Received September 7, 2019, accepted September 18, 2019,

date of publication September 23, 2019, date of current version October 3, 2019.
Digital Object Identifier 10.1109/ACCESS.2019.2942998

The Effect of Employee Participation in Enterprise

Social Media on Their Job Performance
YINGJIE LU AND TAOTAO PAN
School of Economics and Management, Beijing University of Chemical Technology, Beijing 100029, China
Corresponding author: Yingjie Lu (lvyj@mail.buct.edu.cn)
This work was supported in part by the National Natural Science Foundation of China under Grant 71401005 and Grant 71571010, and in
part by the Funds for First-Class Discipline Construction under Grant XK1802-5.

ABSTRACT This study focuses on employee involvement in enterprise social media (ESM) and the impact
of ESM on job performance. Few studies found empirical support for this perspective due to limited
sample size either the difficulty of accessing the data on user behavior in ESM and their job performance.
We addressed this research gap through a data-sharing agreement with a large, high-tech manufacturing
enterprise with internal social media and conducted a cooperative study on the impact of employee behaviors
involved in ESM on their job performance. We divided online activities on ESM into information-sharing
behaviors, information-seeking behaviors and general usage behaviors and then employed hierarchical
regression analysis to investigate how the various usage behaviors of employees on ESM affect their job
performance significantly. We find that information-seeking behaviors on ESM have positive effects on job
performance, but information-sharing behaviors on ESM have negative effects on job performance. Overall,
the usage of social media within enterprise social media use could help improve employee job performance.
Moreover, to test whether the timing of ESM usage (during working hours or off-work hours) may affect
employee job performance, we add a variety of moderator variables into the model. The results show that
information-sharing behaviors, information-seeking behaviors and general usage behaviors during working
hours negatively moderate the relationship between ESM usage and job performance. The findings have
valuable managerial implications for the use of ESM.

INDEX TERMS Enterprise social media, information sharing, information seeking, job performance.

I. INTRODUCTION which was called enterprise social media (ESM). ESM was
With the rapid development of Web 2.0 technologies, a grow- put forward by Leonardi and defined as ‘‘Web-based plat-
ing number of organizations have adopted Web 2.0 applica- forms that allow workers to (1) communicate messages with
tions in the workplace. Some popular social networking sites, specific coworkers in the organization; (2) explicitly indicate
such as Facebook and Twitter, have been routinely employed or implicitly reveal coworkers as communication partners;
by many organizations for organizational external communi- (3) post, edit, and sort text and files linked to them-
cation with partners, including their customers and suppliers, selves or others; and (4) view the messages, connections,
and for internal employee communication and information text, and files communicated, posted, edited, and sorted by
sharing. However, these publicly available social networking anyone else in the organization at any time of their choosing.’’
sites pose apparent risks that may result in serious negative ESM includes corporate blogging, microblogging, and enter-
consequences, including loss of confidential data, misman- prise social network sites to provide employees the abil-
aged and misplaced business records, exposure of company ity to post and share information [3]. ESM platforms run
secrets, and security breaches [1]. on the host organization’s servers, are protected by fire-
Consequently, some organizations implement an integrated walls and are restricted to use by employees due to security
social media platform within the enterprise for internal concerns [4].
knowledge sharing and collaboration among employees [2], Some well-known companies have launched internal
deployments of integrated enterprise social software services
The associate editor coordinating the review of this manuscript and to encourage their employees to exchange information with
approving it for publication was Chunsheng Zhu . each other on the corporate intranet [5]. Examples include

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see http://creativecommons.org/licenses/by/4.0/
137528 VOLUME 7, 2019
Y. Lu, T. Pan: Effect of Employee Participation in ESM on Their Job Performance

