Bateman 2011

This article was downloaded by: [192.122.237.
41] On: 26 June 2018, At: 17:59

Publisher: Institute for Operations Research and the Management Sciences (INFORMS)
INFORMS is located in Maryland, USA
Information Systems Research

Publication details, including instructions for authors and subscription information:
http://pubsonline.informs.org
Research Note—The Impact of Community Commitment

on Participation in Online Communities
Patrick J. Bateman, Peter H. Gray, Brian S. Butler,
To cite this article:

Patrick J. Bateman, Peter H. Gray, Brian S. Butler, (2011) Research Note—The Impact of Community Commitment on
Participation in Online Communities. Information Systems Research 22(4):841-854. https://doi.org/10.1287/isre.1090.0265
Full terms and conditions of use: http://pubsonline.informs.org/page/terms-and-conditions
This article may be used only for the purposes of research, teaching, and/or private study. Commercial use
or systematic downloading (by robots or other automatic processes) is prohibited without explicit Publisher
approval, unless otherwise noted. For more information, contact [email protected].
The Publisher does not warrant or guarantee the article’s accuracy, completeness, merchantability, fitness
for a particular purpose, or non-infringement. Descriptions of, or references to, products or publications, or
inclusion of an advertisement in this article, neither constitutes nor implies a guarantee, endorsement, or
support of claims made of that product, publication, or service.
Copyright © 2011, INFORMS
Please scroll down for article—it is on subsequent pages
INFORMS is the largest professional society in the world for professionals in the fields of operations research, management
science, and analytics.
For more information on INFORMS, its publications, membership, or meetings visit http://www.informs.org
Information Systems Research
Vol. 22, No. 4, December 2011, pp. 841–854
issn 1047-7047 eissn 1526-5536 11 2204 0841 http://dx.doi.org/10.1287/isre.1090.0265
© 2011 INFORMS
Research Note
The Impact of Community Commitment on
Downloaded from informs.org by [192.122.237.41] on 26 June 2018, at 17:59 . For personal use only, all rights reserved.
Participation in Online Communities

Patrick J. Bateman
Williamson College of Business Administration, Youngstown State University, Youngstown, Ohio 44555,
[email protected]
Peter H. Gray
McIntire School of Commerce, University of Virginia, Charlottesville, Virginia 22903,
[email protected]
Brian S. Butler
Katz Graduate School of Business, University of Pittsburgh, Pittsburgh, Pennsylvania 15260,
[email protected]
O nline discussion communities have become a widely used medium for interaction, enabling conversations
across a broad range of topics and contexts. Their success, however, depends on participants’ willingness to
invest their time and attention in the absence of formal role and control structures. Why, then, would individuals
choose to return repeatedly to a particular community and engage in the various behaviors that are necessary
to keep conversation within the community going? Some studies of online communities argue that individuals
are driven by self-interest, while others emphasize more altruistic motivations. To get beyond these inconsistent
explanations, we offer a model that brings dissimilar rationales into a single conceptual framework and shows
the validity of each rationale in explaining different online behaviors. Drawing on typologies of organizational
commitment, we argue that members may have psychological bonds to a particular online community based on
(a) need, (b) affect, and/or (c) obligation. We develop hypotheses that explain how each form of commitment
to a community affects the likelihood that a member will engage in particular behaviors (reading threads,
posting replies, moderating the discussion). Our results indicate that each form of community commitment
has a unique impact on each behavior, with need-based commitment predicting thread reading, affect-based
commitment predicting reply posting and moderating behaviors, and obligation-based commitment predicting
only moderating behavior. Researchers seeking to understand how discussion-based communities function will
benefit from this more precise theorizing of how each form of member commitment relates to different kinds
of online behaviors. Community managers who seek to encourage particular behaviors may use our results to
target the underlying form of commitment most likely to encourage the activities they wish to promote.
Key words: online communities; virtual communities; discussion groups; commitment; online behavior;
Web 2.0; social media; social technologies
History: Soon Ang, Senior Editor; Katherine Stewart, Associate Editor. This paper was received on June 18,
2008, and was with the authors 3 41 months for 3 revisions. Published online in Articles in Advance
January 27, 2010.
1. Introduction participated in an online community (Horrigan 2001),

With widespread adoption of the Internet, conversing resulting in an explosion of text-based conversation.
with others who share similar interests has never been While online discussion communities are increas-
easier. Early in the development of the Internet, online ingly pervasive, at their core they remain voluntary
groups—both within and outside of organizations— structures; whether individuals participate—and in
quickly emerged as dynamic conversation spaces what ways—is largely their own choice (Moon and
(Rheingold 1993, Sproull and Kiesler 1991). A plethora Sproull 2008). Online communities are easily found
of browser-based communication tools has acceler- and accessed, and individuals typically have the
ated the growth of new online discussion communi- option to come and go as they please. This might be
ties, each with its own purpose, leaders, members, expected to lead individuals to engage online commu-
structures, resources, and norms. Of the estimated nities as they do many other websites, through one-off
one billion Internet users (e.g., Hof 2005), 84% have transactions driven by general search and retrieval of
841
Bateman, Gray, and Butler: Impact of Community Commitment on Participation in Online Communities
842 Information Systems Research 22(4), pp. 841–854, © 2011 INFORMS
relevant information (Pirolli and Card 1999). Indeed, key differences in how members actually behave. This
research indicates that many visitors to online com- commitment-based approach also helps to address the
munities soon disappear (Arguello et al. 2006). But practical needs of corporate managers and commu-
there is also evidence that online communities are nity developers who care about why members engage
not purely transient collections of casual foragers, and in particular activities in their community, not why
that most survive because some individuals return they participate in online discussions in general (Kim
repeatedly and invest energy in the ongoing conversa- 2000, Preece 2000).
tion (Bagozzi and Dholakia 2002, Lee and Cole 2003).

