Organizing without formal organization: Group Identification, Goal
Setting and Social Modeling in Directing Online Production
Haiyi Zhu, Robert Kraut, Aniket Kittur
Carnegie Mellon University
Pittsburgh PA 15213
haiyiz@cs.cmu.edu
robert.kraut@cmu.edu
nkittur@cs.cmu.edu
ABSTRACT
ACM Classification Keywords
A challenge for many online production communities is to
direct their members to accomplish tasks that are important
to the group, even when these tasks may not match
individual members’ interests. Here we investigate how
combining group identification and direction setting can
motivate volunteers in online communities to accomplish
tasks important to the success of the group as a whole. We
hypothesize that group identity, the perception of belonging
to a group, triggers in-group favoritism; and direction
setting (including explicit direction from group goals and
implicit direction from role models) focuses people’s
group-oriented motivation towards the group’s important
tasks. We tested our hypotheses in the context of
Wikipedia's Collaborations of the Week (COTW), a group
goal setting mechanism and a social event within
Wikiprojects. Results demonstrate that 1) publicizing
important group goals via COTW can have a strong
motivating influence on editors who have voluntarily
identified themselves as group members compared to those
who have not self-identified; 2) the effects of goals spill
over to non-goal related tasks; and 3) editors exposed to
group role models in COTW are more likely to perform
similarly to the models on group-relevant citizenship
behaviors. Finally, we discuss design and managerial
implications based on our findings.
H.5.3 [Information Interfaces and Presentation]: Group and
Organization Interfaces – Collaborative computing,
Computer-supported cooperative work, Web-based
interaction; K.4.3 [Computers and Society]: Organizational
Impacts – Computer supported collaborative work.
Author Keywords
Online Production Communities, Group Identification,
Governance Mechanisms, Directing Behaviors, Group
Goals.
Permission to make digital or hard copies of all or part of this work for
personal or classroom use is granted without fee provided that copies are
not made or distributed for profit or commercial advantage and that copies
bear this notice and the full citation on the first page. To copy otherwise, or
republish, to post on servers or to redistribute to lists, requires prior
specific permission and/or a fee.
CSCW 2012, February 11–15, 2012, Seattle, Washington.
Copyright 2011 ACM XXX-X-XXXXX-XXX-X/XX/XX...$5.00.
INTRODUCTION
Online production communities are becoming increasingly
important in creating innovative products in the networked
world. These organizations have successfully aggregated
the efforts of millions of volunteers to produce complex
artifacts such as GNU/Linux and Wikipedia. Currently most
large online projects primarily rely on a paradigm of selfdirection in which contributors work primarily on the tasks
they are interested in. This paradigm provides a number of
benefits. Contributors are motivated to work on the tasks in
which they are intrinsically interested in and are likely to
choose tasks in which they already have some expertise [4].
However, this approach breaks down when there are
conflicts between the interests of the contributors and the
interests of the project as a whole. Many people may want
to work on the same popular areas (e.g., an article on
“Barack Obama” in Wikipedia) while ignoring less popular
areas that require work. Contributors may not want to
perform maintenance and other unattractive tasks, even
though these tasks are important to the continued
functioning and health of the project.
Many techniques used in conventional employment
organizations are not effective in managing online
volunteers due to the fundamental characteristics of online
communities, including lack of employment contracts,
weak external incentives, weak interpersonal bonds,
impoverished communication, large size, and high turnover
[20]. For example, if a project tries to exert too much
managerial control, volunteers can simply leave, with fewer
economic or social consequences than if they had quit a job
or left a real-life social group.
Instead, communities must turn to other means of
motivating volunteers to accomplish tasks that are
important for the welfare of the group. One technique is by
leveraging group identification—the perception of
belonging to a group. If volunteers feel that their identities
are tied to the identity of the group, their goals may be more
likely to reflect those that are important to the group
[1,15,19,32]. However, group identification by itself does
not specify which particular tasks to work on.
In contrast, direction setting—for example by specifying
goals—can be an effective mechanism for accomplishing
specific tasks [3,21,22]. However, direction setting by itself
may not be enough. For example, Cosley and his colleagues
found that task recommendations based only on the
community’s needs are less likely to interest members than
those that consider personal needs [9]. These challenges
may become even more pronounced for tasks that are
considered unpleasant or unrewarding.
We hypothesize that group identification and direction
setting can complement each other in managing volunteers’
efforts. Group identification can align the individual
volunteer’s goals with the group’s goals, while direction
setting can channel their effort toward specific group goals.
Thus people who identify themselves as group members
may voluntarily follow directions based on group needs and
perform tasks valued by the group because they believe that
investing effort in these tasks is important for the group and
thus validates their own identity.
This paper describes a mechanism to motivate and manage
volunteers when standard managerial mechanisms deployed
in conventional organizations are not available. This
mechanism combines group identification and direction
setting. Particularly, two sources of direction setting are
investigated – explicit direction based on publicized group
goals and implicit direction based on role modeling. We test
the effectiveness of the mechanism in the context of
WikiProjects, subgroups within Wikipedia. After presenting
the main findings we also discuss design implications for
governance in online communities.
TRADITIONAL MANAGEMENT MECHANISMS
Markets
The market mechanism relies on individuals to make
independent decisions about the tasks they want to
undertake and how they will do them. In contrast to simple
self-direction, market mechanisms use external incentives,
such as price, to regulate participants’ behaviors. Amazon’s
Mechanical Turk, a popular crowd sourcing website, uses
price to encourage subscribers to undertake tasks that
employers care most about. If Wikipedia applied a
monetary market mechanism, it would pay editors more for
editing important but unpopular articles or for engaging in
important but tedious tasks such as maintenance work.
