Upload
joseph-a-bonito
View
216
Download
0
Embed Size (px)
Citation preview
Un modèle local du partage de l’information dans des petits groupes
Joseph A. Bonito
University of Arizona
Résumé
Cet article esquisse un modèle à deux étapes du traitement de l’information dans des petits
groupes, au sein duquel le processus de communication influence la quantité et la qualité de
l’information mentionnée durant la discussion. La première étape du modèle, l’activation, définit
comment l’information devient disponible dans la mémoire à court terme, de façon à offrir une
assise à une éventuelle contribution à la discussion. Il s’agit d’un processus fondé sur la
cohérence; l’activation est fonction de contributions, précédentes et projetées, à la discussion. La
seconde étape, le choix, définit les conditions régissant la contribution ou la dissimulation
d’information active par les membres. Bien que le choix de participer puisse reposer sur plusieurs
conditions préalables, le modèle se concentre sur les effets de la contribution à la discussion
d’information partagée et unique.
A Local Model of Information Sharing in Small Groups
Joseph A. Bonito
University of Arizona
Nachfolgender Artikel skizziert ein Zweistufenmodell der Informationsverarbeitung in
Kleingruppen, bei dem der Kommunikationsprozess, die in einer Diskussion erwähnten
Informationen in ihrer Quantität und Qualität beeinflusst. Der erste Schritt im Modell, nämlich
Aktivierung, beschreibt, wie Informationen im Kurzzeitgedächtnis verfügbar gemacht werden
und so die Grundlage für einen potentiellen Beitrag zur Diskussion liefern. Dieser Prozess basiert
auf Kohärenz; Aktivierung selbst ist eine Funktion von Vorbedingungen und geplanten Beiträgen
zur Diskussion. Der zweite Schritt, die Wahl, beschreibt die Bedingungen unter welchen die
Teilnehmer aktive Informationen beitragen oder zurückhalten. Auch wenn die Entscheidung, sich
in eine Diskussion einzubringen, auf eine Vielzahl von Vorbedingungen zurückzuführen ist,
fokussiert das Modell auf den Beitrag geteilter und einzigartiger Information zu einer Diskussion
und damit verbundene Effekte.
Un Modelo Local de Información Compartida en Pequeños Grupos
Joseph A. Bonito
University of Arizona
Resumen
Este manuscrito esboza un modelo de procesamiento de información en pequeños grupos
de dos pasos, en el cual el proceso de comunicación influye sobre la cantidad y la calidad
de información mencionada durante la discusión. El primer paso en este modelo, la
activación, describe cómo la información se vuelve disponible en la memoria de corto
plazo de manera tal que provee de base para las contribuciones potenciales en la
discusión. Este es un proceso basado en la coherencia; la activación es una función de las
contribuciones precedentes y proyectadas en la discusión. El segundo paso, la opción,
describe las condiciones bajo las cuales los miembros contribuyen ó retienen información
activa. Aún cuando la opción de participar puede ser basada en una congregación de
condiciones predecesoras, el modelo se enfoca en los efectos de la información
contribuyente compartida y única en la discusión.
小组信息分享的局部模式
Joseph A. Bonito
亚利桑那大学
摘要
本文勾画出小组信息处理的二步模式以解释传播过程对讨论中所提及信息的数量和
质量的影响,模式的第一步为激活:描述信息在短期记忆的出现,为参与讨论打下
基础。这是一个相互关联的过程。激活是由讨论先例和计划参与讨论两者所决定
的。模式的第二步为选择:描述了小组成员贡献或保留信息的条件,尽管选择参与
是由一系列先决条件所决定的。本模式着重解释参与讨论和独特信息的效应。
소집단내에서의 정보공유 지역 모델에 관한 연구
Joseph A. Bonito
University of Arizona
요약
본 논문은 커뮤니케이션 과정이 대화도중에 언급된 정보의 양과 질에 영향을 미
치는 가에 관한 소규모 집단내 정보처리의 2 단계모델에 대한 연구이다. 모델의
첫번째 단계는 활성화로 이는 어떻게 정보가 단기간 기억내에서 가능하게 되
는지를 기술하는 것으로, 그렇게 함으로써 대화에 있어 잠재적 공헌을 위한
기반을 제공하게 된다. 이 단계는 응집성에 근거하는 과정인바, 활성화는 대
화에 있어 선행하는 그리고 고안적인 공헌들로서 기능하게 된다. 두번째 단계
인 선택은 구성원들이 활성화된 정보에 기여를 하는가, 아니면 이를 제지하는
가에 대한 상황들을 기술하고 있다. 비록 참여에 있어서의 선택이 다수의 선
행적 조건들에 기초할 수 있으나, 이 모델은 대화에서의 공유된 기여의 효과
와 독특한 정보에 중점을 두고있다.
ORIGINAL ARTICLE
A Local Model of Information Sharing inSmall Groups
Joseph A. Bonito
Department of Communication, University of Arizona, Tucson, AZ 85721
This paper sketches a 2-step model of information processing in small groups in which
the process of communication influences the quantity and quality of information men-
tioned during discussion. The first step in the model, activation, describes how informa-
tion becomes available in short-term memory, such that it provides the basis for
a potential contribution to discussion. It is a process based on coherence; activation is
a function of antecedent and projected contributions to discussion. The second step,
choice, describes the conditions under which members contribute or withhold active
information. Although choice to participate can be based on a host of antecedent condi-
tions, the model focuses on the effects of contributing shared and unique information to
discussion.
doi:10.1111/j.1468-2885.2007.00295.x
Small-group researchers have invested considerable effort into understanding theconditions under which participants mention information, as well as the effects of
sharing information on group outcomes. Because information-sharing research hasits roots in persuasive arguments theory (PAT; see Stasser & Titus, 2003), it is subject
to similar criticisms. For current purposes, the most relevant criticism of PAT wasoffered by Meyers and Seibold (1990), who noted that PAT assumes a ‘‘cognitive-
informational’’ approach, in which the individual is the focus, and argument—including the conditions of its production as well as its effects—is treated as fixed
and stable. They did not argue against such an approach but suggested instead that itshould be complemented by a ‘‘social–interactional’’ perspective, in which argumentis considered a ‘‘jointly produced, socially governed, interactive activity’’ (Meyers &
Seibold, 1990, p. 269), and the focus is on the social unit or group rather than on theindividual. This perspective, they argued, would account for Meyers’ (1989) obser-
vation that cognitive arguments (i.e., the information that people know or are askedto memorize prior to discussion) look very little like the arguments people make
during discussion. The dynamic and emergent character of interaction influencesboth the way in which participants argue as well as the effects of their arguments.
Corresponding author: Joseph A. Bonito; e-mail: [email protected]
Communication Theory ISSN 1050-3293
252 Communication Theory 17 (2007) 252–280 ª 2007 International Communication Association
Models of information sharing, much like their predecessor PAT, assume a cog-nitive-informational approach by focusing on social or social–psychological mech-
anisms that might compel members to mention, or dissuade them from mentioning,information during discussion (for reviews see Stasser & Titus, 2003; Stasser &
Vaughan, 1996; Wittenbaum, Hollingshead, & Botero, 2004). Information is treatedas static, and the conditions under which members mention it during interaction areassumed to be fixed and stable. Although this approach is perfectly reasonable, it
ignores or takes for granted the role of interaction in the mentioning of information.When viewed through a social–interactional lens, information is dynamic and fluid,
and participants are assumed to tailor the mentioning of information to the unfold-ing characteristics of interaction. Thus, interaction is both an influence on and an
outcome of information sharing; the information in one’s contributions is shapedby and shapes the information-based contributions of his or her colleagues.
In what follows, I develop a local model of information sharing, one that has itsroots in a social–interactional perspective. Pavitt and Johnson (1999) used the termslocal and global to describe orientations to coherence during group discussion. A local
orientation refers to the extent to which one’s contributions are substantively relevantto preceding and subsequent contributions (e.g., O’Keefe & Lambert, 1995; Sanders,
1987; Tracy, 1982), whereas a global framework indexes the relation between largersegments of discussion and a given contribution. My use of the term local is consistent
with this definition, although I expand the term global to include the correspondencebetween a contribution to discussion and factors exogenous to it. For example, an
expert on a given topic or problem will often contribute frequently because othermembers defer to his or her expertise (e.g., Franz & Larson, 2002). One’s perceived
expertise and its effects likely remain constant throughout discussion.In adopting a local perspective, I do not reject or ignore cognitive-informational
approaches to the problem of information sharing. Rather, I propose a two-step
model that incorporates elements of both local and global factors in explaininginformation sharing during discussion. As I describe in detail below, the model
assumes that information must first be activated from long-term memory (Stage1) before a member can decide whether to contribute it to discussion (Stage 2).
Activation occurs primarily because of the potential relevance of a given piece ofinformation for the issue or topic under consideration at a particular moment. Thus,
discussion is an influence on the quality and quantity of information that one mightinclude in a contribution. Other factors come into play when a participant decides tocontribute (or not) such information. For example, one might have relevant infor-
mation to provide but will defer from providing it because of the perception thatother members, for whatever reasons, are better able to contribute to discussion.
Information-sharing research and hidden profiles
I begin with a brief description of hidden profiles, focusing primarily on Stasser and
Titus’s (1985) seminal study. I then summarize the research findings, drawing on
J. A. Bonito Local Model of Information Sharing
Communication Theory 17 (2007) 252–280 ª 2007 International Communication Association 253
reviews provided by Stasser and Titus (2003), Stasser and Vaughan (1996), and Wit-tenbaum et al. (2004). My point in doing so is to (a) offer a set of critiques for existing
explanations of information processing and exchange and (b) lay the groundwork forhow such factors figure into a local management model as described below.