Microsoft Yammer, Salesforce Chatter, and IBM’s Con- ESM could promote employees’ interactivity with colleagues
nections, etc. Some companies even developed their own to strengthen mutual trust [20] and work cooperation [21],
ESM platforms, such as the Beehive system developed at which helps improve work performance. However, other
IBM [6], [7] and the Watercooler system developed at HP [8]. research [22] has reported that ESM can also aggravate the
In addition to those multinational enterprises in developed adverse effects of workplace conflict because conflicts pub-
countries, many small and medium-sized enterprises have licized by ESM can deepen misunderstandings between rel-
attempted to use ESM to foster internal collaboration, com- evant parties. These contradictory conclusions maybe result
munication and knowledge interaction among employees, in difficulties in making organizational policies regarding the
especially in some emerging market countries such as China. use of ESM.
For example, in recent years, many Chinese companies tend While many studies have been done about the impact
to use ESM for knowledge management and work coop- of ESM, few studies have found empirical evidence on
eration. A well-known ESM application, DingTalk, a free how employee participation in ESM affects their job perfor-
and all-in-one enterprise communication and collaboration mance. In our studies, we address this research gap through
platform developed by Alibaba Group in 2015, has already a data-sharing agreement with a large, high-tech manufac-
been employed at over 7 million business and organization turing enterprise with internal social media and conducted
clients [9]. Another leading enterprise social software plat- a cooperative study on the impact of employee behaviors
form, Mingdao, was officially launched in January 2014 and involved in ESM on their job performance. We want to know
has become a very popular SaaS product for enterprise inter- whether employee online behaviors significantly affect their
nal collaboration and knowledge management applications in job performance and how enterprise managers should manage
China [10]. The software has been installed by millions of their ESM. Thus, in our study, we try to investigate the
users, more than 4,000 of whom are paying customers. following research questions:
However, to date, ESM has not been widely used in most RQ1: Will employee participation in ESM affect their
organizations, even though Web 2.0 technologies are readily job performance? Which kinds of employee online behav-
available and require little additional investment. One key iors positively influence job performance? Which kinds
factor that prevents widespread adoption of ESM is the lack of employee online behaviors negatively influence job
of a clear understanding of ESM. Many organizations simply performance?
believe that ESM will bring potential advantages, but they had RQ2: How could enterprise managers implement practical
no idea how to effectively utilize ESM to achieve business policies toward the use of ESM to help employees improve
benefits. A global survey found that one-third of organiza- their job performance?
tions reported that they had not seen concrete benefits from
employees actively using these ESM tools to perform their II. LITERATURE REVIEW
job, despite the companies’ widespread deployment of ESM A. EMPLOYEE BEHAVIORS AND MOTIVATIONS IN ESM
in their organizations [11], [12]. Previous studies argued that ESM is a well-liked channel for
Employee involvement in ESM and the impact of ESM employees to obtain information, express views and interact
on job performance have recently become hot academic with others within the organization. They focused on the
issues. ESM usage was generally considered a double-edged analysis of employee behavior patterns of ESM usage and
sword [13], [14]. Some previous studies [15] consider ESM their motivations for participating in ESM. For example,
as a time-waster that will lead to increased interruption and DiMicco et al. [6] conducted qualitative interviews with
decreased productivity. They noted that ESM has become employees involved in a social network site deployed within
an important platform for leisure and entertainment for an IBM intranet and found what motivates employees to use
many employees, where employees discuss interesting top- the site included connecting on a personal level with cowork-
ics they are concerned about. They worried that employees ers, assisting them in their personal career advancement and
might spend much time engaging in ESM doing non-work- convincing others to support their ideas and projects.
related activities such as browsing hot topics, posting to Information-seeking behavior and information-sharing
express their opinions and chatting with other users during behavior are widely considered two distinct types of behav-
working hours, resulting in a decrease in their productiv- iors in a social network environment [23], and different
ity. However, most studies [16], [17] express the opposite behaviors were driven by several different incentives [24].
opinion that ESM could provide more efficient communi- Employees could draw on the knowledge of others by search-
cation channels and convenient information-sharing services ing the work-related information posted by other users, which
where employees can obtain work-related information and helped them become more professionally productive. Brows-
share their work experiences. Evidence has suggested that ing some valuable posts when they encounter obstacles at
ESM have potential for knowledge sharing and problem work often brings employees new ideas [25]. Moreover, many
solving by strengthening communication and collaboration employees prefer to seek some leisure-related information
among employees [18], [19]. on ESM, such as vacation pictures, entertainment gossip, and
In addition, ESM can be used to maintain rela- hot social issues. Although this non-work-related information
tionships with colleagues. Some scholars proposed that might not help the employees improve their productivity,