However, volunteer peer production systems rarely have the
resources to provide external incentives to get important
work done. External incentives may undermine people’s
intrinsic motivation to contribute if they become focused on
the extrinsic rewards [11]. Finally, they may be inconsistent
with the ideology that drives some volunteer communities.
Bureaucratic Control
Three primary controlling strategies evolved in the history
of modern organizations [2]. First was “simple control”,
which represents direct and personal supervision by factory
owners. The second was “technological control”, in which
simple control was reinforced by physical technology such
as the assembly line in traditional manufacturing. The most
familiar is bureaucratic control, which derives control from
hierarchical social relations between supervisor and
supervisee and sets of systemic rules that reward
compliance and punish noncompliance [2]. A supervisor
can legitimately assign employees tasks and rewards and
punish them based on their amount and quality of work.
Bureaucratic control legitimizes the roles of managers, so
that employees see themselves as having an obligation to
adhere to the decisions made by their managers. External
incentives, including monetary rewards such as raises and
bonuses, and social ones including promotions and better
assignments, supplement this legitimacy and are also
important in causing employees to follow the direction of
their managers.
Bureaucratic control has become the primary control
strategy in conventional modern organizations. Some
degree of bureaucratic control exists in online production
communities, as well. For many years, Linus Torvalds had
significant control in the community developing the Linux
operating system. Although by definition managers cannot
use wages as incentives to get volunteers to comply with
their directives, they can motivate contributors through
promotion from rank-and-file positions to more important
ones, such as committer status in open source software
development projects [29] or administrator status in
Wikipedia [6].
However, the effectiveness of bureaucratic control is
limited by other characteristics of online production
communities. As with market mechanisms, online
production communities cannot afford external incentives.
Furthermore, tight managerial control of volunteers,
including regular supervision and communication with
them, is associated with higher turnover rates in offline
volunteer organizations. According to Hager and Brudney,
bureaucratic control may cause their “volunteer experiences
to feel too much like the grind of their daily work rather
than an enjoyable avocation,” [14, p. 9] and thereby drive
them away. In addition, impoverished communication and
weak interpersonal bonds in online communities weaken
the managers’ ability to exert bureaucratic control [10].
INCORPORATING GROUP IDENTITY AND DIRECTION
SETTING
Group identity
Tajfel and his colleagues conducted a series of laboratory
studies in the early 1970s showing that the mere perception
of belonging to a group – that is, social categorization per
se – is sufficient to trigger intergroup discrimination
favoring the in-group [32, 33, 34]. For example, when
assigned to groups on the basis of trivial criteria,
participants tend to award more rewards to in-group
members than outgroup members. Tajfel and his colleagues
introduced the concept of social identity and developed
classic social identity theory. Social identity is “the
individual’s knowledge that he belongs to certain social
groups together with some emotional and value significance
to him of the group membership” [32]. Social identity rests
on intergroup social comparisons, in which members
attempt to establish or confirm ingroup-favoring evaluative
distinctiveness between ingroup and outgroup. Social
identity is motivated by an underlying need for self-esteem
[34]. That is, to the extent people have incorporated the
group’s identity into their personal identities, positive
evaluation of the group results in enhanced self-esteem.
The relationship of social identity and in-group favoritism
plays out in work environments. In offline organizations,
social identity leads individuals to perform behaviors
beneficial to the groups of which they are part (see [1] for a
review). The outcomes associated with social identity
involve cooperation, effort, participation, organizationally
beneficial decision making, intrinsic motivation, task
performance, information sharing, and coordinated action.
Recently work has extended the analysis to online volunteer
communities as well. Kittur and his colleagues [19]
examined the effects of group identification in Wikipedia,
finding that joining a WikiProject (a subgroup in Wikipedia)
was associated with increased production work,
coordination work and citizenship behaviors.
Direction setting: Goal setting & social modeling
The in-group favoritism that results from group
identification alone is often too diffuse to effectively direct
volunteers toward specific actions. Volunteers, who identify
with a group and want to benefit it, have wide latitude in
selecting behaviors that benefit the group. Therefore, we
hypothesize that direction setting could complement group
identification by focusing people’s group-oriented
motivation towards important and necessary tasks for the
group.
Previous researchers interested in increasing contribution in
online communities have often focused on getting
volunteers to provide more of what they already contribute.
For example, Beenen et al. examined the effects of goal
setting in MovieLens [3]. They assigned performance goals
(e.g., number of movies to rate), while allowing volunteers
to self-select specific targets (e.g., which movies to rate).
Cosley and his colleagues designed task recommendation
systems in Wikipedia and MovieLens. However, these
systems focused on matching individuals with tasks they
are already interested in [8,9]. Below we discuss how two
direction setting mechanisms—explicit goal setting and
implicit social modeling—can motivate self-identified
group members to work on tasks important for the group’s
interests, rather than their own interests.
Goal setting
A goal is the object or aim of an action, usually within a
specified time limit [22]. Goal setting can be an effective
technique to direct human attention and efforts toward goalrelevant activities and away from goal-irrelevant activities
[22, 23]. For example, students with specific learning goals
attend to and learn goal-relevant passages better than goalirrelevant passages [30]; similarly, when people receive
feedback, they only improve their performance on
dimensions for which they have goals even when receiving
feedback on multiple dimensions [21]. In addition to the
directive function, goals can motivate high task
performance. Goals have an energizing function – high
goals lead to greater effort than low goals. Goals also affect
persistence – they extend directed effort over time. Finally,
goals also affect action indirectly by leading to strategy
development and action plans for attaining ones’ goal [22].