In a social environment characterized by choice
and low switching costs, why would some individ- 2. Theoretical Background:
uals choose to repeatedly participate in a particu- Commitment Theory
lar discussion community? Furthermore, why would A central focus of research in organizational behav-
some individuals choose to invest additional time ior has been to provide theories of organizational
and energy in behaviors—such as contributing con- membership—that is, examining individuals’ endur-
tent or helping to moderate discussions—that are ing desire to be part of an organization (Mowday
necessary to keep the community’s conversations 1998). One mature body of research has sought to
going? A review of the online community litera- understand how the psychological bonds that arise
ture reveals a range of explanations for why indi- between employees and organizations influence work-
viduals participate in online communities in general. place behaviors (Cohen 2003). Commitment has been
Some explanations highlight the benefits individuals generally described as a psychological bond that
receive from a community, such as access to expert “stabilizes individual behavior under circumstances
advice (Lampel and Bhalla 2007), insights into oth- where the individual would otherwise be tempted to
ers’ beliefs and opinions (Herring 1996), enhanced change that behavior” (Brickman 1987, p. 2). More
reputation (Lakhani and von Hippel 2003), addi- specifically, organizational commitment is a psycho-
tional professional contacts (Christensen and Raynor logical bond that characterizes an individual’s rela-
2003), improved professional status (Hall and Graham tionship with an organization (Meyer and Allen 1991,
2004, Lerner and Tirole 2002), a more positive self- p. 67). Early on, commitment researchers typically
image (Constant et al. 1994), and greater confi- pursued different conceptualizations of psychologi-
dence in their own knowledge (Wasko and Faraj cal attachment, including employees’ sense of depen-
2000). Other explanations focus more on helping oth- dence on the organization, their feelings of attachment
ers, such as a desire to help build a community to the organization, and their sense of obligation to the
(e.g., Blanchard and Markus 2004), contributing to organization (e.g., Mathieu and Zajac 1990). In their
collective goals (Constant et al. 1994, 1996), ensur- seminal work, Meyer and Allen (1991) converged on
ing the continued existence of the community (Wasko the now widely accepted three-part conceptualization,
and Faraj 2000), feelings of camaraderie (Hall and theorizing that organizational commitment is a multi-
Graham 2004), reciprocity (Constant et al. 1994, Wasko dimensional construct consisting of three components:
and Faraj 2005), altruism (Lakhani and von Hippel continuance, affective, and normative commitment.
2003), and even empathy (Preece 1999, Preece and In the organizational context, these three types of
Ghozati 1998). These kinds of motivators are often commitment have been broadly accepted as power-
offered up as competing explanations for why indi- ful predictors of job-related outcomes (Meyer and
viduals participate in online communities in general Herscovitch 2001, Meyer et al. 2002). However, each
(e.g., Wasko and Faraj 2000). But because an individ- has a unique pattern of associations with differ-
ual may behave very differently in different commu- ent employee attitudes and behaviors, including
nities (Jones et al. 2004, Joyce and Kraut 2006), we job performance (Somers and Birnbaum 1998), cit-
seek to answer these questions by using a theoretical izenship behaviors (Organ and Ryan 1995), and
approach that considers both the individual and the absenteeism (Somers 1995). Continuance commitment
community. Our approach draws on organizational leads employees to focus their efforts on preserving
commitment research (Meyer and Allen 1997), which the employment relationship for their own benefit
has sought to provide locally situated explanations for (Becker 1960); they do what they must to main-
why individuals engage in certain behaviors based on tain their position with the organization, and little
the psychological bonds they develop to their orga- else, because they need something that is a result of
nizations. We hypothesize that analogous forms of their involvement (i.e., status, influence, compensa-
community commitment differentially affect specific tion). For example, in a recent meta-analysis (Meyer
types of member behaviors in an online community, et al. 2002), continuance commitment was found to
thereby advancing the literature beyond merely pre- be the only form of commitment that was neg-
dicting whether individuals participate by explaining atively correlated with perceptions of interactional
Information Systems Research 22(4), pp. 841–854, © 2011 INFORMS 843
justice and supervisor support, and the only one to organizations; continuance commitment leads indi-
negatively predict job performance. Affective com- viduals to remain with their organization because
mitment, on the other hand, is based on an emo- they feel a need for the unique benefits which they
tional attachment to the organization, which leads personally derive from that relationship (Meyer and
employees to act in ways that further the organiza- Allen 1991). Sometimes termed “calculative” commit-
tion’s interests (Meyer and Herscovitch 2001). The ment (Swailes 2002), this bond between individual
same meta-analysis noted above shows that affec- and organization is driven by doubts that an alter-
tive commitment was the only form that was posi- nate organization could provide the same level of ben-
tively correlated with supervisor satisfaction, and it efits for the same effort invested (Meyer and Allen
uniquely predicted both absenteeism and job perfor- 1997). Continuance commitment is therefore a con-
mance. Finally, employees who have a strong nor- cept that is broadly applicable because, even in com-
mative commitment feel an obligation to contribute mitment theory, it does not depend on the nature of
to the goals and mission of the organization, which the organization, but rather on the extent to which
makes them more likely to engage in discretionary, an individual believes that he or she derives benefits
nonmandated behaviors, such as organizational cit-
from the relationship that are not available from other
izenship behaviors, because they feel that it is “the
sources (Whitener and Walz 1993). In a community
right thing to do” (Wiener 1982). Although each form
context, members invest time and energy and report
of commitment has a different theoretical rationale,
receiving a range of informational and social bene-
they are not mutually exclusive; indeed, employees
can simultaneously possess different levels of each fits (Lakhani and von Hippel 2003, Ridings and Gefen
type (Meyer et al. 2002). 2004). Community members who value those bene-
Though commonly applied to employment rela- fits and doubt that they could obtain the same bene-
tionships in the management literature, commitment fits elsewhere while incurring acceptable membership
research originally sought to explain why volunteers costs are thus likely to continue participating in ways
at nonprofit organizations varied in their level of that maximize those benefits (Butler 2001). We term
dedication (Becker 1960), making it a particularly this corresponding construct Continuance Community
appropriate theory base for understanding individu- Commitment (Continuance CC) and define it as a bond
als’ voluntary behavior in online communities. Com- between a member and a particular community that is
munity members are not employees, but both have based on the member’s belief that his or her involve-
considerable discretion in their level of engagement ment provides net benefits that are not easily available
and the specific behaviors they choose to empha- elsewhere.
size; an individual’s commitment to a community Affective commitment has been defined as “the
might therefore be expected to affect his or her employee’s emotional attachment to, identification
online behaviors (Wasko and Faraj 2005). However, with, and involvement in the organization” (Meyer
while employee-employer relationships and member- and Allen 1991, p. 67). Individuals with high levels
community relationships are both fundamentally of affective commitment like their organization and
volitional in nature, the differences between them find their association with it to be emotionally ful-
(e.g., de Souza and Preece 2004, Herrmann et al. filling, independent of whether they necessarily like
2004) suggest that each type of commitment will have the particular activities that constitute their jobs. They
analogous, but not identical, effects in online com- tend to identify strongly with the employing firm
munities. Thus, while the basic logic of commitment (Rousseau 1998), share its goals and values (Griffeth
theory—that an individual’s particular mode of com- et al. 2000), and feel as though they belong in the firm
mitment will have unique effects in shaping his or (Porter et al. 1974). Affective commitment may arise
her engagement and behavior—is expected to be con- from social exchanges that lead employees to trust
sistent with the nature of online communities, the their employers (Cook and Wall 1980) and feel fairly
specific connections between constructs must be re-
treated by them (Riketta 2002). Similar affective bonds
theorized for the online community context.
may also form between a member and an online
2.1. Types of Community Commitment community; evidence suggests that some community
Continuance commitment has been defined as “an members may develop such feelings of attachment
awareness of the costs associated with leaving an and identification with a community (Blanchard and
organization” (Meyer and Allen 1991, p. 67) and as Markus 2004) and may in turn come to feel a sense
“the degree to which an individual experiences a of belonging (Markus et al. 2000). Affective Commu-
sense of being locked in place because of the high nity Commitment (Affective CC), therefore, is a bond
costs of leaving” (Jaros et al. 1993, p. 954). Here, between a member and a particular community that is
costs include the loss of social and economic bene- based on the member’s strong emotional attachment
fits that individuals believe are not available at other to that community.
Normative commitment is a third distinct form of Cothrel 2000), it becomes clear that a single com-
organizational commitment that has been defined as mon set of antecedents is unlikely to have identi-
“a feeling of obligation to continue employment” cal impacts across these three very different kinds of
(Meyer and Allen 1991, p. 67) with an organiza- behaviors (Koh et al. 2007). In the following sections
tion. Employees with a high level of normative com- we hypothesize how each of the three kinds of bonds
mitment feel that they ought to remain with an that may form between an individual and a commu-
organization, independent of whether or not they nity impacts a different behavior within that com-
benefit directly from their activities there (Meyer and munity. We do so by situating a range of motivators
Allen 1991). Normative commitment may arise from discussed in the online communities literature in the
employees internalizing the sense of loyalty (Wiener commitment framework described above.
1982) and obligation (Ashforth et al. 1998) held by
their colleagues, or when employees experience a 3.1. Continuance Community Commitment
sense of indebtedness because they believe they have A range of studies have noted that behaviors in online
received benefits from their employer that they are communities are driven by members’ desire to max-
unable to adequately reciprocate (Gouldner 1960). In imize the value they obtain from a community (e.g.,
a community context, members who have benefited Furlong 1989, Ridings and Gefen 2004). This kind of
motivation revolves around a member’s own needs
from others’ friendship and advice may feel indebted
and goals, and the effort required to obtain valuable
to the community, feel a duty to remain members to
information (Jones 1995). To obtain informational ben-
repay the perceived debt (Lakhani and von Hippel
efits from an online community, members read dis-
2003), and as a result have a sense of normative obli-
cussion threads (Welser et al. 2007). But reading takes
gation towards the community (Rheingold 1993). Nor-