VOLUME 7, 2019 137529

 Y. Lu, T. Pan: Effect of Employee Participation in ESM on Their Job Performance

it could satisfy employees’ interests and help them relieve friends or colleagues who are in their social media network
stress. Some studies supported the proposition that ESM is a more convenient and easier [41]. ESM could serve as a com-
good platform to relax for employees with a heavier workload munication channel where explicit and implicit knowledge
and greater pressure [26], [27]. can be effectively transferred among employees. There is evi-
Knowledge contribution behaviors in ESM primarily occur dence suggesting that employees who use social networking
when users are motivated to review the questions posted, sites are 9% more productive than those who do not [36].
choose those they are able and willing to answer, and take Another independent survey of 2,500 employees conducted
the time and effort to formulate and post a response [28]. The in five European countries has shown that 65% of employ-
determinants of information-sharing behaviors are different ees think ESM make them more efficient workers, 65% of
from those of information-seeking behaviors. Information employees think ESM give them more creativity and ideas,
contributors voluntarily help strangers through ESM. They and 38% of employees think ESM help them gain knowledge
have no assurances that those they are helping will ever and create solutions to problems. Employees also reported
return the favor, since lurkers may draw upon the knowledge that social networking has become part of the culture of their
of others without contributing anything in return. There- workplace.
fore, it seems irrational that they contributed much time In addition, some studies considered that social media
and effort without immediate benefits [29], [30]. To under- use in non-work-related activities could help improve job
stand this paradox, some studies focused on the motivations performance because personal surfing and online chatting
for information-sharing behaviors in ESM. According to on social media can serve as one solution to help employ-
Pi et al. [31], there are several different factors affect- ees relax and take breaks from their stressful work sur-
ing the intention and attitude toward knowledge sharing roundings. There is evidence that the employees who are
in ESM, including reputation, expected relationship, sense involved with highly demanding jobs are more likely to
of self-worth and subjective norm. Many users would like access social media compared to those in less stressful posi-
to contribute their knowledge when they perceive that doing tions because employees who experience a high degree of job
so enhances their professional reputations and they have stress need some type of support to help them deal effectively
the experience to share [32]. Additionally, they acquire an with external stressors [41]. Another similar study reported
increased sense of self-worth, which is thought to motivate that 70% of 300 employees think personal surfing on social
them to continuously share their knowledge [33]. In addition, media enables their brain to be refreshed and renewed, lead-
some users hope to create and strengthen ties with colleagues ing to a higher total concentration on their work and, as a
through knowledge sharing [33], [34]. Some studies also result, increasing job productivity [36].
showed that the subjective norm is a significant influencing Moreover, different from the views that ESM is a time-
factor encouraging knowledge-sharing behaviors [31]. waster because it is an online communication platform for
personal business driven by private demands, further anal-
B. ESM AND EMPLOYEE PERFORMANCE ysis found that social media, as a means by which they
With its increasing popularity in a few corporate settings, can maintain personal connections with family, friends, and
there is considerable research related to the use of social colleagues, could enhance employee satisfaction, which is
media in the workplace and its effects on job performance. positively associated with job performance [42]. Good rela-
Some previous studies consider ESM as a time-waster [35]. tionships among colleagues make employees feel connected
According to a survey report from an IT research company to one another, which in turn motivates them to communicate
allowing access to social networking sites in the workplace, and interact more often. In addition, the ability to maintain
77% of employees using social media at work resulted in a personal connections outside the enterprise can be a crucial
1.5% decrease in total employee productivity [36]. There is factor that allows employees to balance their work and life
evidence that most employees use ESM as a tool to socialize activities more effectively. Employees who are fully satisfied
with their friends and families rather than discuss work- in their personal lives are more able to concentrate on their
related issues with colleagues [37], [38]. Therefore, the usage work, which will subsequently help them achieve higher
of ESM inside companies is blamed for reducing the produc- levels of performance. Therefore, ESM as a source of social
tivity of employees due to the addictive and excessive use support will help improve employee performance through the
of ESM in non-work-related activities. Accordingly, some mediating effect of job satisfaction.
studies also reported that many employers consider banning
the use of social media during working hours [39] because C. SUMMARY
they worry that staff is wasting time on websites during Although an increasing number of studies have begun to
working hours and weakening productivity [40]. explore ESM, the focus has mainly been on how to evaluate
While some studies take a more negative view on the its value for the organization in terms of internal communi-
impact of ESM on job performance, several recent studies cation and knowledge management at the macro level [43].
have shown just the opposite. They argued that social media These studies have provided few insights into the effects of
use at work has a positive impact on job performance. Many ESM on individuals at the micro level. There seems to be
employees found that ESM makes obtaining advice from a very limited understanding of personal use of ESM and