The seminal study
The scholarly literature addressed here has its roots in Stasser and Titus’s (1985)
study in which the authors sought to disentangle the effects of normative andinformational influence on decision making. Normative influence refers to social
comparison processes, for example, when one adopts a position because the majorityof members are known to prefer it (see Pavitt, 1993). Informational influence occurs
when the content of members’ substantive (e.g., data-based) contributions persuadesthe group to choose a particular solution. The rationale for Stasser and Titus’s study
rested on PAT (see Burnstein & Vinokur, 1977; Vinokur, Trope, & Burnstein, 1975),which assumes that distributions of arguments among members, as well as the argu-ments’ persuasiveness, push members toward some decisions and away from others.
Thus, if prior to discussion each group member knew every argument for and againsta set of positions, mentioning the argument during discussion would have no effect
on prediscussion preferences. If, however, some members know or possess somearguments that other members do not, and if those arguments were nontrivial, then
hearing such arguments for the first time during discussion would cause members toreconsider their positions. The issue that Stasser and Titus addressed was the fact that
attitude shift could be attributed to the expression of or the learning about thedistribution of preferences in the group or to the persuasiveness of the arguments
for or against a particular solution, or to both.Stasser and Titus (1985) designed a study that would control for both informa-
tional and normative influence. The design, known as the hidden profile, is created
when (a) a set of information is developed such that, when considered as a whole,points to an optimal solution, (b) the information is distributed across members
such that some information is shared (i.e., given to all members) and the remainderis unshared or unique (i.e., given to just one member), and (c) the information each
member receives (which consists of both shared and unique units) points to a sub-optimal solution. Thus, the design produces an initial distribution of preferences
that would lead members to make a poor choice if there were no subsequent dis-cussion. Should members exchange information, however, the unique information(which by design points to the optimal decision or away from the suboptimal ones)
would influence deliberations by permitting the group to develop complete assess-ments of each of the decision alternatives, in principle increasing the likelihood that
the group will select the best choice.The groups in Stasser and Titus’s (1985) study performed poorly. Especially
noteworthy are the findings from the conflict condition (in which the hidden profilecreated two subgroups, each favoring a different suboptimal decision), where groups
not only failed to discover the hidden profile but also chose the optimal solution less
Local Model of Information Sharing J. A. Bonito
254 Communication Theory 17 (2007) 252–280 ª 2007 International Communication Association
frequently than the distribution of individual preferences at the start of discussionwould suggest—the best solution actually lost support. Subsequent research (Stasser,
Taylor, & Hanna, 1989) investigated the content of discussion, reporting that sharedinformation was mentioned and repeated more frequently than was unique infor-
mation (see also Larson, 1997). Even though groups do not always perform wellwhen all information is mentioned, it is nonetheless alarming to learn that membersoften fail to mention information when given the opportunity to do so.
Three explanations for the shared information bias
Researchers have offered three general explanations for the failure of members tomention unique information (Stasser & Titus, 2003; Wittenbaum et al., 2004). The
first is the collective information sampling model (CIS), in which it is assumed thatshared information is more likely to be discussed by the group because of its distri-
butional advantage over unique information. Imagine that all four members ofa four-person group know (or have been given to memorize) information Item A,and that information Item B has been given to just one member. Assuming random
sampling, Item A is more likely to appear in discussion because there are moreopportunities for it to be ‘‘sampled’’ and selected.1 As Wittenbaum et al. noted,
the CIS is meant to show that probability can account, to some extent, for thetendency of groups to discuss shared information.
The second explanation for the shared information bias is preference consistency.Recall that the hidden profile, by design, creates an initial preference that favors
a suboptimal decision, and that shared information, again by design, is instrumentalin developing the initial preference. It is reasonable to assume then that members
more positively evaluate knowledge that is consistent with their initial preferences,and that members will present such knowledge during discussion as evidence orsupport for their claims. Because unique information is designed to contradict initial
preferences, it is less likely to be produced; members tend to make arguments infavor of their positions, not against them. Shared knowledge, because it is largely
consistent with initial preferences, has a built-in advantage over unique informationin terms of being mentioned.
Finally, the social comparison explanation posits that the discovery one hasinformation in common with others confers a sense of legitimacy and credibility on
that knowledge, as well on the persons who mention it (e.g., Wittenbaum, Hubbell,& Zuckerman, 1999). Those who present unique information, however, because suchinformation cannot be verified,2 incur some type of social cost (Stasser & Titus,
2003); the bearer of unique information is perceived, for example, as less credible,which in turn results in a reticence to speak again, or a general disinclination for
other members to provide that member with more speaking turns, or both.
Limitations of the three explanations
The basis for this section can be found in Burke (1974), who opined that acquiring
a speaking turn and filling it are not governed by the same processes. The former is
J. A. Bonito Local Model of Information Sharing
Communication Theory 17 (2007) 252–280 ª 2007 International Communication Association 255
a function of social mechanisms related to turn-taking, whereas the latter is a productof cognitive and communicative issues that underlie message production. Turn-
taking is a collaborative endeavor in which participants, by virtue of a series ofnonverbal (e.g., Kalma, 1992) and verbal (Sacks, Schegloff, & Jefferson, 1974) means,
distribute speaking opportunities. It is often the case that the process is affected byfactors exogenous to discussion (e.g., gender or institutional rank—see Ridgeway &Smith-Lovin, 1999) such that some individuals, because they possess a set of desir-
able social characteristics, are deemed as more likely to contribute in ways thatbenefit the group. Filling a turn, however, is dependent on the activation of infor-
mation on which to base a substantive (i.e., task-relevant) contribution (Weiner &Goodenough, 1977). Two implications follow from this position. First, one cannot
substantively fill a speaking turn if he or she does not have task-based information onwhich to base a contribution to discussion. Second, one may have substantive infor-
mation to contribute but is unable or unwilling to take the floor for various reasonsrelated to the turn-taking process.
When viewed through this lens, the three explanations for the shared informa-
tion bias confound these two distinct but related processes. The CIS, for example,assumes that recalled information is automatically mentioned (see Stasser & Titus,
2003, p. 307). Clearly, recall is based on cognitive processes, whereas its mentioningis due, in part, to the processes of turn-taking. Thus, there is no guarantee that
a recently recalled piece of information will be contributed to discussion. As a casein point, Nijstad, Stroebe, and Lodewijkx (2003) noted that production blocking,
which is defined as having to take turns to express ideas (generally in brainstorminggroups but applicable to other group types), results in fewer ideas being mentioned
because active information begins to decay the longer one has to wait to gain thefloor. In addition, in the case of brainstorming, one is rehearsing ideas while waitingfor a speaking turn, which limits the resources available to recall new ideas. More-
over, the CIS relies on a probabilistic model of recall, whereas models of messageproduction assume that substantive associations among units are responsible, in
part, for activation. Although probability can describe, to a certain extent, distribu-tions of shared and unique information mentioned during discussion, the CIS makes
‘‘strong theoretical assumptions that bear little resemblance to many naturalisticgroup decision-making situations’’ (Wittenbaum et al., 2004, p. 298).
The preference consistency explanation assumes that one’s solution preferenceestablishes a link among the data that support one’s position.3 In addition, support-ing information is salient, which means it is available (in short-term memory) for
inclusion in a contribution. The preference consistency explanation, however, fails toavoid Burke’s conflation problem because it does not explain how one comes to
acquire speaking turns in support of his or her position—simply having a preferencedoes not guarantee that one will be given (or take) the opportunity to express
and provide support for it. It also does not explain any other substantive turns(i.e., ones that have little to do with one’s preference) one might take during discussion
(Wittenbaum et al., 2004).
Local Model of Information Sharing J. A. Bonito
256 Communication Theory 17 (2007) 252–280 ª 2007 International Communication Association
Finally, although the social comparison explanation provides insight into thedevelopment of relevant social perceptions that are consequential for turn-taking, it
does not explain the cognitive mechanisms that fill speaking turns. Thus, one mightbe deemed as highly credible based on his or her contributing shared information,
but credible individuals cannot participate substantively if they do not have activecontent on which to base a contribution.
The preceding leads to the following two observations. First, although recall is
clearly necessary for making substantive contributions, there is currently no theorythat explains which information is recalled and when during discussion. Only subsets
of information are active at any given moment during interaction (O’Keefe & Lambert,1995), and active information eventually decays, especially when other information
is activated later in the discussion (Chafe, 1994; Greene, 1997; Nijstad et al., 2003).4
The second observation is that the underlying processes for the failure to contribute
to discussion are treated the same for those who recall but choose not to mentioninformation and for those who do not recall information at any given moment. Thus,it is not clear if, for example, one chooses not to contribute because of social com-
parison processes or because he or she does not have relevant active information tocontribute during discussion. This is obviously an unsatisfying situation, one that calls
for a more complex picture of the relation between social and communicative pro-cesses underlying the contribution of information to discussion. Below, I present
a model that treats social processes and message-production mechanisms separatelyand then show how each contributes to information sharing in groups.
A local model of information sharing in small groups
In response to Burke’s (1974) argument, the model of information sharing proposedhere distinguishes between two distinct processes. The first is the communicative–
cognitive processes associated with developing potential contributions to discussion.This part of the model derives from O’Keefe and Lambert’s (1995) discussion of
message production during interaction, in which a contribution to discussion is pos-ited to be a function of conditions of relevance associated with interaction. These
conditions, referred to as focus, are in part responsible for the activation of thethoughts on which contributions to interaction are based. O’Keefe and Lambert con-
tended that the most critical aspect of focus is how ‘‘antecedent and projected con-tributions shape what is relevant to say at any particular juncture’’ (p. 75). In general,contributions are fashioned in some way to be responsive to what is said, as well as to
anticipate what others might say in reply (Sanders, 1987). For example, if participantsare discussing politics, then characteristics of politicians and their policies are likely to
be activated and available for expression. It is then up to the participant to determinehow such information should be used at that moment, for example, as a rebuttal or as
a way to anticipate and diffuse an interlocutor’s argument.The second part of the model describes the conditions under which members
contribute recently activated information to discussion. There are many reasons why
J. A. Bonito Local Model of Information Sharing
Communication Theory 17 (2007) 252–280 ª 2007 International Communication Association 257
participants choose to contribute, including a host of individual differences andsocial issues that develop prior to or during group interaction. The model advanced
here, however, builds on previous work on information sharing by positing thatchoice is related to the mentioning of shared or unique information. For example, in
some cases, contributing shared information positively affects perceptions related toturn-taking (e.g., credibility) that, in turn, leads to subsequent opportunities for thecontributor to participate (Wittenbaum et al., 1999). In other cases, unique infor-
mation has a similar effect by creating the impression that the speaker is knowledge-able and persuasive (Larson, Sargis, Elstein, & Schwartz, 2002).