time: adults typically read 200 words per minute (e.g.,
mative Community Commitment (Normative CC) is thus
Saubramanian and Pardhan 2006). Add to this the
a bond between a member and a particular commu-
extra requirement of navigation and search (Zhang
nity that is based on the member’s sense of obligation
and Watts 2008), and it becomes clear that reading
towards that community.
threads is not effortless. Though there may be many
explanations for why members are motivated to do
so, one primary determinant is likely to be a mem-
3. Theory Development ber’s belief that the benefits of reading outweigh the
By drawing on commitment theory, the range of com-
costs (Butler et al. 2007). Continuance CC allows for
peting explanations for participation offered in prior
the examination of such beliefs, by indicating the
work can be understood as pieces of a more coherent
degree to which a member believes the cost/benefit
framework that helps explain why individuals engage ratio associated with one community is superior to
in different kinds of conversational behaviors online. that of other communities.
To produce such a finer-grained model, we consider A member’s level of continuance CC is likely to
three key community behaviors identified in the lit- affect his or her thread-reading behavior in several
erature: reading threads, posting replies, and moder- ways. First, a strong expectation that content will
ating discussions (e.g., Constant et al. 1996, Lakhani be valuable is likely to lead a member to be espe-
and von Hippel 2003, Wasko and Faraj 2000). Because cially persistent in reading threads (e.g., Hsiu-Fen and
it entails consuming content posted by others, reading Gwo-Guang 2006), even if he or she encounters low-
threads is a primary mechanism by which individu- quality content from time to time. While a member
als obtain the direct informational and social benefits who has a low continuance CC might be discour-
available from a community (Butler 2001, Welser et al. aged when he or she encounters content that does
2007). Individuals who post replies are contributing not directly match his or her interests and expec-
new information resources that help others (Lakhani tations, a lack of fit is unlikely to deter a member
and von Hippel 2003) and may also improve the who has a high level of continuance CC (at least
posting individual’s status and reputation (Oreg and in the short run). Such persistence may occur irre-
Nov 2008). Those who engage in informal moderating spective of actual content quality, as cognitive dis-
behaviors work to actively maintain and promote con- sonance (Festinger 1957) would cause such a mem-
versations in the community (e.g., Ahuja and Galvin ber to behave as though the community contained
2003, Butler et al. 2007, Lampe and Resnick 2004) valuable, unique content—whether it actually did or
by guiding discussions towards collectively valued not. This effect would not be expected in a mem-
themes (Kollock and Smith 1996), managing disputes ber with a low level of continuance CC, who would
between users (Butler et al. 2007), and discouraging be more likely to discontinue reading threads when
off-topic posts (Burnett and Bonnici 2003). Once par- faced with content that did not match his or her inter-
ticipation is considered more precisely (Williams and ests and needs (Butler 2001). Second, a member with
high continuance CC may believe that the social costs feel no particular bond or emotional connection with
of leaving a community would be very high, partic- a community, are less invested in its long-term viabil-
ularly if he or she has invested considerable effort ity, and are thus less likely to care enough to reply to
into the community in establishing an identity and others’ postings or explicitly be part of the commu-
learning the specific ways the community functions nity conversations.
(Ma and Agarwal 2007). In conjunction with a per-
Hypothesis 2 (H2). A member’s level of affective CC
ception of greater benefits, the costs associated with
towards a community will positively influence the number
having to recreate an identity and learn the ways of a
new community would lead such a member to spend of replies he or she posts in that community.
more of their time consuming content in the com- 3.3. Normative Community Commitment
munity in which they have established themselves
Some members of an online community feel a sense
(versus another, new community). Together, these two
of indebtedness or loyalty towards the community
theoretical paths (heightened persistence and sunk
(Ridings et al. 2006), which may appear to others as
costs) suggest that members with high continuance
altruism (Wasko and Faraj 2000) or empathy (Preece
CC will read more threads than those with low con-
1999). Such feelings may arise when a member has
tinuance CC.
benefitted from his or her association with a com-
Hypothesis 1 (H1). A member’s level of continuance munity and feels an obligation to repay the commu-
CC towards a community will positively influence the nity, or because he or she believes strongly in the
number of threads he or she reads in that community. community. Underlying each of these motivations is
an implicit normative argument, as some members
3.2. Affective Community Commitment
believe that it is their duty to support the commu-
Online community research provides a range of evi-
nity (Constant et al. 1994, Wasko and Faraj 2005). Such
dence to support the idea that individuals may form
members are likely to behave differently than those
strong emotional attachments to a community (e.g.,
who lack the same sense of loyalty and/or indebted-
Greer 2000), with some members even reporting that
ness (Rheingold 1993). In particular, a member who
they love their online community (Preece 1999). Mem-
feels bound to remain in a community because it is
bers who have developed a strong affective bond
“the right thing to do” is more likely to act in selfless
towards an online community generally like that com-
munity and identify with it (Blanchard and Markus ways to sustain and even strengthen the community
2004), and are therefore more likely to want to be (Hall and Graham 2004). This sense of obligation may
part of the conversations that occur in that commu- lead such a member to engage in behaviors that help
nity (Preece 1999). Because they find their associa- the community, even if doing so incurs costs to the
tion with it to be emotionally fulfilling, individuals member that are not offset by any other direct benefits
with high levels of affective CC are likely to care very (Oreg and Nov 2008).
much about the community and how it is growing Members who have a strong normative commit-
and evolving. ment to a community are more likely to moderate dis-
Individuals are generally more inclined to help cussions in that community (Kim 2000)—that is, to try
those who are part of a group that they like and to control negative behaviors and guide online dis-
care about (Grant 2007), and so members who feel cussions towards positive ends. Conceptually, infor-
a strong emotional attachment to a community are mal moderating behaviors are related to the idea of
more likely to help other members by replying to community justice, which refers to the processes by
their posts. An affective attachment helps counteract which members of geographically bound communi-
cognitive processes that help humans conserve atten- ties take responsibility for self-policing and respond-
tion, but which can limit individuals’ willingness to ing to crime via social control mechanisms that
help strangers (Noddings 1984). Because of their sense enhance community life (e.g., Sampson 1995). Com-
of attachment and belonging, members who have a munity justice builds on the recognition that when cit-
strong affective CC towards an online community are izens engage in civic activism, they discourage crime
more apt to invest their time and effort helping oth- and steer would-be criminals towards socially appro-
ers in the community by responding to their ques- priate behaviors. Proponents of community justice
tions (e.g., Fisher et al. 2006, Wellman and Gulia 1999). are more likely to appear among citizens who feel a
Furthermore, members who identify strongly with a strong sense of loyalty and social obligation towards
community care more about its central conversational their community (Clear and Karp 2000). Their sense
topic or theme, and have a stronger desire to pub- of normative commitment leads them to take per-
licly demonstrate their solidarity with the community sonal responsibility for preserving the viability of
(Blanchard and Markus 2004, Ren et al. 2007) by con- their physical community.
tributing to the conversations that evolve in the com- Similar processes are likely to occur in online com-
munity. Conversely, members with low affective CC munities (Kang et al. 2007). Members who have high
levels of normative CC are more likely to engage in controlled for among independent variables, depen-
leadership behaviors (Blanchard and Markus 2004), dent variables, and other control variables described
because they feel an obligation to do the right thing below.
for their community. Believing that they have ben-
efitted from their membership in the community, a 4. Research Methods
form of generalized social exchange (Ekeh 1974) pro- Data for testing our research model were collected
duces a sense of obligation that leads members to act at BroadForum (a pseudonym), an online discussion
in ways that repay the community as a whole. As a community that had approximately 50,000 registered
result, members with high normative CC are more members and 3.4 million posts during the prior 22
likely to engage in behaviors that protect and enhance months of operation. BroadForum’s business model
the community. By promoting constructive behaviors was based on revenue generated via paid advertising
and discouraging disruptive ones, they help to main- targeted at members. Built on a commercial thread-
tain a sense of cohesion by defining what is acceptable based bulletin board platform (vBulletin), it was pro-
behavior and what is not (Bergquist and Ljungberg moted as a “general discussion” community, with
2001, Burnett and Bonnici 2003). Over time, informal topics including current events, sports, entertainment,
moderating serves to both winnow out unproductive fashion, politics, philosophy, technology, and anime,
members and retain productive ones, socialize new- among many others. Members did not occupy any
comers, and strengthen the community (Ahuja and particular professional, technical, or cultural niche. As
Galvin 2003). Because they feel a bond of loyalty and is common practice in the administration of online
obligation to the community, members who have high communities, the sheer size of the community and
levels of normative CC are more likely to engage in message volume led its manager to group discus-
behaviors that preserve the community. sion threads into categories to help members organize
their conversations around topics of interest. How-
Hypothesis 3 (H3). A member’s level of normative ever, BroadForum maintained its identity as a single
CC towards a community will positively influence the online community, hosted by a single manager. We
extent to which he or she engages in discussion moderating solicited subjects via an invitation message made in a
in that community. new thread (as recommended by Andrews et al. 2003)
that included a description of the project, an endorse-
Though we hypothesize three relationships, one for
ment by the manager, and an invitation to complete
each form of commitment and its distinctive impact
an online survey in exchange for the chance to win a
on a particular community behavior, it is possible that gift certificate from a popular online retailer. Follow-
each form of community commitment could impact up postings were made on days 7 and 11, and data
other online behaviors. To control for the other pos- collection terminated on day 14.
sible relationships not hypothesized here, we include
these other potential associations as controls in our 4.1. Respondents
model. Figure 1 therefore shows the three hypothe- Our invitation was viewed 3,183 times over 14 days,
sized relationships alongside all other paths that we although this tally does not distinguish between
Figure 1 Research Model
Continuance Reading
H1 Age
CC threads
Affective Posting
H2 Gender
CC replies
Normative Moderating
H3 Tenure
CC discussions
Note. Controls shown with dashed lines.