Group goals, which highlight important tasks for the group
as a whole, can direct people’s attention and efforts towards
these tasks and improve their performance on these tasks.
The effects are strongest when people perceive goals as
desirable and important for them and thus are committed to
the goal [22]. As we discussed previously, people who
identify with the group align their own interest with the
group’s interest; therefore they are more likely to invest
their efforts to achieve group goal than people who do not
identify with the group because they believe the goals are
important to the group and thus important for themselves.
Hypothesis 1 (Direct effects of goal setting).
H1a. Highlighting tasks important to the group through
goal setting directs people’ efforts towards these tasks
and improves performance on these tasks. H1b. The
effect is stronger for people who identify with the group
than those who do not identify with the group.
If we assume volunteers’ total efforts are fixed, group goals
would only redistribute their efforts. However, there are
reasons to expect that volunteers’ total efforts will be
increased by group goal setting. Specifically, group goals
might lead to motivational spillover, in which people
increase their efforts on behalf of the group beyond that
demanded by the original goals. Because of expectancy
effects, success and failure on one task may change
motivations for subsequent tasks [18, 23, 28].
Accomplishing group goals can lead to rewards such as
recognition and reputation, activating people to continue
working after the initial task is accomplished. Furthermore,
publicizing group goals may activate people’s awareness of
the group, which then leads to more group relevant
activities and contributions.
Hypothesis 2 (Spillover effects of goal setting).
Group goals increases people’s general contributions to
group-related tasks.
Social modeling
There are often a set of prototypical members in groups
who best embody the features that are valued by the group
[15, 35]. In volunteer associations and online production
groups, the prototypical members are often a small set of
core members who perform large amounts of work, engage
in coordination activities, and have significantly more
knowledge of the group and the community than peripheral
members [26, 27]. The prototypical group members serve
as models, providing cues for what behavior is valued, and
make salient the situational needs for certain actions.
According to social identity and self-categorization theories,
individuals who identify themselves as group members tend
to spontaneously change their behaviors to be more similar
to these prototypical members [15, 35]. In contrast,
prototypical members should have less of an effect on those
who does not consider themselves as group members [15,
35].
However, for social modeling to occur, the prototypical
members should be visible, so that group members can
perceive them as role models and to imitate their behaviors.
Hypothesis 3. (Effects of social modeling)
H3a Exposure to prototypical group members should
lead people to perform more group-valued behaviors that
prototypical members engage in. H3b The effect is
stronger for people who identify with the group than
those who do not identify with the group.
STUDY PLATFORM
article lists. Some Wikiprojects list their most valued
articles in their project pages, encouraging people to
improve these. 3) Contests. Some Wikiprojects set goals
and then reward people who contribute the most to them
over a defined time period. 4) Collaborations of the week
(COTW). Projects set one or two articles need to improve
during a defined time period (usually one week to one
month).
Collaborations of the Week (COTW)
In this paper, we investigated a specific mechanism,
collaborations of the week, which designate one or two
articles to improve in a defined period. Collaborations of
the Week (COTW) are a widely-applied mechanism in
Wikiprojects. As of March 2008, 189 Wikiprojects had
conducted COTWs for at least part of their history.
COTWs usually have two phases - selection and
collaboration. In the selection phase the project chooses one
or two articles on which members will collaborate. In some
projects, the article is chosen through voting. Other projects
rely on the judgment of coordinators for article selection. In
other cases, the choice is made by an automated program.
During the collaboration phase, the project tags the chosen
article(s) with a special template in its talk page (as shown
in Figure 1). This template is visible to all editors who read
the article talk page, not just those who are members of the
Wikiproject. In addition, the project typically announces the
targets of the collaboration on its project pages (as shown in
Figure 2). Some projects also send special reminders to
project members (those editors with names on member list)
Wikiprojects – groups in Wikipedia
We chose Wikiprojects, subgroups in Wikipedia, as the
domain in which to investigate the effects of group
identification and direction setting. Wikiprojects are
collections of editors interested in specific topics such as
military history, psychology, or even copyediting. As of
March 2008, Wikipedia contained more than 2000
Wikiprojects.
Each Wikiproject has dedicated pages (known as project
pages) on which editors can coordinate and organize the
writing and the editing of project-related articles.
Wikiprojects have a loose membership structure; any editor
can participate in project activities and contribute to articles
within projects as they see fit. Editors often express their
identification with a project by adding their name to a
member list or tagging their personal user pages with the
project template. Some projects have explicit coordinators,
who are responsible for coordinating maintenance tasks and
keeping the project functioning.
Wikiprojects employ a variety of techniques to direct
members’ attention to project valued-tasks [19]. These
techniques include: 1) Open task lists or article alerts.
Many Wikiprojects list from dozens to hundreds of open
tasks in their project pages. These lists identify articles that
need to be expanded, assessed, copy-edited or reviewed and
discussions that need more participation. 2) Important
Figure 1. An example template identifying an article as a
collaboration of the week.
Figure 2. A collaboration of the week announcement in a
project page
on their personal talk pages.