Note that the model applies to substantive contributions, ones based on task-specific knowledge. Nonsubstantive contributions are domainless (Weiner & Good-
enough, 1977); their production is not based on possessing some knowledge aboutthe task. They are typically comments that manage the floor, reference discussion
procedures and protocols, or address interpersonal issues within the group, and canbe produced by anyone regardless of the quality and quantity of one’s task-specificinformation resources.
In what follows, I describe the components of the model. Figure 1 represents themodel in diagrammatic form. The elements in boxes refer to states, whereas the lines
are processes. The model also requires some different methodological and analyticalstrategies, some examples of which are discussed below.
Model components
Knowledge
The term knowledge, as used within the local model of information sharing, refers to
data stored in long-term memory.5 Previous work has used the term information torefer to both a datum committed to long-term memory and the content of a contri-bution to discussion (see Propp, 1999, footnote 1). This is problematic for an
activation-based model because it does not adequately distinguish the processesby which data are retrieved from long-term memory and the manner in which they
are included as content in a message. Briefly, the storage and retrieval of items in
Knowledge
Activation
Choice
Activation
Contribution Potential Contribution(s)
Figure 1 Local model of information sharing in groups.
Local Model of Information Sharing J. A. Bonito
258 Communication Theory 17 (2007) 252–280 ª 2007 International Communication Association
long-memory is based on associations or linkages between the items and stimuli thatbegin or continue the activation process (see Greene, 1997, for an example). The
actual content of a contribution to discussion is based on the selection of somecontent over others (assuming that more than one item has been activated) that is
available in short-term memory, as well as the function (e.g., as a rebuttal) to whichsuch information is put. Information, then, refers to the content of a contribution todiscussion (see below).
Connections among knowledge units may be distinguished along two relateddimensions. The first dimension, intermember, concerns the connections of units
across participants. A given unit is shared if Participant A knows in advance of orcomes to learn via discussion that she and Participant B possess it. The item is unique
if only one person possesses it (although, obviously, once an item is mentioned, it isno longer unique). Interunit relationships refer to the connections among knowledge
units irrespective of the group members who might (or might not) possess them.6
Consider how the two units, ‘‘gives multiple-choice exams’’ and ‘‘gives only essayexams,’’ from the ‘‘choose the best candidate’’ task (Larson, Foster-Fishman, & Keys,
1994) might be related. From an intermember perspective, the item concerning essayexams is connected (or shared) if all members possess that information, whereas the
one involving multiple-choice exams is not connected (i.e., is unique) if only onemember has that information. In terms of interunit connections, it is clear that both
are topically related (e.g., teaching practices) and likely have other relations, as shallbe noted below.
Potential contribution
A potential contribution refers to task-relevant information contained in short-termmemory. In addition, it refers to perceived potential uses and outcomes of thatinformation. Continuing with the example above, assume that Participant A has
expressed a preference for the candidate named ‘‘Smith’’ and that Participant B isconsidering whether to contribute the active unit that Smith gives only essay exams.
Clearly, such a contribution would have several potential consequences, includingcasting doubt on the viability of Smith as a candidate and the establishment of
a disagreement between the two participants. Participant B may choose not tocontribute this information at this time (see below for more on ‘‘choice’’), but it
is important to note that B has the resources to do so at this point in the discussion.Given the conflation problems facing research on information sharing noted above,this is an important issue in that one could say that B chose not to contribute, which
is far different from saying that B did not contribute because he or she did not haveactive, relevant information at that time.
Contribution
A substantive contribution to discussion contains task-relevant information. Schegl-off (1995), however, argued that information gets its sense, meaning, or function
because of the social action in which it is embedded (see also Jackson, Jacobs, &
J. A. Bonito Local Model of Information Sharing
Communication Theory 17 (2007) 252–280 ª 2007 International Communication Association 259
Rossi, 1987). A useful way to think of this is to characterize information as static andaction as dynamic. Thus, the same information item can be used in interaction in
a variety of different ways, and a given usage influences one’s interpretation of thedata as well as one’s response to it. For example, consider the datum ‘‘Hartford is the
capital of Connecticut.’’ In one case, it is a perfectly reasonable answer to the ques-tion, ‘‘What is the capital of Connecticut?’’ In another, it can function as a correction(among other things, e.g., a display of status) when uttered in response to the
question, ‘‘Isn’t New Haven the capital of Connecticut?’’ Thus, although the contentof the replies in the two examples is obviously the same, it is nonetheless unsatis-
factory to consider the replies as equal or even similar actions.
Activation
The preceding suggests that the mentioning of information is related to one’s knowl-
edge resources (the quality, quantity, and integration of data in long-term memory)and characteristics of ongoing interaction (what has been said, how, and what isprojected to be said and how). Thus, information is both an influence on and an
outcome of the process by which messages are produced during discussion. As anoutcome, information reflects some correspondence to the knowledge that generated
it. As an influence, it activates certain knowledge units but leaves the remainder atresting levels. These activated knowledge units continue the process by providing the
basis for subsequent contributions to discussion. Importantly, knowledge unitsmight be connected in one or more ways at different levels of abstraction (Greene,
1997; Pennington & Hastie, 1993).It is important to note that the source of activation can be the contributions of
others, one’s own contributions, or the connections between knowledge units inlong-term memory. Following Kenny and Cook (1999; Cook & Kenny, 2005), acti-vation of particular knowledge units via the contributions of one’s colleagues is the
partner effect—activation is from an external source. The actor effect describes acti-vation as a function of one’s own contributions to discussion or by the connections
among knowledge units in long-term memory. The former is an external actor effect,as one’s contributions to discussion are designed to fit in with what has preceded it—
hearing one’s contributions in the context of discussion may activate subsequentthoughts (Levy, 1979). Connections within long-term memory are internal actor
effects, an example of which is train of thought (Nijstad et al., 2003). This is repre-sented in Figure 1 by the arrow from potential contributions to knowledge. Onemight, for example, have as an active item the phrase ‘‘gives essay exams only,’’ which
has the potential to activate other related units. The originally activated unit may ormay not be contributed, but its role in the activation of other information is critical
for developing potential contributions to discussion.
Choice
As noted, choice refers to the decision to contribute substantively to discussion.
Choice applies only to potential contributions; those participants without relevant
Local Model of Information Sharing J. A. Bonito
260 Communication Theory 17 (2007) 252–280 ª 2007 International Communication Association
active information to contribute cannot in principle choose to contribute substan-tively, as there is no potential contribution to consider. For those with active infor-
mation in short-term memory, there are myriad reasons why one might choose tocontribute what they are thinking. Many factors influencing choice are related to
variables exogenous to discussion; several recent reviews have documented these(Bonito & Hollingshead, 1997; Gill, Menlo, & Keel, 1984; Stasser & Vaughan,1996). But other issues related to discussion might also influence choice. Recall
the example above in which Participant B must choose whether to contribute theactive item gives only essay exams. The choice to contribute this information has
much to do with B’s assessment of the situation, including her status relative to thatof Participant A, the extent to which confrontation is desirable or necessary, and
whether other members might prefer Smith or other candidates.
Shared and unique information: A local approach
In this section, I apply the local model to information-sharing research. As noted, thelocal approach distinguishes between activation and choice, with the former based
on message production processes and the latter related to social mechanisms thatunderlie turn-taking. Activation is primarily a local phenomenon; exigencies related
to discussion influence which data points are activated. Choice can be both local andglobal, although global influences on choice can develop from local sources, as when
one’s contributions on a given topic mark him or her as more competent thanothers, and perceptions of competence influence subsequent interaction (Fisek,
Berger, & Norman, 1991). Moreover, the process is serial, with activation precedingchoice; one cannot choose to participate substantively if he or she has no active
information to contribute. My goal here is to show how the three explanations forthe shared information bias (i.e., CIS, preference consistency, and social compari-son) might be incorporated in the local model for explaining particular aspects of
activation and choice.The preference consistency and CIS models are activation-based processes. The
preference consistency approach describes a substantive association among dataunits that is only incidentally related to the intermember relations among units.
Recall that shared knowledge, by design, points to an initial, but suboptimal, solu-tion. In most cases, participants do not know that their preference-supporting data
are shared—they typically discover via discussion that some information is shared,some not.7 Preference-supporting data then are associated because of their valencetoward one’s preferred solution, and valence is often an important dimension of
information processing (see Geva, Mayhar, & Skorick, 2000). Should a membermention preference-supporting information (which, as noted, is likely to be shared),
then other similarly valenced (toward a particular solution) items are likely to beactivated in both self and other.