137530 VOLUME 7, 2019

Y. Lu, T. Pan: Effect of Employee Participation in ESM on Their Job Performance

its impact on job performance [44]. Empirical studies are much valuable work-related information and practical experi-
still somewhat scarce because of limited sample size either ence of others through ESM platforms. They also could try to
the difficulty of accessing the data on user behavior in ESM seek advice and assistance information to create solutions to
and their job performance. Therefore, empirical analysis is problems they encountered in their work. In addition, some
needed to provide further evidence for the important value employees pointed out that ESM could help them think of
of ESM within enterprises, since this proposition remains an innovative ideas for their work. This valuable information
issue under investigation. and knowledge obtained from ESM make employees become
more productive and professional, which leads to better job
III. RESEARCH HYPOTHESES performance.
Earlier studies suggested that the usage of social media within Moreover, some employees prefer to browse entertainment
enterprises could be blamed for employees spending too information or read gossip through ESM that has nothing
much time on online activities unrelated to the workplace, to do with their job. It seems that information seeking with
resulting in a decrease in their job performance. The pre- hedonic motivation is a distraction and interruption to work
conceived notion that ESM is considered time-wasters had and apparently has a negative impact on job performance.
a sizeable impact on employers’ willingness and desire to Several studies, however, argued that these information-
take advantage of social media in the workplace. Some recent seeking behaviors unrelated to their job will help employees
studies, however, have provided different opinions regard- achieve higher job performance. The possible explanation
ing ESM. For example, a survey was done on the opinion is that accessing ESM could allow employees to relax and
that using ESM lowered employee job performance, and the take breaks from their stressful work surroundings and there-
results indicated that the same percentages of respondents fore help increase their job satisfaction. Employees with a
agreed and disagreed with the notion [37]. very high workload in particular can achieve a work-life
Employees involved in ESM engage in different types balance by seeking non-work-related recreation or leisure
of online activities due to different motivations, which may topics. When employees are happy with their jobs, they are
engender a complicated influence on their job performance. more likely to enjoy working, thereby indirectly potentially
Information sharing and information seeking are thought to increasing their job performance. Thus, we propose the fol-
be two types of distinct user online activities. Information- lowing hypothesis:
sharing behaviors on ESM constitute online activities for the Hypothesis 2: The employees who engage in information-
purpose of contributing information, such as posting some seeking activities on ESM would achieve higher performance
messages or viewing other messages. Previous studies found than other employees.
that there are multiple factors affecting the attitude of employ- In addition to the above two types of usage behaviors,
ees towards information sharing on ESM. Employees could employees could benefit from involvement in ESM in many
receive much attention by sharing much work information. other ways, such as building individual social network rela-
Some of them are even identified as experts in a certain tionships. Employee long-term involvement in ESM could
domain by continually sharing their work experience or offer- make them feel more connected within an organization and
ing useful advice to others. Accumulated reputation from more willing to interact with others, discuss novel ideas and
information-sharing behaviors could bring them certain indi- work together in problem solving. In addition, as a commu-
rect benefits, such as becoming known to senior employees nication channel, ESM provides a platform for employees to
and upper management, thereby potentially increasing their communicate non-work-related information and helps build
job performance. In addition to professional reputation, some informal personal relationships. Employee involvement in
intrinsic benefits are considered to motivate employees to ESM makes it more convenient and easier for them to interact
contribute their knowledge, such as self-esteem and feelings with others and perceive more emotional support. Accord-
of commitment to their organization. Some employees per- ing to social capital theory, these positive outcomes through
ceive that helping others with challenging problems is inter- involvement in ESM could have a positive impact on their job
esting work and makes them gain higher self-satisfaction. performance. Thus, we propose the following hypothesis:
Some also consider they have an obligation to help their Hypothesis 3: The employees who participate more in ESM
colleges solve difficulties; however, they also acquired much would achieve higher performance than other employees.
knowledge and improve their problem-solving skills, which Regardless of their motivations for involvement in ESM,
may positively affect their job performance. Thus, we propose employees will inevitably devote much time and effort to
the following hypothesis: online activities. A survey of 237 employees in an IT research
Hypothesis 1: The employees who engage in information- company found that 77% of respondents who were allowed
sharing activities on ESM would achieve higher performance to access social media during working hours lose an average
than other employees. of 1.5% employee productivity, indicating that working fewer
Information-seeking behaviors on ESM refers to online hours would have an obvious negative effect on job perfor-
activities for the purpose of obtaining information, such as mance. Nevertheless, we tried to find the best approaches
searching for valuable knowledge or browsing specific online being used to effectively utilize ESM to help employees
communities to find topics of interest. Employees could gain improve their job performance while avoiding unfavorable