We chose to examine the Collaborations of the Week as
group goal setting mechanism for a number of reasons:
• COTWs are a project goal setting mechanism that
highlights tasks crucial for the Wikiproject. For example,
some projects explicitly claim that the goal of
collaborations is to “fill the gap” of the Wikiproject [38];
collaboration targets are typically articles rated as high
importance but having low quality [40]. Furthermore,
COTWs have many properties of effective goals,
according to the goal setting theory [22]. Compared with
a diffuse open task list, for example, COTWs set specific,
concrete and time-limited requirements for editors. The
limited number of articles and defined time period focus
editors’ attention on these articles, potentially leading to
both production and social benefits.
• COTWs are also social events. COTWs focus volunteers
towards specific targets during a defined period,
providing opportunities for volunteers to discuss plans
and progress with each other, and potentially to influence
each other. According to a small survey we conducted
with COTW participants, COTWs are “a chance to get to
meet your collaborators and their interests”. COTW
participants are “virtually surrounded by peers who are
into the topic and you all have the common goal of
sharing knowledge together”.
• COTWs are salient. Notices for COTWs are prominent
on project pages, thereby attracting people who care
about the project, and on the talk pages of the articles
which are targets of the collaboration, thereby attracting
editors interested in the specific article. Also, the effects
of COTWs are amenable to analysis. Firstly, COTWs are
widely-used so we can obtain sufficient data for analysis.
Secondly they have clear-cut start times and end times.
We can compare editors’ behavior on the same articles
when they are the subjects of collaborations and at other
times.
We included in our sample editors who had edited the
collaborated target articles either during the collaboration
period (week or month) or the pre- and post-collaboration
period (week or month). We assume that all of these editors
were aware of the event, at least from the advertisement
notice on the article talk page.
To test the direct effects of group goals, we examined
whether these editors’ contributions increased during the
goal period (the period when the articles are selected as
collaboration targets) compared to the non-goal period (the
pre- and post-collaboration period). For the effects of group
identification, we further investigated whether the
contribution increase during the goal period was larger for
editors who self-identified as group members than for those
who did not.
1.2 Dependent Variable
Goal-relevant Contribution: We measured editors’
contributions towards goal-related articles through their
revision count on that article. Revisions are a measure of
editors’ effort, indicating the number of changes they make
to articles during a period of time. Each revision comprises
a set of editing actions, for example adding, changing,
deleting or reverting text, references or illustrations, or
communicating with other editors. In this analysis, the
dependent measure was the number of revisions the editor
made to the COTW articles or their associated discussion
pages.
METHOD
1.3 Independent Variables
Data Collection
Goal period: Collaborations of the Week are explicit group
goals that designate one or two articles as targets of work
during a defined time period. When editors revise and add
to these articles during that period, we consider that they
are following the group’s goals. However, editing other
articles or editing the COTW articles at other periods did
not fulfill the group goals in this context. To assess the
effectiveness of these goals, we compared contributions
towards the same target articles in different time periods –
pre-collaboration, during collaboration and postcollaboration. In the analysis, pre-, during and postcollaboration periods were of the same length. For example,
if the collaboration lasted one week, pre-collaboration is the
week before the start of collaboration; while postcollaboration includes the week after the end of the
collaboration. In particular, the dummy variable “Goal
period” in our analysis was defined as 0 during the precollaboration and post-collaboration periods, and 1 during
the collaboration period.
In the following analysis, we used a complete download
provided by the MediaWiki Foundation from Wikipedia’s
inception to March 2008 (approximately 182 million
revisions). To handle this data volume, we used the Yahoo!
M45 computing cluster running Hadoop and Pig. Among
the 189 projects that ever used COTW for goal setting, we
chose projects with at least five collaborations that had
explicit time periods and complete collaboration histories.
We also excluded redirected projects and two collaborationoriented projects which do not have their own topics. The
remaining 26 projects carried out a total of 618
collaborations, which lasted 17.7 days on average.
The 26 projects were large and important ones in Wikipedia.
They include eight of the ten largest projects in Wikipedia.
On average, each project encompassed 26,553 articles
(median = 4,632) and 471 members (median = 255.5).
Overall, these 26 projects contained 68.5% of all articles
associated with any project in Wikipedia.
ANALYSIS AND RESULTS
1. Direct Effects of Goal Setting (H1)
1.1 Analysis Strategy
H1 predicts that, although any editor can participate in the
Collaborations of the Week, people who identify
themselves as group members in particular will be
especially likely to work more on goal-related articles.
Group identification: Originally, we operationalized
people who identified with the group as those who edited
the project member lists. However, we found edits to the
project members list page were not always a good indicator
of group identification, as members often added the names
of others to the page (e.g., if the page was copied or
repurposed from another source). Therefore, we determined
self-identified group members to be all editors who have
edited any project page, under the assumption that editors
who are involved in the organization of project activities
are more likely to consider themselves group members. We
used a dummy variable to indicate group identification: 0
indicates the editor has not identified as a group member,
while 1 indicates the editor has identified as a group
member.
1.4 Control Variables.
Goal length: the number of weeks the collaboration lasts.
Project articles: the total number of articles in the scope of
the project during the given period.
Project members: the total number of project members
during the given period.
1.5 Statistical Model
We conducted an editor-level analysis, with revision count
of contributors to the article as the dependent variable.
Because revision counts are count data with a non-normal
distribution truncated at zero, we used a negative binomial
regression model. Because the analysis compared the
contributions from the same editor in different time periods
and one of the explanatory variables is constant for an
individual, we used random effects methods to deal with the
panel data set [16].