unique and repeat views. A total of 741 members Table 1 Survey Items
accessed the survey site, meaning that at least 23.3%
Continuance community commitment
of those who viewed the invitation clicked through
CC1a I am sure that there are other sites where I could find the
to the survey. Of these click-throughs, 324 (43.7%) same content and services that I get at this site. [r]
subsequently went through the entire survey. A con- CC2a I keep coming to this site because there are few alternative
servative estimate1 is thus that at least 10.1% of sites available.
potential respondents completed the survey (although CC3 If I stopped coming to this site, it would take me a long time
to find a site that could replace it.
this figure would increase if the invitation was viewed CC4 There are very few other places where I could find the kind of
multiple times by the same individual). We deleted useful content and services that I get from this site.
responses made by those under 18 years old and those CC5 The content of this site is too valuable for me to stop visiting.
with large amounts of missing data, resulting in a Normative community commitment
final data set of 192 adult members (ages ranged from NC1 I feel an obligation to continue visiting this site.
18 to 53, with a mean of 23.5). Tenure as a registered NC2 I would feel guilty if I stopped visiting the site now.
member ranged from six to 671 days, with a mean NC3a This site deserves my loyalty.
NC4 I keep coming to visit this site because I have a sense of
of 360 days and median of 390 days. Eighty-four per- obligation to it.
cent of respondents were male. The largest proportion NC5 I visit this site partly out of a sense of duty.
of respondents reported spending between two and Affective community commitment
four hours daily using the Internet and visiting online AC1 I feel like a part of the group at this site.
communities. AC2 I have a real emotional attachment to this site.
AC3 This site has a great deal of personal meaning for me.
4.2. Measures AC4 I feel a strong sense of belonging to this site.
AC5 I feel a strong connection to this site.
We developed our survey instrument following
Moderating discussions
Dillman’s (2000) approach. We adapted items from
MO1 I try to settle disputes between users.
Meyer and Allen’s (1997) organizational commitment MO2 I encourage users not to post messages that are off-topic
scales to create community commitment equivalents, (i.e., hijack) from the original thread.
and items from Butler et al. (2007) to measure moder- MO3 I reprimand other users’ inappropriate behavior.
ating behaviors. Items that measured continuance CC Gender What is your gender?
were analogous to organizational continuance com- Age What is your year of birth?
mitment measures, focusing on key ideas that had a
Items dropped from final analysis.
meaning in both contexts, including individuals’ per-
ceptions of loss on departure, availability of viable
alternatives, scarcity of similar benefits elsewhere,
All attitudinal items were measured on Likert scales
and perceptions of unique value. Items adapted to
measure the community-oriented normative analogue anchored on “1 = strongly disagree” and “7 = strongly
focused on the idea of loyalty, obligation, guilt, agree.”
and duty. Finally, items developed to measure affec- We gathered archival data from BroadForum’s
tive community commitment used ideas of perceived server logs for 16 weeks after the completion of the
group membership, emotional attachment, personal survey: the number of threads read and the number
meaningfulness, and a sense of belonging, all present of replies posted over this period. The archival mea-
in the source scales. In each case, our goal was to cap- sure of threads read is the summation of the num-
ture the original theoretical concept as it would be ber of threads displayed on each member’s monitor
expressed in an online community context. All items over those 16 weeks, while the measure of replies
were further refined via a card sort procedure and a posted is the summation of the number of times a
pilot study performed in three different online com- member responded to another’s post over the same
munities, with a total of 285 completed responses, and 16-week period. While not all members were active
informal discussions with community managers and during each of those 16 weeks, summing responses
members. The final set of items is shown in Table 1. over this period produces variables that reflect longer-
term tendencies, rather than being confounded by
1
Following an established procedure for calculating response rates short-term bursts of activity. These archival data were
in surveys of online communities (Ridings et al. 2002) produced
a response rate of 25.9% of individuals who accessed the survey skewed, as is typical in community behavior data
site. While we know nothing about those who read the invitation (Butler 2001, Jones et al. 2004); we therefore calculated
but chose not to access the survey site, our ability to track how the logarithm of each archival variable to increase
many times invitations were viewed and how many surveys were
the normality of the data by collapsing the distance
started and completed allowed us to calculate a reliable response
rate, thereby providing better metrics than are typical in online data between the values in the long tail of the distribution
collection (Andrews et al. 2003). (a long tail is common in studies of online community
Table 2 Means, Standard Deviations, and Correlations 4n = 1935
Response Composite
mean Std. dev. reliability AVE 1 2 3 4 5 6 7 8 9
1 Gender 0084 n.a. n.a. n.a. n.a.