VOLUME 7, 2019 137531

 Y. Lu, T. Pan: Effect of Employee Participation in ESM on Their Job Performance

consequences. An alternative solution to the problem of ESM For many years, the company has made great efforts to
usage is to only allow employees to participate in ESM in promote information communication and knowledge sharing
their spare time or off-work hours instead of during work- within the enterprise. The managers thus attempted to make
ing hours, since employee online activities on ESM during full use of social media for communication and coordination
working hours significantly affect their job performance. The among the employees. However, with the popularity of the
choice to limit access to ESM is based on the condition online forum among the employees, management has serious
that employees who were more likely to engage in vari- doubts concerning whether the social media platform could
ous types of online activities during working hours would improve employee performance and, if so, to what extent.
achieve lower performance than others would without usage Therefore, we conducted a cooperative study to explore the
preference. Thus, we propose the following three hypotheses: effect of employees participating in ESM on their job perfor-
Hypothesis 4-a: The employees who are more willing mance in the context of this company.
to share information on ESM during working hours would In our study, one of the challenges is to establish a data
achieve lower performance than those without preference for collection approach that is appropriate to measure employee
the timing of information-sharing activities. engagement in ESM. The most common data collection
Hypothesis 4-b: The employees who are more willing methods currently used in social media research include
to seek information on ESM during working hours would surveys, interviews, experiments, online data collection and
achieve lower performance than those without preference for so on. In recent studies about social media, some studies
the timing of information-seeking activities. use employees’ online behavior data instead of their self-
Hypothesis 4-c: The employees who are more willing to use report survey data to measure the employee engagement in
ESM during working hours would achieve lower performance enterprise internal social media [45]–[48]. Considering that
than those without preference for the timing of usage. we aim to assess three types of online behaviors, including
information-sharing behaviors, information-seeking behav-
IV. METHODS iors and general usage behaviors, we decided to use online
A. RESEARCH CONTEXT AND DATA COLLECTION behavioral data on users’ online traces of activities and
The empirical study was conducted in a large state-owned interactions in the ESM platform. There are three reasons
high-tech spacecraft equipment manufacturing company with for this. First, methodologically, we wish to capture actual
over 10,000 employees in China. The company has become behaviors of employee engagement in ESM, not the employ-
one of the leading international spacecraft designers and ees’ attitudes toward and perceptions of their engagement
manufacturers over the last four decades, providing fully in ESM. Thus, online behavioral data are more appropriate
integrated solutions of all kinds of spacecraft equipment for our study compared to data from self-report methods such
and systems for global customers. A social media platform, as surveys or interviews. Second, studies [49] have found
that is, an online discussion forum, was deployed within the a large disparity between self-report online behaviors and
company in 2005. The online forum is restricted to the actual usage. Often, users significantly underestimate their
employees of the company and supports interpersonal com- online engagement because they may have trouble accurately
munication, coordination and collaboration inside the com- accounting for the full amount of time spent online. Our
pany. Any employee could have access to the online forum data collection approach can solve the problem and make the
on the company intranet to browse and search the messages results more accurate. Finally, sampling techniques used in
posted by other users. Only the registered users are allowed the surveys or interviews limit generalizability of findings
to initiate a new topic or reply to other posts in a discussion due to their small sample size. In contrast, the large amount of
thread after logging in. All the messages posted by users are behavioral data created by online users makes the experiment
accessible to all the employees of the company and are stored results more credible and convincing.
in a database on the internal server. In our study, the available data for empirical analysis
The online forum was classified into 66 discussion boards were collected from two sources: online employee behavior
according to their topics, where users could have access to all information stored in the internal ESM database and offline
kinds of information that they were interested in, including employee information from the human resource management
work-related knowledge, problem-solving experiences, per- system.
sonal life trivia, entertainment, leisure and so on. Because On one hand, the ESM database contains detailed informa-
little restriction was imposed on the use of the online forum tion about employee posts, where each post is uniquely iden-
by the company, employees could engage in all kinds of tified. The additional necessary attributes were also stored,
online activities, including information seeking, information including author, post title, post body content, thread to which
sharing and communicating with others at any time. After the post belongs, and timestamp, indicating the date and
over ten years of development, the site has 5,640 registered time it was posted. However, it is difficult to collect data on
users who generated a total of 682,036 posts. Currently, more employee information-seeking activities, as the system just
than half of the employees in the company have been involved contains an entry for every access to the sites of registered
in the online forum, indicating that it has become an important users in the server log file and was not used to keep track
channel of communication for the company. of user online behaviors. To solve this problem, we were

137532 VOLUME 7, 2019

Y. Lu, T. Pan: Effect of Employee Participation in ESM on Their Job Performance

TABLE 1. A summary description of all variables.