1.6 Analysis Results
Figure 3 shows the average number of revisions per editor
on collaboration targets in different time periods. We found
that people in general contributed more during
collaboration periods, but the effect is dramatically larger
Revision Counts
on Collaborated Articles
7
Non self-identified
editors
6
5
Self-identified
group members
4
for those who identified with the group: editors who
identified with the group contribute approximately three
times more during the collaboration period than they did
before the collaboration period, and four times more than
editors who did not identify with the group.
The negative binomial regression model with random
effects methods predicting revision counts on COTW
articles tests the significance of these results we ran. The
results of the analysis are shown in Table 1, with the effects
reported as Incidence Rate Ratios (IRR), which can be
interpreted as the ratio change of the dependent variable
when increasing an independent variable by one unit. The
model assumes that contributions from non-self-identified
editors during non-collaboration periods are the baseline
level. During collaboration periods, non-self-identified
editors increased their contributions 107%, while selfidentified editors increased 405% compared to baseline.
The main effect (PGoal_period < 0.001) and interaction effect
(PGoal_period*Group_identification < 0.001) are both highly
significant. These results support H1, suggesting that
COTWs have a strong motivating effect on contribution,
and the effect is especially strong for editors who identify
with the project.
The results also suggest that the number of weeks a COTW
lasts has a slight negative effect on contributions. Although
statistically significant, the size of this effect is quite small,
suggesting care must be taken in making too much of it.
Factors such as the total number of project articles and
project members do not have significant effects. Together,
these results suggest that the group goal settings coupled
with projects is robust and applies across variations in the
length of goal period, and project characteristics.
2. Spillover Effects of Goal Setting (H2)
2.1 Analysis Strategy
The previous analysis demonstrated that group goals set via
Collaborations of the Week energized editors, especially
self-identified project members. We now examine whether
accomplishing these COTW-set group goals influence these
project members’ editing contributions beyond the targets
of the group goal.
3
2
1
0
Pre-Collaboration
Predictors
Collaboration
Post-Collaboration
IRR.
Std.
P>|z|
Err.
Figure 3. Average revision counts on collaboration target articles
1.424
.301 of editors.
<.001
identification
inGroup
different
time periods from
different types
(1-self-identified;
0-not identified)
2.066
.036
<.001
Goal period
* Group identification
2.975
.102
<.001
Goal length
.996
.001
0.002
Project members
1.000
2.16e-5
<.001
Project articles
1.000
1.07e-7
.105
Goal period
(1- collaboration period;
0 – pre & post collaboration)
Log likelihood
-42894.534
Table 1. Negative binomial regression model predicting goal
relevant contributions (revision counts on collaboration
target articles). IRR: the ratio change of the dependent
variable by increasing an independent variable by a unit.
We examined the 26 projects in different time periods. We
investigated whether the projects received more
contributions on goal-irrelevant articles when group goals
were posted compared to the period when there were no
group goal goals at all.
2.2 Dependent Variable
Non-related contributions: the average number of
revisions done by each self-identified project member on all
articles in the scope of a given project (including associated
discussion pages) in a given month, excluding the revisions
on COTW target articles.
2.3 Independent Variable
Goal period: a dummy variable indicating whether the
project posted COTW goals in a given month. Even though
all of the projects in the sample used COTWs at some time
in their histories, they used them in only 46% of the months
in the dataset.
2.4 Control Variables
Project articles: number of articles in the project.
Project members: total number of project members signed
up before the end of the given month.
Project coordination activity: number of revisions made
to the project pages in the given month. Since these project
pages are where editors organize and discuss project
activities, this variable reflects the overall activity of the
group during the time period. We used this variable to
control for other project activities which might influence
contribution towards the project.
Project age: number of months the project has been in
existence, starting month one (the month when the project
was created). We used this variable to control for the
maturity of the project which might influence how much
effort people will devote towards the project.
2.5 Statistical Model
For reasons similar to those for the previous analysis, we
also applied a negative binomial regression model with
random effects to fit the data.
2.6 Analysis Results
The results reveal that the presence of a Collaboration of
the Week substantially increased the average number of
edits done by project members (IRR = 2.14, P<0.001). The
effect is substantial: the presence of COTW goals induced
project members to approximately double their
contributions on non-target articles. To put this in context,
during the month the project posted COTW goals, selfidentified group members on average made 9 edits to the
collaboration target articles and 60 more edits to other
articles in the scope of the project compared to non-COTW
month. Thus it appears that employing shared group goal
mechanisms such as COTWs can have large benefits to
contributions to the project that go beyond the articles
identified as collaboration targets.
IRR
SE
P value
Setting goals
2.140
.097
<0.001
Project
Activities
1.000
3.61e-05
<0.001
Project
Members
1.000
9.13e-05
0.002
Project
Articles
1.000
4.26e-07
<0.001
Project Age
1.043
Log likelihood
0.002
<0.001
-3121.773
Table2. Negative binomial regression model with random
effects predicting goal-irrelevant group-related contributions
3. Effects of Social Modeling (H3)
3.1 Analysis Strategy
Group goal presents explicit direction setting while social
modeling is more implicit. When editors work together to
accomplish group goals, they can be exposed to
prototypical project members, who may serve as role
models, and whose behavior provides implicit direction to
others (especially self-identified group members).