2 Age 2305 704 n.a. n.a. −0039∗∗ n.a.
3 Tenure (days) 360 209 n.a. n.a. 0033∗∗ −0017∗ n.a.
b
4 Continuance CC 5013 1042 0089 0073 −0008 0004 0002 0085
5 Normative CC 3050 1064 0093 0078 0001 −0001 0009 0030∗∗ 0088
6 Affective CC 4070 1040 0094 0076 0001 −0003 0013 0051∗∗ 0052∗∗ 0087
7 Reading threadsa 1,694 1,708 n.a. n.a. 0040∗∗ −0027∗∗ 0034∗∗ 0017∗ 0004 0017∗ n.a.
8 Posting repliesa 471 1,570 n.a. n.a. 0030∗∗ −0021∗ 0015 0018∗ 0009 0024∗∗ 0081∗∗ n.a.
9 Moderating discussions 4042 1043 0088 0071 −0012 0004 −0004 0024∗∗ 0035∗∗ 0047∗∗ 0014 0022∗ 0084
a
Means and standard deviations for these variables are reported for raw variables; correlations are reported for logged versions of archival variables.
b
Diagonal elements are the square root of average variance extracted.
∗
p < 0005; ∗∗ p < 0001 (two-tailed tests).
phenomena). We also gathered demographic data— 4.3. Data Analysis and Results
gender (1 = male, 0 = female), age (years), and tenure We tested our hypotheses by examining the size and
(days)—and included these variables as controls in significance2 of structural paths in the PLS analysis,
our models. with all significant paths shown in Figure 2.
The data were employed in a simultaneous test The analysis revealed that only continuance CC
of structural and measurement models using Partial significantly predicted thread-reading behavior (H1,
Least Squares (PLS Graph, version 3.00). In addition = 0018, p < 0001). The model also shows that affec-
to modeling the hypothesized relationships and con- tive CC was the only form of commitment that
trolling for other nonhypothesized relationships and significantly predicted reply-posting behavior (H2,
variables shown in Figure 1, we also controlled for = 0011, p < 0005). Finally, the PLS analysis indi-
cates that normative CC (H3, = 0016, p < 0005)
the possibility that replying and moderating behav-
significantly predicted discussion-moderating behav-
iors would be a direct function of reading—that is,
ior, and that (though not hypothesized) affective CC
individuals who read more threads might reply or
( = 0033, p < 0001) also significantly impacted moder-
moderate more simply because of their larger volume
ating behavior. Other control variables were also sig-
of reading. By controlling for this effect, we were able nificantly associated with different online behaviors.
to more precisely examine the proportion of variance Thread reading significantly predicted reply-posting
in replying or moderating that was a function of com- ( = 0084, p < 0001) and discussion-moderating behav-
munity commitments. We assessed the adequacy of ior ( = 0017, p < 0005). Gender significantly predicted
the measurement model using three common tests of threads read ( = 0030, p < 0001) and negatively pre-
convergent validity (Chin 1998). First, after dropping dicted discussion-moderating behavior ( = −0014,
three items that loaded poorly (CC1, CC2, and NC3), p < 0005). Tenure significantly predicted thread read-
all loadings of the remaining items on their intended ing ( = 0021, p < 0001), and negatively predicted reply
constructs were greater than 0.7. Second, we assessed posting ( = −0016, p < 0001). Age did not significantly
the internal consistency of each construct using com- influence any behavior. Beta coefficients for the full set
posite reliability and found the lowest to be 0.88. of paths tested in our model, including nonsignificant
Third, we calculated the average variance extracted relationships, are provided in Table 3.
for each scale; all scales exceeded Chin’s (1998) guide-
line of 0.5, meaning that at least 50% of the variance
in indicators was accounted for by its respective con-
5. Discussion and Implications
The online communities literature features a range
struct. Table 2 provides the results of these measure-
of sometimes contradictory claims about different
ment model analyses. To assess discriminant validity,
factors that are thought to motivate online partici-
we also conducted an exploratory factor analysis and pation in general (e.g., Blanchard and Markus 2004,
examined the correlations of items and constructs and Lakhani and von Hippel 2003, Wasko and Faraj 2000).
found that none of the cross-loadings exceeded 0.35.
We also noted that the square root of AVE for each
2
PLS produces standardized regression coefficients for structural
construct (see Table 2) exceeded all respective inter-
paths. Bootstrapping techniques, a nonparametric approach for
construct correlations, providing further evidence of estimating the precision of paths, were used to test for significance
discriminant validity. using 500 resamples.
Figure 2 Results of PLS Analysis
Reading
Continuance threads
0.18** Age
CC R2 = 27%
0.84** 0.30**
0.17*
Affective Posting
0.12** replies 0.21** Gender
CC
R2 = 69%
0.33**
– 0.16**
– 0.14*
Normative Moderating
0.16* discussions Tenure
CC
R2 = 29%
Note. Significant paths only: p < 0005 (*) and p < 0001 (**).
By adopting a commitment framework, we demon- commitment in an organizational setting sometimes