The CIS, as noted, does not make any assumptions regarding the substantiveconnections among knowledge units. Instead, it assumes that shared items have
a distributional advantage within a group that makes those items more likely to
J. A. Bonito Local Model of Information Sharing
Communication Theory 17 (2007) 252–280 ª 2007 International Communication Association 261
be recalled and mentioned than are unique data. The CIS is, however, capable ofexplaining activation in limited cases precisely because the model does not consider
substantive associations among data. One place where the CIS potentially fits is topicswitches. If a given topic is exhausted, the selection of information on which to start
the next topic may depend, in part, on some sort of probabilistic process that favorsshared information. Once the group moves on to the new topic, however, the CIS isno longer a viable explanation, in terms of a local model, of subsequent contributions
(at least until the next topic shift occurs).Social comparison explanations account for choice within the local model. My
concern here is with developmental (i.e., within discussion) aspects of choice, ratherthan preexisting member or task characteristics that might predispose one to par-
ticipate. And although there are several means by which choice might develop ingroups (e.g., Shelly & Troyer, 2001), I focus on intermember distributions of data
because they have the potential to affect the choice to contribute active information.Researchers have offered several different models for the effects of mentioned
information on factors that plausibly influence choice. For example, Kameda, Oht-
subo, and Takezawa (1997) argued that members are distinguished by their cognitivecentrality. Cast in social network terms, the cognitively central participant has the
most information ties with the other members of the group; those on the peripheryhave fewer ties. As Kameda et al. noted, one’s centrality score ‘‘represents the number
of arguments that [he or she] shares with other members’’ (p. 298). Cognitivelycentral members play a role in mapping the content of a person’s contribution to
perceptions of that person’s ability and competence. Wittenbaum et al.’s (1999)mutual enhancement model similarly argues that participants who mention shared
information often are the beneficiaries of positive affective and credibility percep-tions. In addition, Larson et al. (2002) noted that, under some circumstances, thementioning of unique information was positively associated with perceptions of
persuasiveness. In any case, increased social standing often leads to more speakingopportunities in groups.
Finally, Wittenbaum et al.’s (2004) motivated information-sharing model isa much different approach to explaining information sharing, although it too is best
conceptualized as a problem in choice. They argued that previous explanations ofinformation sharing made untenable or unrealistic assumptions regarding the con-
ditions under which participants might mention information (e.g., cooperation).Wittenbaum et al.’s model assumes that participants strategically mention informationin pursuit of individual goals, and that the form in which information is mentioned
can be purposefully indirect or veiled. Still, the model assumes a global approach inwhich member goals influence what is said and when, but it does not describe the
conditions under which information is retrieved from long-term memory.
Value of intermember connections
One last issue to address is the extent to which intermember connections are salient
for and affect choice during discussion. As Wittenbaum et al. (2004) noted, models
Local Model of Information Sharing J. A. Bonito
262 Communication Theory 17 (2007) 252–280 ª 2007 International Communication Association
of information sharing assume, no doubt because of the intellectual and methodo-logical indebtedness to PAT, that unique information is more important, hence
more interesting, than is shared information. This is certainly true in hidden-profiledesigns, but other types of knowledge distributions exist in real-world group settings,
and the value of shared and unique data may vary depending on the circumstances.Three scenarios are plausible. The first, taken for granted, occurs when circumstancesdictate that participants know or assume prior to discussion whether information is
differentially distributed within the group. For example, members of a faculty searchcommittee know, or at least assume, that each has access to the same pool of
curricula vitae. The second scenario is informed, in which members might be toldor know in advance that some information is shared, as happens in some research
designs (Stasser, Steward, & Wittenbaum, 1995), as well in organizational teams inwhich some members are responsible for certain subsets of information pertaining to
the task (e.g., Sigman, 1984). The remaining scenario is discovery, in which partici-pants assume little about potential preexisting informational differences; memberscome to learn of such differences incrementally, and perhaps subtly, though con-
tributions to discussion. This scenario is typical of many hidden-profile designs, aswell as initial phases of work groups.
Each scenario plausibly influences the role of unique and shared information onchoice to participate. Shared information in the taken for granted scenario above
may have little effect, as participants already know that most of the information isshared. Mentioning shared information results in little, if any ‘‘bounce’’ in percep-
tions associated with participation. Unique information, however, learned via pro-fessional or personal contacts, for example, might have more weight in terms of
affecting choices to participate. The informed scenario primes members to evaluateinformation differences, which may influence the effects of shared and unique infor-mation. For example, shared information might instigate the development of sub-
groups, whereas unique information has the potential to be persuasive. Discovery,especially in the development of group norms and expectations, might have differ-
ential, but useful, consequences for the mentioning of unique and shared informa-tion; one who mentions unique information might be perceived as influential,
whereas the mentioning of shared information perhaps increases group cohesion.
Information exchange during interaction: An example
In this section, I briefly analyze a pair of transcripts from an experiment on in-formation sharing (Bonito, DeCamp, Coffman, & Ruppel, 2006). The purpose of
the analysis is to apply concepts of the local model, especially those that describethe relation between discussion and activation, to group interaction. Given that
researchers have been working on information sharing for over 20 years, usingwell-established models and research designs, it might seem out of place to offer
a qualitative evaluation of information-sharing discussions. Qualitative assessmentsare often useful during early stages of model and methodological development
(Edmondson & McManus, in press), but such work, to my knowledge, has never
J. A. Bonito Local Model of Information Sharing
Communication Theory 17 (2007) 252–280 ª 2007 International Communication Association 263
been done on information-sharing discussions. Although ultimately my goal is tooffer a set of propositions (see below) that are amenable to quantitative analysis,
close examination of interaction provides insight into the phenomenon that is oftenmissed by reliance on standard designs and coding methods. As Bakeman and Gott-
man (1986) noted:
We need to observe, and we need to do it very well, with imagination, with
boldness, and with dedication. If we do not observe, we shall never see what isthere. If we never see what is there, we shall never see the patterns in what is
there. Without the patterns, there will never be the kind of theory that we canbuild with. (p. 200)
The two transcripts are from two different discussions in which the ‘‘candidate’’ taskwas used. Both groups contained four participants, all of whom were female.
Although I feel that these are representative transcripts, there is little doubt thatexaminations of other discussions from this data set, or data from other similarstudies, would reveal characteristics not seen in this one. Still, the goal is to highlight
features of the local model as they apply to information sharing in groups. It shouldbe noted that the analysis of discussion data, obviously, cannot show the effects of
discussion on attitudes and cognitions related to choice. Entirely different methodsand analyses are necessary, and some suggestions are provided below.
The contributions in the transcripts are presented at the turn level rather than insmaller units (e.g., thought units), and the punctuation is conventionally placed (see
Edwards, 1993). The text in italics is unique (i.e., given to just one member prior todiscussion to memorize), the text in bold shared (all members received the same
information), and the underlined text was mentioned previously in the discussion.Participants were told prior to discussion that the information sheets were notidentical but they were not told anything about the actual distribution of the data.
Thus, participants had no way of knowing which information was shared and whichwas unique. The line numbers refer to speaking turns.
Group 2005-2-22-2
1 A: I liked number two.8
2 B: Martin was the first one. So Smith was the second one.
3 D: Jones was third.4 B: I thought Martin was the best because in the instructions it said not to base
it on an easy teacher, base it on other things. I just thought he had more
experience and he was an expert in communication. I think it is realistic thatyou can miss 2 to 3 days. That is pretty cool because some teachers are
really strict on attendance policy and you get points off and it didn’t seemlike he was anything like that.
5 D: Yeah, that is cool.6 B: I think multiple-choice exams are pretty self explanatory. Pretty easy as
long as you know the information you will be able to pinpoint. Whereas
Local Model of Information Sharing J. A. Bonito
264 Communication Theory 17 (2007) 252–280 ª 2007 International Communication Association
the other ones had essay formats which makes it a little tougher on thestudent. I think Jones was the only one who gave one exam.
7 A: The final.9
8 B: Yeah, I don’t like that because then you don’t see how you are doing. If you
screw up on that then you are screwed.9 D: Yeah.
10 A: The thing I liked about Martin no Smith was the essays I kind of liked.
Just because it is something different. With multiple choice you eithermemorize it or you didn’t. That might not be too great. I also liked how they
said the lectures were interesting. I am kind of big on that. I am just trying togo to class everyday this semester and if I do and I get something out of the
lecture it will actually keep my attention with the course.11 C: Yeah.
12 B: Yeah.13 D: Yeah.14 A: Then if I actually know what is going on and stuff. That is what I really liked
about that one.15 B: That is true because Martin did show a lot of videos, which I think can be
a little boring. But at the same time I think Smith is the one that didn’tallow outside speakers and sometimes the outside speakers give you dif-
ferent views on things.16 A: That is true.
17 B: Rather than hearing the same voice every time.
To begin, A and B discover at the start that they have different preferences; C andD do not offer, or even hint at, their opinions. Not surprisingly, then, A and Bprovide most of the substantive turns during this segment. The information con-
tributed to this discussion results in participants orienting to only two of severalpotential themes. The first concerns the desirability of testing formats (multiple
choice and essay) in Lines 6–10, the second (Lines 10–17) the ‘‘interestingness’’ ofthe two candidates, Smith and Martin. Notice that other potentially relevant themes,
including expertise and attendance policies, both mentioned by B in Line 4, do notgain any traction at this point in the interaction.
The concern with interestingness merits further analysis because it shows howparticipants might use relevant information to make their points. Participant Abegins a comparatively lengthy turn in Line 10 by picking up the issue of testing
practices as advanced by B in Line 6. Martin, B’s preference, uses multiple-choiceexams, which B likes because (a) ‘‘it is something different’’ and (b) they are ‘‘pretty
easy as long as you know the information you’ll be able to pinpoint,’’ whereas essayexams make it ‘‘tougher on the student.’’ Participant A contests this argument by
noting that multiple-choice formats depend on memorization and that not havingmemorized the right material ‘‘might not be too great.’’ The implication in her
argument is that essay exams allow some leeway in what counts as an acceptable
J. A. Bonito Local Model of Information Sharing
Communication Theory 17 (2007) 252–280 ª 2007 International Communication Association 265
or correct answer, whereas multiple-choice formats do not. Rather than continuingto pursue this point, A takes another tack, offering a unique information item
regarding Smith’s skill in giving interesting lectures, which, she argues, will helpher pay attention during class lectures. Although in one sense this contribution
signifies a change in topic, it is also (a) linked to the general issue of teachingpractices and (b) is another point in support of her choice, Smith.