authorized to make the secondary development on the sites we used two approaches to ensure the participating employ-
to obtain the necessary information of all the user online ees were fully informed about what the research involved and
activities by tracking and recording every user click on the its risks and benefits. On one hand, an official notification
sites. Specifically, we did not use registered user ID as the elaborating on the purpose of the research and how the data
user identifier because many users were more likely to anony- will be used was sent by email to each participating employee,
mously login in the sites as guests when they had no intention and asked the recipient to reply via email to confirm that
of posting. Instead, we used the IP address to identify the it had been read. On the other hand, an informed consent
users, as the office computer of each employee was typically page with a ‘‘click to accept’’ button was sent to ESM users
associated with a unique IP address. through the messaging system in the ESM platform, allowing
On the other hand, additional offline employee information participants to click the button to express their understanding
could be obtained from the human resources management and agreement on our research.
system, including employee demographics (gender, age, and We obtained personal information on 5,540 employees
education background, etc.), position within the company, job from the list provided by the human resources department,
performance and so on. Note that the human resources depart- among whom 3,689 employees were involved in ESM,
ment only provided basic information from three years earlier accounting for 66.58% of all employees. After deleting the
due to the sensitivity of employee performance data, but the employee samples with incomplete observations or missing
available data are significant enough for our empirical anal- values, we used additional processing to make connections
ysis. In addition, the human resources department removed between online behavior information and offline informa-
sensitive personal information and provided anonymized tion on employees by their unique identities and allocated
employee profile data for reasons of privacy. IP addresses, yielding a dataset consisting of 2,606 observa-
In addition, we think the requirements for informed con- tions from 1,432 employees for the empirical analysis.
sent must be considered in our study to reduce potential
ethical risks. However, it is much more difficult to obtain B. VARIABLES DESCRIPTION
written informed consent online than in traditional settings. Table 1 provides a summary description of all variables.
A possible solution for this difficulty was that we could waive Job performance, as the dependent variable in our empirical
the signed document of informed consent and take alternative analysis, should be measured in terms of the results produced
approaches, such as allowing a procedure in which human by employees. Previous research has relied on merit pay as
subjects click a button on an online form to indicate they a measure of worker outcome. However, the sample in our
have read and understood the consent form. In our research, study consisted of different kinds of employees based on their

VOLUME 7, 2019 137533

 Y. Lu, T. Pan: Effect of Employee Participation in ESM on Their Job Performance

job positions, including production workers, process workers, In addition to the aforementioned independent variables,
R&D engineers, and administrative workers and managers. there are other important factors that may affect employee job
Compared to the manual workers, whose performance relies performance. Traditionally, employees’ position and tenure
on their total output or output per hour based on the quantity in the company are thought to have significantly influ-
of goods and services produced per hour of labor input, enced their job performance in many studies. Moreover,
it is impractical to measure the output of knowledge workers some previous studies have considered that employees’ basic
and managers, as knowledge work is not easily observ- demographic characteristics, including age, gender, educa-
able or measurable and therefore is more difficult to evalu- tion background and so on, have an influence on their
ate. Thus, many studies adopt survey questionnaire methods job performance. Therefore, in our research, the following
to measure job performance by constructing a self-report four variables—age, gender, education, and tenure—were
questionnaire. However, measurement of job performance employed as control variables. Age was measured as a con-
remains a major challenge due to the poor reliability and tinuous variable. Gender was measured as a dummy variable
validity of measures used on self-assessment questionnaires. with 1 for male and 0 for female. Education was measured