strate the utility of conceptualizing these in a more leads individuals to engage in fewer noninstrumen-
systematic fashion as member-community bonds that tal behaviors, our data did not support such an effect
predict specific discussion behaviors. Our results con- in a community setting. The fact that continuance CC
firm that each form of community commitment has did not affect replies or informal moderation provides
unique explanatory power and is not interchangeable evidence that only those online behaviors that pro-
with others in nature, scope, or impact. This research vide the most direct benefits are motivated by mem-
may help community managers to better target their bers’ belief that they rely on the community for the
member development and retention efforts towards unique value it creates. Although organizational con-
producing the kinds of bonds that will keep the com- tinuance commitment is often seen as undesirable
munity active, focused, and evolving. Below, we elab- (Meyer and Allen 1997), our results suggest that con-
orate on these contributions to research and practice. tinuance CC is not a bad thing. Despite the fact that
members who only consume content are sometimes
5.1. Implications for Research characterized as free riders (Kollock and Smith 1996),
This study advances online community research by continuance CC seems to produce members who try
offering a coherent model of member-community harder to maximize the unique benefits of the com-
commitments and their differential behavioral effects. munity for themselves personally and thus form an
Consistent with commitment theory, all three kinds audience that may attract content providers, which is
of member-community bonds can operate simultane- key to community growth (Butler 2001). Differences in
ously in a community setting and produce different how benefits are allocated may explain these different
kinds of behavioral outcomes. This provides a new effects: In an organizational context, status, influence,
baseline on which future research can build more and income are often relatively insensitive to variation
powerful and precise theoretical models of engage- in employees’ effort level (at least in the short run),
ment in online communities. while online communities produce immediate bene-
First, the connection between continuance CC and fits in direct proportion to effort invested. Organiza-
thread-reading behavior confirms the very focused tional commitment researchers therefore may seek to
outcomes of members’ sense of instrumental depen- build on these findings by investigating whether com-
dence on an online community. While continuance pensation style or the ability to adapt jobs to individ-
ual needs significantly moderates the negative effects
Table 3 All Path Coefficients of continuance commitment in the workplace.
Continuance Affective Normative Reading Second, the impact of affective CC on replies posted
CC CC CC threads Age Gender Tenure attests to the importance of emotional attachment and
Reading threads 0018 ∗∗
0008 −0008 −0012 0030∗∗
0021∗∗
identification in online communities. Members’ affec-
Posting replies −0003 0012∗∗ 0002 0084∗∗ 0000 0001 −0016∗∗ tive CC seems to lead them to want to help others
Moderating 0009 0033∗∗ 0016∗ 0017∗ 0002 −0014∗ −0010 who are part of their community by engaging in con-
discussions
versation with them. The development of affective CC
∗
p < 0005; ∗∗ p < 0001 (two-tailed tests). in members may therefore be an important step that
not only engages members in a community’s activities important future challenges in reconciling the differ-
more deeply, but also helps ensure the long-term suc- ent foci of obligations that may exist in online com-
cess of the community by making it more likely that munities. In particular, the strong impact of affective
questions will receive responses (Arguello et al. 2006). CC on moderating behaviors suggests that emotional
One unexpected finding was the powerful effect of attachment may account for at least some of what
affective CC on informal moderating behaviors. A other studies have attributed to obligation.
potential explanation for this may be that whatever The integrative theoretical approach developed in
the specific characteristics that produce an affective this work also provides guidance for scholars inter-
bond—and these are likely to vary from group to ested in modeling online communities. Prior stud-
group—members who have strong affective bonds ies have tended to treat theories of need, affect, and
to a community may be more likely to engage in obligation as competing explanations for behaviors
behaviors that contribute to maintaining those char- in online communities. For example, an exploratory
acteristics over time (Ren et al. 2007). Because these study by Wasko and Faraj (2000) asked members of
characteristics are often perpetuated through social an online community why they participated in it, and
structures, members who seek to sustain the desir- respondents gave a wide range of reasons (includ-
able aspects of a community are likely to voluntar- ing anticipated benefits, enjoyment, and moral obli-
ily engage in behaviors that reinforce such commu- gation, among others). However, Wasko and Faraj
nity goals, values, and social structure (Kim 2000). saw these as alternate explanations for participat-
As such, the emotional connection that leads them to ing in online communities generally, and, based on
identify with the community and hold its values and the strength of the responses they gathered, con-
goals as important may lead them to be more will- cluded that a variety of individually held beliefs
ing to act in relatively selfless ways to help sustain about moral obligation were the most important fac-
those values by acting as an informal group mod- tors in explaining online behaviors. Our study takes
a key step forward by advancing a comprehensive
erator. However, our results show that affective CC
set of member-community bonds based on a well-
did not lead members to read more threads—possibly
established parallel literature, theorizing the unique
because of a plateau effect or because members who
impacts of these bonds, and documenting the spe-
have a strong affective CC express their bond by
cific kinds of bonds that lead members to perform
investing more effort into the conversations that they
specific behaviors. In a subsequent study, Wasko
are already part of, rather than seeking out new con-
and Faraj (2005) posited that individuals who were
versations. Overall, members who have a strong affec-
more committed to a professional network of prac-
tive attachment to a community engage in a broader
tice would contribute more knowledge. However,
range of community-focused behaviors than origi-
their operationalization of commitment attempted to
nally anticipated as a result of their identification and combine aspects of affective, normative, and con-
attachment with the community. tinuance logic into one measure, which ultimately
Third, while the impact of normative CC on mod- failed to support their hypotheses. By fully elabo-
erating behaviors was supported, its impact was rating the community versions of each commitment
weaker than the literature on reciprocity and altruism construct, our research provides a plausible explana-
in online contexts sometimes implies (e.g., Constant tion for Wasko and Faraj’s lack of significant find-
et al. 1994), as normative CC was a weaker predic- ings. Rather than finding that commitment has no
tor of moderating behaviors than was affective CC. effect on members’ behavior, our more precise model
Furthermore, normative CC did not significantly pre- in fact reveals that it has a rich array of effects.
dict thread-reading or reply-posting behaviors. Taken Researchers who might have discarded the commit-
together, this calls into question arguments such as ment construct as unimportant should reconsider this
those offered by Preece (1999) and Wasko and Faraj important set of ideas when developing models of
(2000) that generalized that reciprocity and a sense of online behaviors and their antecedents. Our results
obligation are primary drivers of online community suggest that any attempt to explain online behaviors
behaviors. Methodological variation may be partly to will be incomplete unless it considers all three forms
blame for these differences: Studies often differ with of commitment.
respect to the entity that is the target of an indi- Future research that examines the synergistic
vidual’s sense of obligation (another individual ver- impacts of different levels of each kind of commit-
sus a specific group or community versus generalized ment in a community (for instance, via simulation
altruism or reciprocity without a focus). Although our techniques) could go far in illuminating the dynamic
results provide some support for the presence of a process of community formation. Because we were
generalized sense of indebtedness and obligation that not able to observe how different forms of commu-
affects informal moderator behaviors, it also suggests nity commitment developed over time, a variety of
interesting questions remain about their nature and 5.2. Implications for Community Managers
evolution. First, is there a typical temporal sequence Our results highlight a complex problem faced by
in the evolution of a member’s community commit- community managers: how to encourage members
ments? For instance, does an initial period of con- to engage in the various behaviors that are key to
tinuance CC, with its individually focused style, act community viability. Building affective CC is clearly
as a mechanism to keep new members coming back, important, but such relationships may be more dif-
possibly leading to the development of greater affec- ficult to develop, especially on a large scale. Mem-
tive CC and its associated helping behaviors later?
bers primarily motivated by continuance CC are also