Because of the multiple issues addressed by A in Lines 10 and 14, B has five (and
probably more) relevant response options. The five are (a) revisit the argumentabout what constitutes desirable testing practices, (b) address the assertion that
Smith’s lectures are interesting (because B did not have that information and shecould have opted to explore the issue further), (c) discuss the extent to which
interesting lectures improve student performance, (d) contest or agree with theimplication that Smith is a more interesting lecturer than Martin, and (e) move
on to another issue or topic. It is clear, however, where B stands on testing issues (seeLine 6), so it makes less sense for her to revisit that argument. In addition, it isevident that she opts not to contest that Smith’s lectures are interesting (which, in the
context of the ‘‘informed’’ nature of the information, is sensible). She continues onwith the interestingness issue because, as will be shown, she has more to say about it,
especially the implication that Smith is the more interesting instructor.B begins the turn by appearing to agree (‘‘that’s true’’) with the notion that
interesting lectures keep students’ attention focused on the material. She continues,however, using the term ‘‘because’’ to link the agreement (‘‘that’s true’’) with the
shared item regarding Martin’s proclivity to show videos during class, which shecontends is ‘‘a little boring.’’ Participant A never directly stated that Smith was
a more interesting lecturer than Martin, but B’s remark, especially the use ofdiscourse marker ‘‘because’’ (Schiffrin, 1987), seems to both make the issue moreexplicit (because it offers a direct comparison of teaching practices that is unflat-
tering to Martin) and cede the point to A that Smith is more interesting. But B isapparently not so willing to give in on the interestingness issue, as she mentions
that Smith did not bring in outside speakers (a shared information item), which,she contends, suggests that Smith’s teaching practices are not as interesting as
would first appear (‘‘rather than hearing the same voice every time’’). Participant Athen agrees with this point, which seemingly leaves unresolved the issue of who is the
more interesting instructor.The preceding demonstrated that the mentioning of information is based on
interitem connections, ones relevant to characteristics of ongoing interaction,
including substantive issues (e.g., topic) as well as pragmatic ones (e.g., rebuttingarguments). The local model does not, however, preclude that intermember associ-
ations affect interaction, but it assumes that the relationship differs from thatbetween interitem connections and discussion. Rather than a kind of topical coher-
ence characteristic of interitem associations, intermember connections—when madesalient during discussion—often lead to other kinds of exchanges (e.g., repair, ver-
ification) that are sometimes important for subsequent participation and group
Local Model of Information Sharing J. A. Bonito
266 Communication Theory 17 (2007) 252–280 ª 2007 International Communication Association
outcomes. The following extract, which begins 14 lines into the discussion, showsone way in which participants might address intermember issues during interaction.
Group 2005-2-28-5
14 B: The only reason I didn’t choose Smith was because like I said I was writingthem out I was like oh wait but I guess the only reason I didn’t was becauseof the easy A and then also that he didn’t have any outside speakers. I
thought that would be bad too.15 D: That he was always late to class.
16 B: Is that what it said?17 A: Oh I didn’t know that.
18 C: I didn’t see that.19 A: Mine didn’t say that.
20 B: Yeah mine didn’t either. It said that Martin was always on time but Smithcould talk to people after class.
21 A: I know we weren’t supposed to be personal but it seemed like Martin was
only interested in what he was studying and he had no outside knowledgeor he didn’t use communication outside in the world and I think that is
important.
Participant B, who initially preferred Martin, begins to list information (which Ahad mentioned previously) in Line 14 that in her estimation disqualifies Smith as
a desirable candidate. Participant D (Line 15), who has been silent up to this point,responds with a unique item (‘‘Is always late to class’’) that continues the campaign
against Smith. Rather than pursue the theme, however, the other three participantsfollow this remark by noting that, in effect, they did not receive that item (‘‘late toclass’’) to memorize. These remarks appear to be based on intermember associations
and function to highlight information differences (ones that they were told prior tothe task might exist). But the concern with information differences does not last very
long. It would have been relevant at that juncture for participants to further exploretheir information differences, and in fact, Participant B, in Line 20, appears to do just
that. She lists two items (both of which happen to be unique, although one isrepeated), prefacing them with ‘‘Yeah mine didn’t either. It said . . . ’’ with the
‘‘it’’ referring to the information sheet she was given to memorize prior to thediscussion. If this were an attempt to simply list her information resources, thenany item would have sufficed to begin the list. But it is not coincidental that she
would begin with ‘‘Martin was always on time’’ because it is directly related to D’sclaim (Line 15) that Smith is always late. The second item in that contribution is yet
another point against Smith, as it addresses what is apparently his limited availabilityto students. Thus, rather than become a listing of information resources, which
would have been a consequence of intermember associations, the discussion driftsback to the original theme, one related to the desirability of candidates. The theme
continues when Participant A (in Line 20), a Smith advocate, turns attention toMartin’s potential problems, noting that Martin seems too immersed in his research
J. A. Bonito Local Model of Information Sharing
Communication Theory 17 (2007) 252–280 ª 2007 International Communication Association 267
and that he ‘‘doesn’t use communication in the outside world’’ (which hearkens tothe knowledge item ‘‘has no experience practicing communication in the real
world’’).In sum, this relatively cursory look at some extracts from the discussion high-
lights the role of interitem connections for information sharing but also notes whenintermember connections affect discussion. Regarding the second extract, it is clearthat information differences do not become much of an issue during that part of
discussion, but there is no way of deducing the extent to which such differencesmanifested themselves as relevant cognitions and attitudes regarding participants.
For example, in the second transcript, Participant D was largely silent until intro-ducing the unique item (and it was unknown to her that it was unique), which
changed the focus, however briefly, of the discussion. One wonders how D wasperceived by her colleagues at that time and if such perceptions translated into
opportunities for her subsequent participation. Transcriptions rarely reveal that kindof information. Instead, other techniques are necessary, and I provide some possi-bilities for assessing them below.
Propositions
My goal in the previous section was to demonstrate, via analysis of discussion tran-
scripts, that the mentioning of information is related to the characteristics of unfold-ing interaction. The purpose of this section is to offer a set of propositions that are
amenable to quantitative analysis. Testing of the propositions depends crucially onthe researcher identifying potentially salient substantive connections among knowl-
edge units, as well as distributing information within groups in theoretically relevantways. One would then examine discussion-relevant topics or issues that might cor-respond with those connections. For example, one would predict that substantively
relevant unique information is more likely to be mentioned than is less relevantunique information. Of course, choice plays a role as well in the sense that members
perceived as more able or competent are more likely to contribute active informationthan are participants who are perceived as less competent. In what follows, I offer
a set of propositions that might be derived from the model. Because the model treatsactivation and choice as two distinct parts of the process, I address each component
separately.
Activation
As noted, activation refers to the transfer of knowledge items in long-term memoryto short-term memory where they become the basis for potential contributions to
discussion. The local model posits that activation is based on interitem relationships,and that both discussion and searches through long-term memory initiate the pro-
cess. Because intermember associations are not directly responsible for activation ofnew units, shared and unique units become active because their content is relevant to
the issue under discussion at the moment.
Local Model of Information Sharing J. A. Bonito
268 Communication Theory 17 (2007) 252–280 ª 2007 International Communication Association
Proposition 1: Shared knowledge is likely to be activated when its substantive
characteristics are connected to characteristics of ongoing discussion.
Proposition 2: Unique knowledge is likely to be activated when its substantive
characteristics are connected to characteristics of ongoing discussion.
Proposition 3: Shared knowledge is likely to be activated when its substantive
characteristics are related to the characteristics of other items stored in long-term
memory.
Proposition 4: Unique knowledge is likely to be activated when its substantive
characteristics are related to the characteristics of other items stored in long-term
memory.
Proposition 5: Shared information is more likely than unique information to be
mentioned when topic switches are relevant.
Choice
Choice refers to the decision to contribute active information to discussion.
Although choice can be related to variation in many preexisting situational factorsor to a host of individual differences, the propositions here focus on the effects
of information uniqueness on decisions to contribute to discussion. Drawing oninformation-sharing research, the local model assumes that the contribution ofshared and unique items affects the perceptions that contributors have of each other.
For example, cognitively central participants have increased social standing becausethe information content of their contributions is similar to those of other members
(Kameda et al., 1997). Furthermore, perceptions relevant to turn-taking influencesubsequent decisions to participate (Wittenbaum & Bowman, 2004). Thus, cognitive
centrality should be positively associated with decisions to contribute informationduring discussion.
The preceding, however, is complicated by several factors. The first is that muchof the work on information sharing specifically (and on participation in general)
involves zero-history groups, which means that the relation between participationand turn-taking develops over time (Shelly & Troyer, 2001). In such cases, partici-pation early in a discussion influences relevant perceptions, which in turn are con-
sequential for subsequent interaction (Bonito, 2006). In contrast, continuing groupslikely have existing structures in place that might moderate the relation between the
mentioning of information and perceptions related to turn-taking (Skvoretz, 1981).This claim is consistent with both structurational approaches to groups (e.g., Poole,
Seibold, & McPhee, 1985) and with Kenny’s (2004) model of person perception,which shows that perceptions tend to level out over time, regardless of the triggering
behaviors in question. The second factor is, as noted above, that the value of uniqueand shared information, in terms of their influence on relevant perceptions andsubsequent interaction, varies as a function of a group’s context (Wittenbaum et al.,
J. A. Bonito Local Model of Information Sharing
Communication Theory 17 (2007) 252–280 ª 2007 International Communication Association 269
2004). For example, if each member knows or assumes that everyone in the grouphas the same information, mentioning shared information is likely to have a negli-
gible impact on relevant perceptions and subsequent participation.These observations lead to the following propositions regarding the choice to
contribute active information to discussion:
Proposition 6: The mentioning of shared or unique information influences the
development of task-relevant judgments or perceptions related to choice.
Proposition 7: As task-relevant judgments or perceptions related to choice develop, they
influence subsequent choices to participate.
