices support for one option; the group unexpectedly learns that decision time is scarce), with comparisons between information and preference statements made before and after the manipulation. Measurement of member preference among options, conducted both publicly and privately, can occur at specified times during the discussion, such as before and after manipulations of the type just proposed, with the proviso that the necessary interruptions in discussion flow might disrupt the processes under examination. Two other possibilities are described in Waldron and Cegala (1992). In one, group members would interact with one another electronically while verbalizing their concurrent thoughts about the discussion; in the other, group members individually observe a recording of a previous discussion in which they were involved and report their memories of their concurrent thoughts. In both cases, the reported thoughts of each member would be temporally matched to group discussion content in transcripts of the discussions. All of these possibilities would benefit from collaborations among researchers from multiple disciplines, in particular pairing those with expertise in experimental research with those skilled in content analysis.
Even so, given that comparison and argumentation processes reinforce one another and that verbal compliance can turn into true persuasion, teasing apart the individual impacts of each component may be beyond present-day methodology. In this case, computer simulation provides an alternative route for exploring the implications of this model. A good role model is Stasser’s (1988) DISCUSS simulator, which is founded on many of the same presumptions concerning prediscussion information and the dissemination of that information during discussion as the present model.
Despite its errors of commission and omission, I hope that the Simplified Model of Group Social Influence reflects a more realistic account of group discussion processes than past efforts. Perhaps future research using alternatives to the group polarization and hidden profile paradigms, adopting either the methodological proposals made here or others yet to be imagined, may someday allow for the relaxation of some or all of the simplifying assumptions behind the model and inspire more realistic replacements for it.
Footnotes
This article is part of the special issue: 2014 Annual Review Issue, Small Group Research, Volume 45(6).
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.