If so, how does normative CC fit into the evolu- important, because they serve as an audience and
tion of member commitments? Second, how do inter- are probably easier to attract. However, they are less
actions between groups of people characterized by likely to give back to the community, as their behav-
these different kinds of commitments contribute to iors are driven by their own needs. Their value might
the operation of a community as a functioning whole? not be immediately apparent, but managers who can
In particular, there seem to be parallels between the appreciate them as part of the diversity necessary to
commitment/behavior patterns observed in our find- keep a healthy community growing are likely to do
ings and the core-periphery structure that is often a better job of ensuring that the community does not
discussed in the context of online communities (e.g., become too introspective and exclusionary. Instead of
Kim 2000). There are clearly some intriguing possibil- trying to understand the “ideal” or “average” com-
ities for building new connections between social net- munity member—as much past research has done—
work and social psychological approaches to online our results suggest that managers should appreciate
behavior, where individual-level insights may com- the range of psychological bonds that members may
plement network-level findings to produce a more experience. It may not be necessary to encourage each
integrated, multilevel theory of community activity member to engage in all the major behaviors that
and dynamics. keep a community functioning; instead, managers
Control variables in our model also suggest some
may be able to stimulate desired behaviors by nur-
interesting avenues for future research. First, in our
turing the associated underlying psychological bonds.
sample, men were much more likely to read threads,
Although we did not investigate community commit-
while women were more likely to moderate discus-
ment antecedents, we describe some potentially valu-
sions. This stands in contrast to research that finds,
able interventions below.
for example, that men and women are equally likely
Managers who seek to expand their audience could
to both lurk and post (Preece 2004), but is consis-
enhance members’ perceptions of the unique value of
tent with other research demonstrating that men and
the community. This could begin with a survey or
women communicate differently online (Gefen and
Ridings 2005). Second, our results suggest that mem- focus group, or with an analysis of topic popularity to
bers with longer tenure are more likely to read but identify the kind of content that is uniquely valuable
less likely to reply to threads. Claims that longer- to the member base; doing so might help managers
tenure members are more likely to take on leader- reposition a community towards the types of valu-
ship roles (Kim 2000) may be confusing tenure with able content that are not available elsewhere. Man-
commitment; our findings suggest that it is the form agers might then create new discussion areas, solicit
of attachment that counts, not merely the length of targeted contributions to stimulate conversations, or
tenure. seed controversial discussions to encourage postings
Looking forward, a natural extension to this study on high-value topics. Managers could also commu-
would be to examine the factors that influence how nicate with members to ensure that all are aware of
new members, as they are socialized into an online the unique benefits available, for instance by post-
community, form various kinds of commitments. ing short examples of high-value content not avail-
The organizational commitment literature suggests a able elsewhere, or by retelling stories of members
broad range of antecedents to each form of commit- who benefitted greatly from community content. Such
ment, some of which may also apply in community efforts are likely to increase continuance CC and pos-
settings (e.g., shared values, trust, supportiveness). itively influence thread-reading behavior.
It is also possible that certain desires and goals Managers who want to create more in-depth dis-
expressed by members, such as enhancing one’s rep- cussion by increasing reply-posting behavior may tar-
utation or increasing one’s power and authority in get members’ levels of affective CC. For example,
the community, could affect individuals’ commit- if a community has a diffuse or unfocused identity,
ment levels. Additionally, other impacts of each form it may be difficult for members to identify strongly
of commitment (e.g., satisfaction, intention to leave) with it. Any effort a manager can make towards cre-
could be examined to build a more complete nomo- ating a clear and consistent community identity (com-
logical network. mon values, interests, or goals) is likely to increase
affective CC in members who see parallels between concerns researchers’ ability to compare our results
the group’s identity and their own, which will in turn concerning reply-posting behavior with other pub-
encourage reply-posting behavior. Alternately, affec- lished studies of message posting in general. Because
tive bonds are likely to be enhanced when members we do not hypothesize or test determinants of posting
feel they are supported and treated fairly and trust the in general but rather focus on a very specific type of
community manager. A range of initiatives to encour- posting (replying), our findings may not be directly
age mutual respect and accountability could therefore comparable to prior studies of more general posting
enhance affective CC and subsequent reply-posting behaviors. We see this as a necessary by-product of
behavior. our efforts to help online community research become
Finally, managers who wish to encourage infor- more specific in what it models. The increasing the-
mal moderating behaviors have two options for doing oretical precision that comes from being able to con-
so—namely, via affective CC and via normative CC. ceptualize replying as a more specific behavior makes
Members’ affective bonds seemed to have a greater it worthwhile for us to pursue this approach over a
effect on informal moderating than did their sense general and undifferentiated “posting” variable, at the
of obligation. This suggests that attempts to pro- cost of direct comparability.
mote moderating may be more effective when accom-
plished through indirect approaches—for instance, by 5.4. Conclusion
building members’ sense of identification with the Does the kind of commitment that a member devel-
community, as described above. However, managers ops towards an online community affect his or
who wish to enhance normative CC might undertake her behaviors? Our research suggests that the three
communications that stress the underlying “right- established forms of organizational commitment pro-
ness” of the community’s cause, or highlight stories of vide an important integrating framework for under-
members who displayed loyalty and who are greatly standing these bonds and their distinct behavioral
respected as a result. Initiatives such as these, which outcomes. By theorizing the community-specific out-
may increase normative CC, are likely to increase comes of affective, normative, and continuance CC,
informal moderating behaviors. this work provides a more coherent perspective on
online communities as interlocking sets of behaviors
5.3. Limitations and commitments that together support long-term
As with any empirical study, this work is subject to community viability. As such, this research provides a
limitations, the first of which has to do with method- basis for locally situated models of member behaviors
ology. Two of our three dependent variables were that will help managers enhance their communities
archival data, and one was self-reported; the results and offer researchers a foundation for future efforts
for H3 are therefore subject to the typical limitations to refine our understanding of members’ behaviors in
of cross-sectional, survey-based research. More gener- online communities.
ally, the fact that our data came from a single com-
munity (albeit a general interest community) limits Acknowledgments
the generalizability of our results, a matter that can The second author’s contribution to this research was par-
only be addressed through replication. Similarly, our tially funded by the McIntire School of Commerce Founda-
sample was made up of self-selected respondents— tion’s Dean’s Fund for Excellence.
we only have data on members who chose to read
our invitation and complete the survey. Thus, the
References
sample may not be representative of all community
Ahuja, M. K., J. E. Galvin. 2003. Socialization in virtual groups.
members, and may underrepresent those individuals J. Management 29(2) 161–185.
who do not read much content. As such, our sam- Andrews, D., B. Nonnecke, J. Preece. 2003. Electronic survey
ple may be biased towards members who were highly methodology: A case study in reaching hard-to-involve Inter-
committed. However, because respondents vary sig- net users. Internat. J. Human-Comput. Interaction 16(2) 185–210.
nificantly in their commitment scores, online behav- Arguello, J., B. Butler, E. Joyce, R. Kraut, K. S. Ling, X. Wang.
2006. Talk to me: Foundations for successful individual-group
iors, and tenure, it seems unlikely that our findings interactions in online communities. CHI 2006: Proc. ACM
are purely the result of such bias. Although it is not Conf. Human-Factors Comput. Systems, ACM Press, New York,
inconsistent with our focus on explaining why indi- 959–968.
viduals return over time to a community, the possi- Ashforth, B. E., A. M. Saks, R. T. Lee. 1998. Socialization and new-
bility of self-selection bias makes it more likely that comer adjustment: The role of organizational context. Human
Relations 51(7) 897–926.
our findings would apply to members who are regu-
Bagozzi, R. P., U. M. Dholakia. 2002. Intentional social action in
lar visitors, and less likely that they would apply to virtual communities. J. Interactive Marketing 16(2) 2–21.
members who have visited only a single time, or who Becker, H. S. 1960. Notes on the concept of commitment. Amer. J.
visit very infrequently. A final potential limitation Sociol. 66(1) 32–42.
Bergquist, M., J. Ljungberg. 2001. The power of gifts: Organizing Hall, H., D. Graham. 2004. Creation and recreation: Motivating col-
social relationships in open source communities. Inform. Sys- laboration to generate knowledge capital in online communi-
tems J. 11(4) 305–320. ties. Internat. J. Inform. Management 24(3) 235–246.
Blanchard, A. L., M. L. Markus. 2004. The experienced “sense” of Herring, S. C. 1996. Computer-Mediated Communication: Linguis-
a virtual community: Characteristics and processes. Data Base tic, Social, and Cross-Cultural Perspectives. John Benjamins,
Adv. Inform. Systems 35(1) 64–79. Philadelphia.
Brickman, P. 1987. Commitment. B. Wortman, R. M. Sorrentino, Herrmann, T., I. Jahnke, K.-U. Loser. 2004. The role concept as a
eds. Commitment, Conflict, and Caring. Prentice-Hall, Engle- basis for designing community systems. F. Darses, R. Dieng,
wood Cliffs, NJ, 1–18. C. Simone, M. Zackland, eds. Cooperative Systems Design. IOS
Burnett, G., L. Bonnici. 2003. Beyond the faq: Explicit and implicit Press, Amsterdam, 163–178.
norms in usenet newsgroups. Library Inform. Sci. Res. 25(3) 333. Hof, R. D. 2005. The power of us: Mass collaboration on the Internet
Butler, B. S. 2001. Membership size, communication activity, and is shaking up business. Bus. Week (June 20), 74–82.
sustainability: A resource-based model of online social struc- Horrigan, J. B. 2001. Online communities: Networks that nurture
tures. Inform. Systems Res. 12(4) 346–362. long-distance relationships and local ties. L. Rainie, S. Fox, eds.
Butler, B., L. Sproull, S. Kiesler, R. Kraut. 2007. Community effort in Pew Internet American Life Project, Washington, DC, 1–28.
online groups: Who does the work and why? S. Weisband, L. Hsiu-Fen, L., L. Gwo-Guang. 2006. Determinants of success for
Atwater, eds. Leadership at a Distance. Lawrence Erlbaum Asso- online communities: An empirical study. Behav. Inform. Tech.
ciates, Mahwah, NJ, 171–194. 25(6) 479–488.
Chin, W. W. 1998. The partial least squares approach for structural Jaros, S. J., J. M. Jermier, J. W. Koehler, T. Sincich. 1993. Effects
equation modeling. G. A. Marcoulides, ed. Modern Methods for of continuance, affective, and moral commitment on the with-
Business Research. Lawrence Erlbaum Associates, Mahwah, NJ, drawal process: An evaluation of eight structural equation
295–336. models Acad. Management J. 36(5) 951–995.
Christensen, C. M., M. E. Raynor. 2003. Why hard-nosed executives
Jones, S. G. 1995. Understanding community in the information
should care about management theory. Harvard Bus. Rev. 81(9)
age. S. G. Jones, ed. Cybersociety: Computer-Mediated Communi-
66–74.
cation and Community. Sage Publications, London, 10–35.
Clear, T. R., D. R. Karp. 2000. Towards the ideal of community
Jones, Q., G. Ravid, S. Rafaeli. 2004. Information overload and the
justice. NIJ J. (October) 20–28.
message dynamics of online interaction spaces: A theoretical
Cohen, A. 2003. Multiple Commitments at Work: An Integrative model and empirical exploration. Inform. Systems Res. 15(2)
Approach. Lawrence Erlbaum Associates, Hillsdale, NJ. 194–210.
Constant, D., S. Kiesler, L. Sproull. 1994. What’s mine is ours, or Joyce, E., R. E. Kraut. 2006. Predicting continued participation in
is it? A study of attitudes about information sharing. Inform.
newsgroups. J. Comput.-Mediated Comm. 11(3) 723–747.
Systems Res. 5(4) 400–421.
Kang, I., K. C. Lee, S. Lee, J. Choie. 2007. Investigation of online
Constant, D., L. Sproull, S. Kiesler. 1996. The kindness of strangers:
community voluntary behavior using cognitive maps. Comput.
The usefulness of electronic weak ties for technical advice.
Hum. Behav. 23(1) 111–126.
Organ. Sci.: J. Inst. Management Sci. 7(2) 119–135.
Kim, A. J. 2000. Community Building on the Web. Peachpit Press,
Cook, J., T. Wall. 1980. New work attitude measures of trust,
Berkeley, CA.
organizational commitment and personal need nonfulfillment.
J. Occupational Psych. 53(1) 39–52. Koh, J., Y.-G. Kim, B. Butler, G.-W. Bock. 2007. Encouraging partic-
de Souza, C.S., J. Preece. 2004. A framework for analyzing and ipation in virtual communities. Comm. ACM 50(2) 68–73.
understanding online communities. Interacting Comput. 16(3) Kollock, P., M. Smith. 1996. Managing the virtual commons: Coop-
579–610. eration and conflict in computer communities. S. Herring, ed.
Dillman, D. A. 2000. Mail and Internet Survey: The Tailored Design Computer-Mediated Communication: Linguistic, Social and Cross-
Method, 2nd ed. John Wiley & Sons, New York. Cultural Perspectives. John Benjamins, Amsterdam, 109–128.
Ekeh, P. 1974. Social Exchange Theory: The Two Traditions. Harvard Lakhani, K., E. von Hippel. 2003. How open source software works:
University Press, Cambridge, MA. “Free” user-to-user assistance. Res. Policy 32(6) 923–943.
Festinger, L. 1957. A Theory of Cognitive Dissonance. Stanford Lampe, C., P. Resnick. 2004. Slash(dot) and burn: Distributed mod-
University Press, Stanford, CA. eration in a large online conversation space. ACM Comput.
Fisher, D., M. Smith, H. T. Welser. 2006. You are who you talk to: Human Interaction Conf. 2004, Vienna.
Detecting roles in usenet newsgroups Proc. 39th Hawaii Internat. Lampel, J., A. Bhalla. 2007. The role of status seeking in online
Conf. System Sci. Kauai, HI. communities: Giving the gift of experience. J. Comput.-Mediated
Furlong, M. S. 1989. An electronic community for older adults: The Comm. 12(2) 100–121.
seniornet network. J. Comm. 39(3) 145–153. Lee, G. K., R. E. Cole. 2003. From a firm-based to a community-
Gefen, D., C. M. Ridings. 2005. If you spoke as she does, sir, instead based model of knowledge creation: The case of the Linux
of the way you do: A sociolinguistics perspective of gender dif- kernel development. Organ. Sci.: J. Inst. Management Sci. 14(6)
ferences in virtual communities. Data Base Adv. Inform. Systems 633–649.
36(2) 78–92. Lerner, J., J. Tirole. 2002. The simple economics of open source.
Gouldner, A. W. 1960. The norm of reciprocity. Amer. Sociol. Rev. J. Indust. Econom. 50(2) 197–234.
25(2) 165–167. Ma, M., R. Agarwal. 2007. Through a glass darkly: Information
Grant, A. M. 2007. Relational job design and the motivation to make technology design, identity, verification, and knowledge con-
a prosocial difference. Acad. Management Rev. 32 393–417. tribution in online communities. Inform. Systems Res. 18(1)
Greer, B. G. 2000. Psychological and social functions of an e-mail 42–67.
mailing list for persons with cerebral palsy. CyberPsychology Markus, M. L., B. Manville, C. Agres. 2000. What makes a virtual
3(2) 221–233. organization work—Lessons from the open source world. Sloan
Griffeth, R. W., P. W. Hom, S. Gaertner. 2000. A meta-analysis Management Rev. 42(1) 13–26.
of antecedents and correlates of employee turnover: Updated Mathieu, J. E., D. M. Zajac. 1990. A review and meta-analysis of
moderator tests, and research implications for the next millen- the antecedents, correlates, and consequences of organizational
nium. J. Management 26(3) 463–488. commitment. Psych. Bull. 108 171–194.
Meyer, J. P., N. J. Allen. 1991. A three-component conceptualiza- Ridings, C. M., D. Gefen, B. Arinze. 2002. Some antecedents and
tion of organizational commitment. Human Resource Manage- effects of trust in virtual communities. J. Strategic Inform. Sys-
ment Rev. 1(1) 61–89. tems 11(3–4) 271–295.
Meyer, J. P., N. J. Allen. 1997. Commitment in the Workplace: The- Ridings, C. M., D. Gefen, B. Arinze. 2006. Psychological barriers:
ory, Research, and Application. Sage Publications, Thousand Lurker and poster motivation and behavior in online commu-
Oaks, CA. nities. Comm. AIS 18 329–354.
Meyer, J. P., L. Herscovitch. 2001. Commitment in the workplace: Riketta, M. 2002. Attitudinal organizational commitment and job
Toward a general model. Human Resource Management Rev. performance: A meta–analysis. J. Organ. Behav. 23(3) 257–266.
11(3) 299–326.
Rousseau, D. M. 1998. The “problem” of the psychological contract