Proposition 8: The relation between mentioning information and task-relevant
judgments or perceptions related to choice is moderated by group history.
Proposition 9: The relation between mentioning information and task-relevant
judgments or perceptions related to choice is moderated by information distributions
within the group.
Methodological and analysis considerations
The model advanced here raises several potentially interesting methodological and
analysis issues. Because activation is central to the model, it is important to measurethe amount and kind of active information possessed by participants over time. Also,
given that activation is thought to be a function of contributions to discussion, itrequires the examination of the relation between active information and discussion
topics or events. Obviously, one cannot stop discussion at regular intervals or fol-lowing some relevant event (e.g., a topic switch) and ask participants to report ontheir cognitions. A possible design is stimulated recall in which the researcher video-
tapes discussion and then has participants view the recording (Waldron & Cegala,1992). The researcher can stop the tape at relevant junctures or at regular intervals,
and participants can respond to protocols or survey instruments regarding percep-tions and cognitions of interest. Research examples include Waldron’s (1997) study
of conversational goals during AIDS discussions and Bonito’s (2006) study of therelationship between perceptions of ability and participation over time during group
discussions. Although there are concerns regarding the validity of such designs (seeWaldron & Cegala, 1992), there is currently no better way to examine the relationbetween cognitions and discussion over time.
Because discussion is assumed to influence the activation of knowledge from long-term memory, it is important to evaluate discussion content and to assess its relation-
ship to activation. There is no correct way to do this, as there are many valid ways tocode discussion (O’Keefe, 1987), depending on (among other things) the research
question and the nature of the task. The example used throughout this paper, the‘‘choose the best candidate’’ task, has topical and other elements that might differ
from other tasks. The task is intellective in nature, which means that some of the
Local Model of Information Sharing J. A. Bonito
270 Communication Theory 17 (2007) 252–280 ª 2007 International Communication Association
information ‘‘naturally’’ points away from the correct answer, and that other infor-mation highlights it. Given this, it makes sense to code for valence, as well as by topic
or issue (e.g., the items regarding testing practices). Other tasks are more judgmentalin nature; there is no obvious relation between information and the solutions the
group might choose. For example, the ‘‘psychological profile task’’ (Bonito, 2001) asksparticipants to form judgments of a target person based on a description of hisbehaviors. Because there is no correct judgment, and ultimately no correct solution
to the task, it makes less sense to code discussion for valence. It is perhaps moredesirable to code discussion for items that reflect elements of impression formation.
Linking information to discussion is obviously a difficult task. One method isto identify topic shifts during discussion. Several authors (see Crow, 1983; Sanders,
Spooren, & Noordman, 1992; Schiffrin, 1987) have approached the problem in slightlydifferent ways, but both systems depend on identifying markers that either signify
connections among items (e.g., the use of ‘‘because’’ noted in the analysis of the dis-cussions above) or mark shifts into different substantive issues. Regardless of the methodchosen, the identification of substantive or topical characteristics of discussion is integral
to developing testable questions from the propositions. For example, if participantsdiscuss teaching practices, then participants who possess the data related to that topic
are more likely to contribute than are members who do not have such knowledge.One goal of information-sharing research is the ability to predict the conditions
under which unique information is more or less likely to be mentioned. According tothe model advanced here, the activation of shared and unique items is largely depen-
dent on the substantive (i.e., interitem) connections among units. Thus, it behoovesresearchers to manipulate information so that unique and shared items correspond
to particular substantive characteristics of information. For example, consider thethree items, ‘‘gives multiple-choice exams,’’ ‘‘gives essay exams,’’ and ‘‘is 30 to 40years of age’’ from the ‘‘choose the best candidate task.’’ Assume that the multiple-
choice item is unique and that the other two are shared. The mentioning of the essayexam item is more likely to activate the item about multiple-choice exams than the
age item simply because the multiple-choice item is more relevant than the age item.The age item may become relevant later, when some other topic (e.g., personal
characteristics) emerges; if the topic does not emerge, however, then the age itemis less likely to be activated and mentioned. Scholars need to investigate more clearly
the arrangement of substantive and intermember relations in order to better predictthe outcomes of interest. This may entail designs in which the same information itemis unique but topically related to other items in one condition, and is shared but not
substantively related to other items in another condition.The local model presupposes a multilevel approach to the study of participation
generally and to information sharing specifically. Although there are several extantapproaches to multilevel modeling of groups (e.g., Bonito, 2002; Gonzalez & Griffin,
2002; Hayes, 2006), I present Cook and Kenny’s (2004) modeling of family assessmentsas an example of issues facing the analysis of groups. Based on Kenny’s (1994)
approach to social perception, the model posits four general effects. The first two,
J. A. Bonito Local Model of Information Sharing
Communication Theory 17 (2007) 252–280 ª 2007 International Communication Association 271
perceiver effect and target effect,10 are individual-level effects. The perceiver effectdescribes the consistency of a participant’s behavior toward or perceptions of other
members of the group (e.g., a person rates his or her colleagues high on a givendimension). The target effect refers to the behaviors directed toward or the perceptions
of a particular member by his or her colleagues (e.g., a target receives consistently high[or low] ratings). The relationship effect is at the level of the dyad, and refers to theunique behaviors or perceptions between a pair of individuals within the group, above
and beyond perceiver and target effects. Finally, the group effect refers to the dynamicsof the group that affect behavior within it. For example, overall participation in some
groups is less than in others, and participants in low-participating groups will obvi-ously speak less frequently than members in high-participating groups.
The implications for the model on studies of information sharing are potentiallyprofound. For example, researchers often examine group means for the mentioning
of information items (e.g., Stasser & Stewart, 1992; Stasser, Vaughan, & Stewart,2000). Although this is often justified because the interventions are at the group level,such analyses ignore potential lower-level phenomena. Some individuals might be
more inclined to participate than others, and some individuals are more likely thanothers to be given and to take speaking turns. Pairs of speakers often dominate the
floor during group discussion (Parker, 1988; Stasser & Taylor, 1991), leaving othermembers relatively silent. Moreover, such analyses do not tell us the extent to which
behaviors of interest are associated within groups; the reporting of intraclass corre-lations is still unfortunately rare in group research. One example is Bonito’s (2003)
analysis of information sharing in mediated groups, in which he noted that thementioning of unique information was positively associated within groups under
certain conditions. Within a local model, such associations are expected; the localmodel posits that the mentioning of certain items is related to the mentioning ofsimilar items by others. It would be interesting to evaluate if within-group associa-
tions better explain outcomes that do analyses of group means.
Conclusions
The purpose of this paper was to offer a local model of information sharing asa complement to existing models. The local model described above is rooted in
social–interactional approaches to group process (Meyers & Seibold, 1990) and isultimately a model of coherence for small-group discussion. Information resourcesprovide different relevant themes for discussion, and the ones participants ultimately
choose establish the coherence of the exchange.11 Although relevance is the basis foractivation, it is not always the case that participants contribute active information.
Thus, the model contains a choice component that posits the conditions under whichmembers are more or less likely to contribute active information to discussion. For
current purposes, choice is affected by previously contributed unique and sharedinformation, although in principle choice is affected by a plethora of exogenous (to
discussion) factors, as well as other discussion-based issues.
Local Model of Information Sharing J. A. Bonito
272 Communication Theory 17 (2007) 252–280 ª 2007 International Communication Association
One would think that information seeking would be an important factor inmodels of information sharing but such is not the case. Information exchange in
these models is assumed to be a passive enterprise, in which participants, for a varietyof reasons, either choose or choose not to contribute information. But research on
information seeking reminds us that participants often actively seek out information,usually to reduce uncertainty, in a variety of contexts, including organizational(Morrison, 2002), group (Durham, Locke, Poon, & McLeod, 2000), interpersonal
(Knobloch & Solomon, 2002), and health (Brashers et al., 2000). In fact, mosthidden-profile manipulations create uncertainty by alerting participants to unequal
distributions of information but without specifying the nature of the distribution.One would think that some participants would engage in information-seeking
behaviors under such conditions. Models of information sharing need to accountfor information-seeking behaviors and their effects in small-group discussion.
This is not a model of decision making per se, as it does not trace the relationbetween the process and the outcomes. In that sense, it falls short of Hewes’s (1986,1996) quite stringent criterion for demonstrating that communication unambigu-
ously affects group decision making. Many putatively communication effects onoutcomes, Hewes has argued, can be attributed wholly or in part to factors exoge-
nous to discussion (e.g., group composition), which reduces communication tonothing more than a conduit through which preexisting preferences, arguments,
tactics, and the like are made manifest (Corman & Kuhn, 2005). Going a step farther,Hewes proposed a socioegocentric model of discussion, in which the baseline
assumption is that contributions to discussion are unconnected, in the sense thatany given message occurs without regard to what precedes and follows it. This
baseline is meant to be evaluated against actual discussion, which, one would hope,does not exhibit such egocentric characteristics. Several scholars (Corman & Kuhn,2005; Pavitt & Johnson, 1999) have taken up this challenge by evaluating the extent
to which group interaction is coherent. This paper continues that line of work, inapplying the notion of coherence to information sharing in groups, and focuses
primarily on how information is related to the process.Given the preceding, and following Pavitt (1993), this paper offers a softer, or
perhaps more encompassing, role for communication in group discussion byeschewing ‘‘prediction’’ for ‘‘mattering.’’ Pavitt showed how communication matters
to discussion in a variety of ways, including an increase in shared awareness, theestablishment and modification of procedures, and the assignment of weights todecision alternatives, all of which are potentially consequential for group outcomes.