Considering the measurement should be feasible, reliable, using a four-level (1 through 4) ordinal scale (1 = High
valid and easy-to-obtain, an annual performance evaluation School degree or less; 2 = Bachelor’s degree; 3 = Master’s
of each employee conducted by the company was used in our degree; 4 = Doctoral degree or above). Job tenure was mea-
study to measure job performance. The outcomes were rated sured as a continuous variable representing the number of
as excellent, good, average, fair and poor. The performance years that the employee had worked for the company.
rating was determined through a deliberative process involv- In addition to the above factors, we try to understand
ing employee’s peers (colleagues), and supervisors based on whether the timing of ESM usage may affect employee job
the accomplishment of duties and tasks that were specified performance. Many researchers believed that wasting the
in their job descriptions. The annual evaluation provided an daily working hours of employees on ESM will cause lower
objective and comprehensive evaluation of job performance worker productivity. However, a few studies found empirical
by soliciting feedback from a variety of points of view and evidence to support the argument that employees who were
was therefore used as the measure of job performance. more likely to engage in online activities during working
The independent variables in our empirical model are all hours would achieve lower performance than others did with-
measures of different usage patterns of ESM from the online out preference for the timing of usage. A variety of moderator
discussion forum in the company. As mentioned above, infor- variables have been identified in our study to examine the
mation seeking and information sharing on ESM are thought moderating effect of the timing of ESM usage on the rela-
to be two distinct online behaviors that could be measured tionship of ESM usage and job performance, as follows.
using different independent variables. Considering that the (1) To test the hypothesis 4-a that the employees who were
information-sharing behaviors on ESM mainly include post- more willing to engage in information-sharing activities on
ing new messages and replying to previous messages, we used ESM during working hours would achieve lower performance
the variable Share to represent the number of posts written by than would the general population, we identify the moderator
one user to measure information-sharing behaviors. variable Share_working to represent the proportion of work-
In contrast, it remains a major challenge in measuring ing time to the total time spent in information-sharing activ-
the information-seeking behavior because it is difficult to ities. Share_working was measured using the percentage of
observe and track actual seeking activities on online forum the number of posts written by the employee during working
sites. A valid proxy measure of information-seeking behavior hours.
is deemed necessary. Considering that users have to click (2) To test the hypothesis 4-b that the employees who were
a post’s subject with an URL that links to one discussion more willing to engage in information-seeking activities on
thread to see the information that interests them, it may be ESM during working hours would achieve lower performance
inferred that the clicking behavior on the post means user than would the general population, we identify the moder-
seeking the information on the post. Hence, the number ator variable Seek_working to represent the proportion of
of clicks on posts’ subjects, represented as variable Seek, working time to the total time spent in information-seeking
might be taken as a proxy measure of information-seeking activities. Seek_working was measured using the percentage
behavior. of the number of post subjects clicked by the employee during
Similarly, for the measure of user involvement in ESM, working hours.
we could not directly observe when the employees were (3) To test the hypothesis 4-c that the employees who
using ESM. However, it seems inevitable for users to click were more likely to use ESM during working hours would
on a button or a link on the sites no matter what online achieve lower performance than would the general popu-
activities the users are engaged in. We thus capture and record lation, we identify the moderator variable Use_working to
all users’ clicking behaviors used to request new pages from represent the proportion of working time to the total time
the server. The number of total click activities, represented as spent in using ESM. Use_working was measured using the
variable Use, might be taken as a valid proxy measure of the percentage of the number of total clicks by the employee
level of involvement in the online forum sites. during working hours.