Meyer, J. P., D. J. Stanley, L. Herscovitch, L. Topolnytsky. 2002. considered. J. Organ. Behav. 19(1) 665–671.
Affective, continuance, and normative commitment to the Sampson, R. J. 1995. The community. J. Q. Wilson, J. Petersilia,
organization: A meta-analysis of antecedents, correlates, and eds. Crime. Institute for Contemporary Studies, San Francisco,
consequences. J. Vocational Behav. 61(1) 20–52. 193–216.
Moon, J. Y., L. S. Sproull. 2008. The role of feedback in managing Saubramanian, A., S. Pardhan. 2006. The repeatability of mnread
the Internet-based volunteer work force. Inform. Systems Res. acuity charts and variability at different test differences. Optom-
19(4) 494–515. etry Vision Sci. 83(8) 572–576.
Mowday, R. T. 1998. Reflections on the study and relevance of orga- Somers, M. J. 1995. Organizational commitment, turnover, and
nizational commitment. Human Resource Management Rev. 8(4) absenteeism: An examination of direct and interaction effects.
387–401. J. Organ. Behav. 16(1) 49–58.
Noddings, N. 1984. Caring: A Feminine Approach to Ethics and Moral Somers, M. J., D. Birnbaum. 1998. Work-related commitment and
Education. University of California Press, Berkeley. job performance: It’s also the nature of the performance that
Oreg, S., O. Nov. 2008. Exploring motivations for contributing to counts. J. Organ. Behav. 19(6) 621–634.
open source initiatives: The roles of contribution context and Sproull, L., S. Kiesler. 1991. Connections: New Ways of Working in the
personal values. Comput. Human Behav. 24(5) 2055–2073. Networked Organization. MIT Press, Boston.
Organ, D. W., K. Ryan. 1995. A meta-analytic review of attitudi- Swailes, S. 2002. Organizational commitment: A critique of the
nal and dispositional predictors of organizational citizenship construct and measures. Internat. J. Management Rev. 4(2)
behavior. Personnel Psych. 48(4) 775–802. 155–178.
Pirolli, P., S. K. Card. 1999. Information foraging. Psych. Rev. 106(4) Wasko, M. M., S. Faraj. 2000. “It is what one does”: Why people
643–675. participate and help others in electronic communities of prac-
Porter, L. W., R. M. Steers, R. T. Mowday, P. V. Boulian. 1974. Orga- tice. J. Strategic Inform. Systems 9(2–3) 155–173.
nizational commitment, job satisfaction, and turnover among Wasko, M. M., S. Faraj. 2005. Social capital and knowledge contri-
psychiatric technicians. J. Appl. Psych. 59(5) 603–609. bution. MIS Quart. 29(1) 35–57.
Preece, J. 1999. Empathy online. Virtual Reality 4(1) 74–84. Wellman, B., M. Gulia. 1999. Net surfers don’t ride alone: Virtual
Preece, J. 2000. Online Communities: Supporting Sociability, Designing communities as communities. P. Kollock, M. Smith, eds. Com-
Usability. John Wiley & Sons, Chichester, UK. munities in Cyberspace: Perspectives on New Forms of Social Orga-
Preece, J. 2004. Etiquette online: From nice to necessary. Comm. nization. University of California Press, Berkeley, 167–194.
ACM 47(4) 560–061. Welser, H. T., E. Gleave, D. Fisher, M. Smith. 2007. Visualizing the
Preece, J., K. Ghozati. 1998. In search of empathy online: A review signatures of social roles in online discussion groups. J. Soc.
of 100 online communities. Proc. 1998 Assoc. for Inform. Sys- Structure 8(2) 1–32.
tems Americas Conf., Association for Information Systems (AIS), Whitener, E. M., P. M. Walz. 1993. Exchange theory determinants of
Atlanta, 92–94. affective and continuance commitment and turnover. J. Voca-
Ren, Y., R. Kraut, S. Kiesler. 2007. Applying common identity and tional Behav. 42(3) 265–281.
bond theory to design of online communities. Organ. Stud. Wiener, Y. 1982. Commitment in organizations: A normative view.
28(3) 377–408. Acad. Management Rev. 7(3) 418–428.
Rheingold, H. 1993. The Virtual Community: Homesteading on the Elec- Williams, R. L., J. Cothrel. 2000. Four smart ways to run online
tronic Frontier. Addison-Wesley, Reading, MA. communities. Sloan Management Rev. 41(4) 81–91.
Ridings, C. M., D. Gefen. 2004. Virtual community attraction: Why Zhang, W., S. A. Watts. 2008. Capitalizing on content: Information
people hang out online. J. Comput.-Mediated Comm. 10(1) Article adoption in two online communities. J. Assoc. Inform. Systems
4. http://jcmc.indiana.edu/vol10/issue1/ridings_gefen.html. 9(2) 73–94.

Bateman 2011

Uploaded by

Copyright:

Available Formats

Bateman 2011

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Bateman 2011

Uploaded by

Copyright:

Available Formats

This article was downloaded by: [192.122.237.

41] On: 26 June 2018, At: 17:59

Information Systems Research

Research Note—The Impact of Community Commitment

To cite this article:

Full terms and conditions of use: http://pubsonline.informs.org/page/terms-and-conditions

Copyright © 2011, INFORMS

Please scroll down for article—it is on subsequent pages

Participation in Online Communities

1. Introduction participated in an online community (Horrigan 2001),

tion (Bagozzi and Dholakia 2002, Lee and Cole 2003).

Figure 1 Research Model

Note. Controls shown with dashed lines.

Table 2 Means, Standard Deviations, and Correlations 4n = 1935

1 Gender 0084 n.a. n.a. n.a. n.a.

Figure 2 Results of PLS Analysis

By adopting a commitment framework, we demon- commitment in an organizational setting sometimes

bers primarily motivated by continuance CC are also

Rousseau, D. M. 1998. The “problem” of the psychological contract

You might also like