Thus, models and descriptions of group communication processes might legiti-mately focus on consequences that are not directly related to group outcomes but
that nonetheless play a role in how a group, whose members likely vary in terms ofknowledge, ability, and preferences, progresses from the beginning stages of discus-
sion to an output that more or less reflects the collective will of the group.My claim, in keeping with the spirit of mattering as described above, is that the
process of information sharing, and thus the quality of the collective information
J. A. Bonito Local Model of Information Sharing
Communication Theory 17 (2007) 252–280 ª 2007 International Communication Association 273
database itself, is a local phenomenon, in the sense that a contribution containing a bitof information has some consequences for the contributions that follow. Because
participants typically do not make incoherent or random contributions to discussion(Pavitt & Johnson, 1999), it is logical to assume that the information contained in
a given utterance influences subsequent contributions, including the type, if any, ofinformation presented in them. Because there are several different options for main-taining coherence in subsequent contributions (i.e., several different bits of informa-
tion might be coherently presented), it is not always knowable before the fact howparticipants will orient and respond to substantive comments in discussion. But such
orientations and responses are consequential for the collective database.This perspective also serves as a useful counterargument to the rather invidious
upshot to the mediated view that communication is responsible for process loss.Imagine that each group member brings unique and useful information to a discus-
sion but that not all of that information is presented during the discussion. In fact,this is an all-too-common finding in most research involving information pooling, atleast for tasks in which prediscussion distributions of knowledge are manipulated by
or known to the researcher (Stasser & Titus, 2003; Stasser & Vaughan, 1996). Becausethe amount of information contributed to discussion falls short of the summed
information resources across the individual members, researchers assume that theprocess of communicating somehow constrained or otherwise prevented members
from contributing all of what they know to discussion.It is hard to imagine how members come to contribute and assess information
without a constitutive role for communication. Although shared and unique infor-mation clearly influence outcomes, the model proposed here assumes that they play
virtually no role in the activation of information; rather, the uniqueness of infor-mation is posited to affect choice. The local model, especially the activation com-ponent, suggests that unique and shared information ‘‘hitch a ride’’ on the bases
(e.g., semantic, function, and structure) for coherence. Without such a process, andin the absence of any type of telepathy, coherence provides the means by which
information is contributed to discussion. Without it, information would exist asknowledge units within members, with little possibility of being communicated to
and acted upon by other members.
Acknowledgments
The author thanks Robert E. Sanders, Mary DeCamp, Erin Ruppel, and three anon-ymous reviewers for their helpful comments.
Notes
1 The formula for determining the likelihood that a piece of information will be men-
tioned is p(D) = 1 2 [1 2 p(R)]n, where p(R) is the probability that an individual will
recall and mention a given piece of information, p(D) is the probability that an item
will be discussed by the group, and n is group size (Stasser & Titus, 1987).
Local Model of Information Sharing J. A. Bonito
274 Communication Theory 17 (2007) 252–280 ª 2007 International Communication Association
2 The term verify as used in the literature does not refer to the process by which any given
piece of information is deemed to be true. Instead, it refers to something of a social
process (e.g., cohesiveness) where the mentioning of a shared item generates the
knowledge that others know the same thing. In that sense, a shared item that is not
factually true (but unknown to be such by members) may nonetheless become verified
and reap the same social benefits as does factually true information.
3 It also explains to a certain extent why contradicting information (which is usually
unique) often fails to change initial preferences. Greitemeyer and Schulz-Hardt (2003)
argued that the principle of cognitive economy limits the amount of effort one is
willing to expend in evaluating contradicting information and incorporating it into his
or her initial preference. In short, it is easier to stick with one’s first assessment rather
than change it.
4 This point, made in contrast to Stasser’s assumption that each information item is
equally likely to be recalled, seems self-evident and begs the question of why one need
make the assumption in the first place. To be fair, the assumption is a necessary
constraint because of the inherent complexity in Stasser’s models of information
sharing and participation. Without it, the model parameters would quickly multiply,
making for an overly complex, and perhaps not very useful, set of prediction formulas.
5 O’Keefe and Lambert (1995) use the term propositional base to refer to data in long-
term memory. My use of the term knowledge here is in keeping with Propp’s dis-
tinction, which she meant to apply to the problem of information sharing in groups.
O’Keefe and Lambert’s terminology is meant to apply to interaction and message
production in general.
6 Most models of message production (e.g., Greene, 1997; Wilson, 1995) assume that the
strength of an association between units must exceed some threshold in order for
retrieval to occur.
7 It is a common practice in hidden-profile designs for the researchers to inform par-
ticipants that some data are shared and others of it not. But participants are usually not
told exactly which data are shared.
8 The information was presented in columns, one column per candidate. The first three
lines of this extract show participants identifying the candidates’ names by the column
in which each appeared.
9 The last comment in Line 6 and A’s fragment in Line 7 both refer to the same shared
item (‘‘gives a final only’’) but I chose not to identify Line 7 as repeated. Technically,
A is not repeating B’s comment but is instead providing elaboration. B was correct in
pointing out that Jones gave only one exam during a semester but was not as specific as
she could have been because, as A points out, that one exam is the final.
10 Cook and Kenny use the terms perceiver and target to refer to perceptual phenomena,
whereas the more general actor and partner can refer to behaviors or perceptions. I use
perceiver and target rather than actor and partner here to keep the discussion distinct
from my description of activation above.
11 Relevance and coherence are often used interchangeably, but as Sanders (1987) has
noted, this is not accurate. Relevance refers to the possible ways in which a contribution
might be related to its antecedent and potential contributions. Coherence is the specific
interpretation of an utterance given its ‘‘commonalities with both its antecedents and
consequents’’ (p. 84). Relevance refers to the several ways in which an utterance might
J. A. Bonito Local Model of Information Sharing
Communication Theory 17 (2007) 252–280 ª 2007 International Communication Association 275
be meaningful, whereas coherence is an achievement—it refers to the meaning or
function that an utterance comes to have given its place or orientation in interaction
(cf. Thagard, 1989).
References
Bakeman, R., & Gottman, J. M. (1986). Observing interaction: An introduction to sequential
analysis. Cambridge, UK: Cambridge University Press.
Bonito, J. A. (2001). An information-processing approach to participation in small groups.
Communication Research, 28, 275–303.
Bonito, J. A. (2002). The analysis of participation in small groups: Methodological and
conceptual issues related to interdependence. Small Group Research, 33, 412–438.
Bonito, J. A. (2003). Information processing and exchange in mediated groups:
Interdependence and interaction. Human Communication Research, 29, 533–559.
Bonito, J. A. (2006). A longitudinal social relations analysis of participation in small groups.
Human Communication Research, 32, 302–321.
Bonito, J. A., DeCamp, M. H., Coffman, M., & Ruppel, E. K. (2006, November).
A longitudinal analysis of information processing and information sharing in small groups.
Paper presented at the annual meeting of the National Communication Association,
San Antonio, TX.
Bonito, J. A., & Hollingshead, A. B. (1997). Participation in small groups. Communication
Yearbook, 20, 227–261.
Brashers, D. E., Neidig, J. L., Haas, S. M., Dobbs, L. K., Cardillo, L. W., & Russell, J. A. (2000).
Communication in the management of uncertainty: The case of persons living with HIV
or AIDS. Communication Monographs, 67(1), 63–84.
Burke, P. J. (1974). Participation and leadership in small groups. American Sociological
Review, 39, 832–843.
Burnstein, E., & Vinokur, A. (1977). Persuasive argumentation and social comparison
as determinants of attitude polarization. Journal of Experimental Social Psychology,
13, 315.
Chafe, W. L. (1994). Discourse, consciousness, and time. Chicago: University of Chicago Press.
Cook, W. L., & Kenny, D. A. (2004). Application of the social relations model to family
assessment. Journal of Family Psychology, 18, 361–371.
Cook, W. L., & Kenny, D. A. (2005). The actor-partner interdependence model: A model of
bidirectional effects in developmental studies. International Journal of Behavioral
Development, 29(2), 101–109.
Corman, S. R., & Kuhn, T. (2005). The detectability of socio-egocentric group speech:
A quasi-Turing test. Communication Monographs, 72(2), 117–143.
Crow, B. K. (1983). Topic shifts in couples’ conversations. In K. Tracy & R. T. Craig (Eds.),
Conversational coherence (pp. 136–156). Beverly Hills, CA: Sage.
Durham, C. C., Locke, E. A., Poon, J. M. L., & McLeod, P. L. (2000). Effects of group goals
and time pressure on group efficacy, information-seeking strategy, and performance.
Human Performance, 13(2), 115–138.
Edmondson, A. C., & McManus, S. E. (in press). Methodological fit in management field
research. Academy of Management Review.
Local Model of Information Sharing J. A. Bonito
276 Communication Theory 17 (2007) 252–280 ª 2007 International Communication Association
Edwards, J. A. (1993). Principles and contrasting systems of discourse transcription.
In J. A. Edwards & M. D. Lampert (Eds.), Talking data: Transcription and coding in
discourse research (pp. 3–31). Hillsdale, NJ: Erlbaum.
Fisek, M. H., Berger, J., & Norman, R. Z. (1991). Participation in hetero and homogeneous
groups: A theoretical integration. American Journal of Sociology, 97(1), 114–142.
Franz, T. M., & Larson, J. R. J. (2002). The impact of experts on information sharing during
group discussion. Small Group Research, 33, 383–411.
Geva, N., Mayhar, J., & Skorick, J. M. (2000). The cognitive calculus of foreign policy decision
making: An experimental assessment. Journal of Conflict Resolution, 44, 447–471.
Gill, S. J., Menlo, A., & Keel, L. P. (1984). Antecedents to member participation within
small groups: A review of theory and research. Journal for Specialists in Group Work,
9(2), 68–76.
Gonzalez, R., & Griffin, D. (2002). Modeling the personality of dyads and groups. Journal of
Personality, 70, 901–924.
Greene, J. O. (1997). A second generation action assembly theory. In J. O. Greene (Ed.),
Message production: Advances in communication theory (pp. 151–170). Mawah, NJ:
Erlbaum.