According to prior research, social modeling may be a
useful way to influence a particularly important kind of
contribution: citizenship behavior [31]. Citizenship
behavior has been defined by Organ [25] as the types of
“extra-role” behaviors that are not explicitly recognized by
the formal reward system, but are vital to the continued
functioning of the organization. For example, the central
and most valued work in Wikipedia is creating good quality
articles. Adding content to articles is not sufficient.
Established editor brag about the number of articles they
have brought to “featured article” status. In contrast,
maintenance tasks, such as copy-editing, formatting
citations, welcoming newcomers, reverting vandalisms, and
assessing articles, are actually important to wikipedia as a
whole, but less explicitly value or rewarded. Wikipedians
wrote of them as “tedious, often unrewarding, and usually
unappreciated” tasks [39]. Many of the non-writing
wikiwork identified by Kriplean et al [17], such as teaching
rewarding welcoming others, finding sockpuppets,
reverting vandalism, assessing articles and creating
templates, comprise citizenship behavior in Wikipedia. In
the analyses below, we treat reverting vandalism and article
assessment as representative citizenship behaviors.
We define prototypical members as those who were the
heaviest contributors in project pages and at the same time
participated in collaborations of the week in a given period.
We selected regular editors as non-prototypical members
who also participated in COTWs at least once. To measure
the influence of role models, we calculated the correlation
between their citizenship behaviors with the citizenship
behaviors of regular editors, considering 1) whether the
regular editors identified themselves as project members or
not, and 2) whether the regular editors participated in
COTWs in the given period or not. According to hypothesis
3, the correlation between prototypical members’ behaviors
and the behaviors of the regular editors will be higher when
the regular editors participated in COTWs than during other
periods. Furthermore, people who self-identified as group
members and participated in COTWs should have the
highest correlation with prototypical members.
3.2 Dependent variables: citizenship behaviors
Anti-vandalism correlation: Vandalism is defined as “any
addition, removal, or change of content made in a deliberate
attempt to compromise the integrity of Wikipedia” [36].
Anti-vandalism is the behavior of reverting the vandalized
version to a previous state. Following previous research
[19], we quantified anti-vandalism as edits annotated with
common vandalism-fighting comments, such as “Reverting
3.3 Analysis Results
The results are shown numerically in Table 3, and
graphically in Figure 4 and Figure 5. For assessments, the
results are consistent with the Hypothesis 3. Compared to
editors who did not participate in collaborations of the week,
editors who were exposed to prototypical members through
the Collaborations of the Week performed more similarly to
prototypical members in terms of helping assess articles.
Editors who self-identified as group members and
participated in the Collaborations of the Week acted most
similar to prototypical members (r=0.36), compared to selfidentified members in other months (r=0.24) or to non-selfidentified editors, either in the month participating
collaborations (r=0.08) or other months (r= 0.07).
For anti-vandalism, editors who participated in
collaborations also behaved more similarly to prototypical
members (average r=0.11) compared to editors who did not
participate (average r=0.06). Surprisingly, however, the
difference between participants and non-participants has
higher among editors who did not identify as group
members (non-self-identified editors: r=0.13 versus r=0.05)
compared to those who did (self-identified members: r=0.09
versus r=0.06). Thus we have mixed results about the
interaction effects of group identification and social
modeling in the case of vandalism reversion. One possible
explanation for the latter findings is that, reverting
vandalism, although an important citizenship behavior, is
not an activity that is strongly identified with any particular
Wikiproject. This suggests that social modeling may not be
effective for behaviors that are not specific to the group.
Additional research is needed to further understand the
Editors who did not
identify as group
members
Editors who selfidentified as group
members
Correlation
with
prototypical
group
members
The month
when
participated
in COTWs
The month
when not
participate
in COTWs
The month
when
participated
in COTWs
The month
when not
participate
in COTWs
Assessment
0.3631
0.2378
0.0759
0.0697
Antivandalism
0.0852
0.0599
0.1292
0.0525
Table 3. The correlations of the behaviors of regular editors
with the behaviors of prototypical project members.
0.4
Assessment Corelation
Assessment correlation: Each article within the scope of
a Wikiproject can have a quality rating and an importance
rating in its Wikiproject template. Assessing an article
involves adding or changing the rating of an article.
Assessing articles is an important task for Wikiproject in
order to recognize excellent contribution and identify
important topics in need of further work; there have been
over 2.1 million assessments made over the history of
Wikipedia, with most being driven by individual projects.
Similar to the anti-vandalism correlation, we measured this
variable by 1) calculating the (log transformed) number of
revisions done by the editor which change the rating of any
article within the project, and 2) calculating the correlation
of regular members with prototypical members.
Regular editors
Month not participate in
COTWs
Month participate in
COTWs
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
Non self-identified editors
Self-identified group members
Figure 4. The correlation of regular editors’ assessment with
the prototypical project members’ assessment
0.14
Anti-vandalism Correlation
vandalism” or variants such as “rvv”. We measured this
variable in two steps. First, we calculated the (log
transformed) number of revisions with anti-vandalism
comments on articles within the project done by each
editor in the given month. We then used this data to
calculate the correlation of regular members with
prototypical members. To compute this correlation, we
matched regular members with prototypical members
whom they would meet if they participated in COTWs in
that month (multiple editors can match the same
prototypical member in a given month).
0.12
0.1
Month not participate in
COTWs
Month participate in
COTWs
0.08
0.06
0.04
0.02
0
Non self-identified editors Self-identified group members
Figure 5. The correlation of regular editors’ anti-vandalism
with the prototypical project members’ anti-vandalism
mechanism of social modeling in these settings: why are
models more likely to influence other community members
on some citizenship behaviors but not others?