Greitemeyer, T., & Schulz-Hardt, S. (2003). Preference-consistent evaluation of information
in the hidden profile paradigm: Beyond group-level explanations for the dominance of
shared information in group decisions. Journal of Personality & Social Psychology, 84,
322–339.
Hayes, A. F. (2006). A primer on multilevel modeling. Human Communication Research, 32,
385–410.
Hewes, D. E. (1986). A socio-egocentric model of group decision-making. In R. Y. Hirokawa
& M. S. Poole (Eds.), Communication and group decision-making (pp. 265–291). Beverly
Hills, CA: Sage.
Hewes, D. E. (1996). Small group communication may not influence decision making: An
amplification of socio-egocentric theory. In R. Y. Hirokawa & M. S. Poole (Eds.),
Communication and group decision making (2nd ed., pp. 179–212). Thousand Oaks,
CA: Sage.
Jackson, S., Jacobs, S., & Rossi, A. M. (1987). Conversational relevance: Three experiments on
pragmatic connectedness in conversation. Communication Yearbook, 11, 323–347.
Kalma, A. (1992). Gazing in triads: A powerful signal in floor apportionment. British Journal
of Social Psychology, 31(1), 21–39.
Kameda, T., Ohtsubo, Y., & Takezawa, M. (1997). Centrality in sociocognitive networks and
social influence: An illustration in a group decision-making context. Journal of Personality
& Social Psychology, 73, 296–309.
Kenny, D. A. (1994). Interpersonal perception: A social relations analysis. New York:
Guilford Press.
Kenny, D. A. (2004). PERSON: A general model of interpersonal perception. Personality &
Social Psychology Review, 8, 265–280.
Kenny, D. A., & Cook, W. (1999). Partner effects in relationship research: Conceptual issues,
analytic difficulties, and illustrations. Personal Relationships, 6, 433–448.
Knobloch, L. K., & Solomon, D. H. (2002). Information seeking beyond initial interaction:
Negotiating relational uncertainty within close relationships. Human Communication
Research, 28, 243–257.
J. A. Bonito Local Model of Information Sharing
Communication Theory 17 (2007) 252–280 ª 2007 International Communication Association 277
Larson, J. R. (1997). Modeling the entry of shared and unshared information into group
discussion: A review and BASIC language computer program. Small Group Research, 28,
454–479.
Larson, J. R., Foster-Fishman, P. G., & Keys, C. B. (1994). Discussion of shared and unshared
information in decision-making groups. Journal of Personality and Social Psychology,
67, 446–461.
Larson, J. R., Sargis, E. G., Elstein, A. S., & Schwartz, A. (2002). Holding shared versus
unshared information: Its impact on perceived member influence in decision-making
groups. Basic & Applied Social Psychology, 24(2), 145–155.
Levy, D. M. (1979). Communicative goals and strategies: Between discourse and syntax. In
T. Givon (Ed.), Syntax and semantics: Vol. 12. Discourse and semantics (pp. 183–212).
New York: Academic Press.
Meyers, R. A. (1989). Persuasive arguments theory: A test of assumptions. Human
Communication Research, 15, 357–381.
Meyers, R. A., & Seibold, D. R. (1990). Perspectives on group argument: A critical review of
persuasive arguments theory and an alternative structurational view. Communication
Yearbook, 13, 268–302.
Morrison, E. W. (2002). Information seeking within organizations. Human Communication
Research, 28, 229–242.
Nijstad, B. A., Stroebe, W., & Lodewijkx, H. F. M. (2003). Production blocking and idea
generation: Does blocking interfere with cognitive processes? Journal of Experimental
Social Psychology, 39, 531–548.
O’Keefe, B. J., & Lambert, B. L. (1995). Managing the flow of ideas: A local management
approach to message design. Communication Yearbook, 18, 54–82.
O’Keefe, D. J. (1987). Message description. Paper presented at the Annual Meeting of the
Speech Communication Association, Boston.
Parker, K. C. (1988). Speaking turns in small group interaction: A context-sensitive event
sequence model. Journal of Personality and Social Psychology, 54, 965–971.
Pavitt, C. (1993). Does communication matter in social influence during small group
discussion? Five positions. Communication Studies, 44, 216–227.
Pavitt, C., & Johnson, K. K. (1999). An examination of the coherence of group discussions.
Communication Research, 26, 303–321.
Pennington, N., & Hastie, R. (1993). Reasoning in explanation-based decision making.
Cognition Reasoning and Decision Making, 49(1), 123.
Poole, M. S., Seibold, D. R., & McPhee, R. D. (1985). Group decision-making as
a structurational process. Quarterly Journal of Speech, 71(1), 74–102.
Propp, K. M. (1999). Collective information processing in groups. In L. R. Frey (Ed.), The
handbook of group communication theory and research (pp. 225–250). Thousand Oaks,
CA: Sage.
Ridgeway, C. L., & Smith-Lovin, L. (1999). The gender system and interaction. Annual Review
of Sociology, 25, 191–216.
Sacks, H., Schegloff, E. A., & Jefferson, G. (1974). A simplest systematics for the organization
of turn-taking for conversation. Language, 50, 696–735.
Sanders, R. E. (1987). Cognitive foundations of calculated speech. Albany, NY: SUNY Press.
Sanders, T. J., Spooren, W. P., & Noordman, L. G. (1992). Toward a taxonomy of coherence
relations. Discourse Processes, 15(1), 1–35.
Local Model of Information Sharing J. A. Bonito
278 Communication Theory 17 (2007) 252–280 ª 2007 International Communication Association
Schegloff, E. A. (1995). Discourse as an interactional achievement III: The omnirelevance of
action. Research on Language & Social Interaction, 28, 185–211.
Schiffrin, D. (1987). Discourse markers. Cambridge, UK: Cambridge University Press.
Shelly, R. K., & Troyer, L. (2001). Emergence and completion of structure in initially
undefined and partially defined groups. Social Psychology Quarterly, 64, 318–332.
Sigman, S. J. (1984). Talk and interaction strategy in a task-oriented group. Small Group
Research, 15(1), 33–51.
Skvoretz, J. (1981). Extending expectation states theory: Comparative status models of
participation in N-person groups. Social Forces, 59, 752–770.
Stasser, G., & Stewart, D. (1992). Discovery of hidden profiles by decision-making groups:
Solving a problem versus making a judgment. Journal of Personality and Social Psychology,
63, 426–434.
Stasser, G., Stewart, D., & Wittenbaum, G. M. (1995). Expert roles and information exchange
during discussion: The importance of knowing who knows what. Journal of Experimental
Social Psychology, 31, 244–265.
Stasser, G., & Taylor, L. A. (1991). Speaking turns in face-to-face discussion. Journal of
Personality and Social Psychology, 60, 675–684.
Stasser, G., Taylor, L. A., & Hanna, C. (1989). Information sampling in structured and
unstructured discussions of three- and six-person groups. Journal of Personality and
Social Psychology, 57(1), 67–78.
Stasser, G., & Titus, W. (1985). Pooling of unshared information in group decision making:
Biased information sampling during discussion. Journal of Personality and Social
Psychology, 48, 1467–1478.
Stasser, G., & Titus, W. (1987). Effects of information load and percentage of shared
information on the dissemination of unshared information during group discussion.
Journal of Personality and Social Psychology, 53(1), 81–93.
Stasser, G., & Titus, W. (2003). Hidden profiles: A brief history. Psychological Inquiry, 14,
304–313.
Stasser, G., & Vaughan, S. I. (1996). Models of participation during face-to-face unstructured
discussion. In E. H. Witte & J. H. Davis (Eds.), Understanding group behavior
(pp. 165–192). Mahwah, NJ: Erlbaum.
Stasser, G., Vaughan, S. I., & Stewart, D. D. (2000). Pooling unshared information: The
benefits of knowing how access to information is distributed among group members.
Organizational Behavior and Human Decision Processes, 82(1), 102–116.
Thagard, P. (1989). Explanatory coherence. Behavioral & Brain Sciences, 12, 435–502.
Tracy, K. (1982). On getting the point: Distinguishing ‘‘issues’’ from ‘‘events, ’’ an aspect of
conversational coherence. Communication Yearbook, 5, 279–301.
Vinokur, A., Trope, Y., & Burnstein, E. (1975). A decision-making analysis of persuasive
argumentation and the choice-shift effect. Journal of Experimental Social Psychology,
11(2), 127–148.
Waldron, V. R. (1997). Toward a theory of interactive conversational planning. In J. O.
Greene (Ed.), Message production: Advances in communication theory (pp. 195–220).
Mahwah, NJ: Erlbaum.
Waldron, V. R., & Cegala, D. J. (1992). Assessing conversational cognition: Levels of cognitive
theory and associated methodological requirements. Human Communication Research,
18, 599–622.
J. A. Bonito Local Model of Information Sharing
Communication Theory 17 (2007) 252–280 ª 2007 International Communication Association 279
Weiner, S. L., & Goodenough, D. R. (1977). A move toward a psychology of conversation.
In R. O. Freedle (Ed.), Discourse production and comprehension (pp. 213–225). Norwood,
NJ: Ablex.
Wilson, S. R. (1995). Elaborating the cognitive rules model of interaction goals: The problem
of accounting for individual differences in goal formation. Communication Yearbook, 18,
3–25.
Wittenbaum, G. M., & Bowman, J. M. (2004). A social validation explanation for mutual
enhancement. Journal of Experimental Social Psychology, 40(2), 169–184.
Wittenbaum, G. M., Hollingshead, A. B., & Botero, I. C. (2004). From cooperative to
motivated information sharing in groups: Moving beyond the hidden profile paradigm.
Communication Monographs, 71, 286–310.
Wittenbaum, G. M., Hubbell, A. P., & Zuckerman, C. (1999). Mutual enhancement: Toward
an understanding of the collective preference for shared information. Journal of
Personality and Social Psychology, 77, 967–978.
Local Model of Information Sharing J. A. Bonito
280 Communication Theory 17 (2007) 252–280 ª 2007 International Communication Association