DISCUSSION
Lessons from Collaborations of the Week
Despite the success of Collaborations of the Week in
Wikipedia, many Wikipedia projects that successfully used
them ultimately abandoned them. In our data, only 13 of the
26 projects that started to use Collaborations of the Week
continued to use them throughout the period of our research
(as of March 2008). According to interviews with project
leaders [37] the explanation is not related to their
effectiveness but instead to the bureaucratic cost of running
them. Like any recurrent event, they need an organizer
responsible for managing the collaboration process, such as
monitoring the nomination progress and maintaining the
announcement. In addition, groups and organizers need
appropriate strategies to choose collaboration targets. These
problems suggest opportunities for computer support for
coordinating the collaboration process, such as helping to
choose collaboration targets and announcing and running
the collaboration process.
Critics might suggest that computer supported coordinated
goal setting is not as optimal as goals selected by group
organizers or voted by members. However, goal-setting
theory suggests that all these types of goal selection can be
equally effective as long as group members become
committed to the group goal [24] and furthermore some
projects have already implemented an automated topic
selection program which chooses targets from a collectively
maintained list [40].
Although Collaborations of the Week are occasions for
social interaction and modeling, their design could enhance
these attributions. For example, some Wikiprojects has
instituted temporally synchronous editing sessions for
project members to get together to work on common tasks,
with the explicit purpose of increasing social interaction.
Managerial implication
Although these results were obtained in the context of
projects within Wikipedia, we believe that the basic idea of
combining group identification and direction setting, as an
unobtrusive management method, may generalize to other
kinds of online communities and offline organizations. For
example, these ideas may work well in organizations
emphasizing creative work, such as art design or scientific
research, where strong managerial control may harm
creativity. Deadlines for major releases in many open
source software projects serve similar functions.
There may be limits to the applicability of group goal
setting, which simply highlight tasks important for the
group. If these tasks involve high coordination costs, the
benefits of adding more effort may be offset by the
difficulties of coordinating that effort; or, as Brooks aptly
states, “Adding manpower to a late software project makes
it later” [5]. However, in the cases when group goal setting
can be used, our results suggest it is remarkably powerful
and leads to benefits not only to the targeted goals but also
to other group-relevant tasks.
Compared to group goal setting, which focuses attention on
a specific set of tasks, social models may be especially
effective in drawing in peripheral members and training
them in a wide range of subtle behaviors. Therefore, we
recommend practitioners pay close attention to encouraging
the desired behaviors from core members and then
providing social opportunities (such as communication
channels and collaboration tasks) for core members to
interact with and potentially influence the others.
CONCLUSION
This paper investigated how combining group identification
with direction, either explicit direction through group goals
or implicit direction through social modeling, can motivate
volunteers in online communities to accomplish tasks
important to the success of the group. We tested our
hypotheses in the context of subgroups within Wikipedia
(Wikiprojects), examining a common group activity
(Collaborations of the Week). Our results demonstrate that
1) highlighting important group goals can have a strong
motivating influence on editors who have self-identified as
group members compared to comparable others who have
not self-identified; 2) the positive effects spill over to nongoal related tasks; and 3) editors exposed to prototypical
group members are more likely to behave similarly to those
members on group-relevant citizenship behaviors than
editors not exposed to prototypical members.
REFERENCES
1. Ashforth, B.E., Harrison,S.H., and Corley, K.G (2008)
Identification in Organizations: An Examination of Four
Fundamental Questions. Journal of Management.
2. Barker, J. (1993). Tightening the iron cage: Concertive
control in self-managing teams. Administrative Science
Quarterly, 38(3), 408-437.
3. Beenen, G., Ling, K., Wang, X., Chang, K., Frankowski,
D., Resnick, P., and Kraut, R. E. (2004). Using social
psychology to motivate contributions to online
communities. In CSCW 2004
4. Benkler, Y. (2002). Coase's Penguin, or Linux and the
Nature of the Firm. Yale Law Journal, Vol. 112.
5. Brooks, F.P. (1975). The mythical man month. AddisonWesley Longman Publishing Co., Inc. Boston, MA, USA
6. Burke, M., & Kraut, R. E. (2008). Mopping up: Modeling
Wikipedia Promotion Processes. In CSCW 2008.
7. Cheney, G., & Tompkins, P. (1987). Coming to terms with
organizational
identification
and
commitment.
Communication Studies, 38(1), 1-15.
8. Cosley, D., Frankowski, D., Terveen, L., and Riedl, J.,
2007. SuggestBot: using intelligent task routing to help
people find work in wikipedia. In Proceedings of the 12th
international conference on Intelligent user interfaces (IUI
'07).
9. Cosley, D., Frankowski, D., Terveen, L., and Riedl, J.,
2006. Using intelligent task routing and contribution
review to help communities build artifacts of lasting value.
In Proceedings of the SIGCHI conference on Human
Factors in computing systems (CHI '06)
10. Daft, R., & Lengel, R. (1984). Information richness: A new
approach to managerial behavior and organizational design.
Research in Organizational Behavior, 6, 191-233.
11. Deci, E. L., Koestner, R., & Ryan, R. M. (1999). A metaanalytic review of experiments examining the effects of
extrinsic rewards on intrinsic motivation. Psychological
Bulletin, 125(6), 627-668.
12. Feather N.T.; Rauter K.A., (2004) Organizational
citizenship behaviours in relation to job status, job
insecurity, organizational commitment and identification,
job satisfaction and work values. Journal of Occupational
and Organizational Psychology, 77(1), 81-94.
13. Forte, A., & Bruckman, A. (2008). Scaling Consensus:
Increasing Decentralization in Wikipedia Governance.
Hawaii International Conference on System Sciences,
Proceedings of the 41st Annual, 157-157.
14. Hager, M., & Brudney, J. (2004). Volunteer management
practices and retention of volunteers. The Urban Institute
15. Hogg, M. A., & Terry, D. J. (2000). Social identity and
self-categorization processes in organizational contexts.
The Academy of Management Review, 25(1), 121-14.
16. Kennedy, P. (2001). A guide to econometrics (4th ed.).
Cambridge, MA: MIT Press.
17. Kriplean, T., Beschastnikh, I., and McDonald, D.W.(2008).
Articulations of wikiwork: uncovering valued work in
wikipedia through barnstars. In Proceedings of the 2008
ACM conference on Computer supported cooperative
work (CSCW '08).
18. Kernis, M., Zuckerman, M., Cohen, A., & Spadafora, S.
(1982). Persistence following failure: The interactive role
of self-awareness and the attributional basis for negative
expectancies. Journal of Personality and Social
Psychology, 43(6), 1184-1191.
19. Kittur, A., Pendleton, B., & Kraut, R. (2009). Herding the
cats: The influence of groups in coordinating peer
production In WikiSym 2009.
20. Kraut, R. E. & Resnick, P. (Under contract). Evidencebased social design: Mining the social sciences to build
online communities. Cambridge, MA: MIT Press.
21. Locke, E. A., & Bryan, J. (1969). The directing function of
goals in task performance. Organizational Behavior and
Human Performance, 4, 35– 42.
22. Locke, E. A., & Latham, G. P. (1990). A theory of goal
setting and task performance. Englewood Cliffes, NJ:
Prentice-Hall.
23. Locke, E. A., & Latham, G. P. (2002) Building a
Practically Useful Theory of Goal Setting and Task
Motivation American Psychologist, Vol. 57, No. 9, 705–
717
24. O'Leary-Kelly, A., Martocchio, J., & Frink, D. (1994). A
review of the influence of group goals on group
performance. Academy of Management Journal, 37(5),
1285-1301
25. Organ, D. W., & Ryan, K. (1995). A meta-analytic review
of attitudinal and dispositional predictors of organizational
citizenship behavior. Personnel Psychology, 48(4), 775802.
26. Pearce, J.P., (1993) Volunteers: the organizational behavior
of unpaid workers, Published by Routledge, 11 New Fetter
Lane, London
27. Preece, J. and B. Shneiderman (2009) “The Reader-toLeader Framework: Motivating Technology-Mediated
Social Participation,” AIS Transactions on HumanComputer Interaction, (1) 1, pp. 13-32
28. Quintela, Y., & Donovan, J. (2008). A model of
motivational spillover. In K. H. Kiefer (Ed.), Applied
Psychology Research Trends. New York: Nova Science,
141-160.
29. Roberts, J. A., Hann, I.-H., & Slaughter, S. A. (2006).
Understanding the Motivations, Participation, and
Performance of Open Source Software Developers: A
Longitudinal Study of the Apache Projects. Management
Science, 52(7), 984-999.
30. Rothkopf, E., & Billington, M. (1979). Goal-guided
learning from text: Inferring a descriptive processing model
from inspection times and eye movements. Journal of
Educational Psychology, 71, 310 –327.
31. Smith, C. A., Organ, D. W., Near, J. P. (1983)
Organizational citizenship behavior: Its nature and
antecedents. Journal of Applied Psychology, Vol 68(4),
Nov 1983
32. Tajfel, H. 1972. Social categorization (English translation
of "La cat6gorisation sociale"). In S. Moscovici (Ed.),
Introduction a la psychologie sociale, vol. 1: 272-302.
Paris: Larousse.
33. Tajfel, H., Billig, M. G., Bundy, R. P., Flament,C., 1971
Social categorization and intergroup behaviour, European
Journal of Social Psychology, Volume 1, Issue 2.
34. Tajfel, H., & Turner, J. C. 1979. An integrative theory of
intergroup conflict. In W. G. Austin & S. Worchel (Eds.),
The social psychology of intergroup relations: 33-47.
Monterey, CA: Brooks-Cole.
35. Turner, J. C. 1985. Social categorization and the selfconcept: A social cognitive theory of group behavior. In E.
J. Lawler (Ed.), Advances in group processes: Theory and
research, vol. 2: 77-122. Greenwich, CT: JAI Press.
36. Vandalism.
http://en.wikipedia.org/wiki/Wikipedia:Vandalism
37. Wikipedia:Wikipedia Signpost/Archives/WikiProject
report.
http://en.wikipedia.org/wiki/Wikipedia:Wikipedia_Signpos
t/Archives/WikiProject_report
38. Wikipedia:WikiProject_Anime_and_manga/Collaborati
on_of_the_Week
http://en.wikipedia.org/wiki/Wikipedia:WikiProject_Ani
me_and_manga/Collaboration_of_the_Week
39. Wikipedia:WikiProject_Council/Guide
http://en.wikipedia.org/wiki/Wikipedia:Wikiproject_Co
uncil/Guide
40. Wikipedia:WikiProject_Video_games/Collaboration_of
_the_week
http://en.wikipedia.org/wiki/Wikipedia:WikiProject_Vid
eo_games/Collaboration_of_the_week