137534 VOLUME 7, 2019

Y. Lu, T. Pan: Effect of Employee Participation in ESM on Their Job Performance

TABLE 2. Descriptive statistics of employee demographics. TABLE 3. Descriptive statistics of the independent variables.

TABLE 4. Descriptive statistics of employee performance rating.

of which 22.49 (74.66%) posts were written during work-

ing hours. The average number of posts read by one user
was 1461.92, of which 1044.33 (71.43%) posts were read
C. ANALYSIS METHOD during working hours. The average number of uses of ESM
In this research, we employed hierarchical regression analy- by one user was 4123.05, of which 3005.95 (72.91%) times
ses to investigate how the various usage behaviors of employ- occurred during working hours.
ees on ESM affect their job performance significantly. To test Table 4 shows the distribution of performance rating on
all hypotheses, we built various models with performance as a five-point scale: poor, fair, average, good, and excellent.
the dependent variable based on different sets of independent Generally, employee’s peers and supervisors tend not to give
variables. a poor performance evaluation. Only 2 employees in the
In the first stage, model 1 contains only four control vari- company received the lowest rating, and they are not in the
ables (Age, Gender, Education, and Tenure) as a benchmark sample. Similarly, only 3 employees in the sample are rated
for the following models. In the second stage, the indepen- ‘‘fair’’. In addition, a vast majority of the employees received
dent variables (Share, Seek, and Use) were added to the an average or above performance rating; 359 employees
regression models to test the effect of different usage patterns (13.78%) received an ‘‘excellent’’ performance rating.
of ESM on employee job performance (hypotheses 1-3).
In the third stage, the moderator variables (Share_working, B. CORRELATION ANALYSIS
Seek_working, and Use_working) were added to the regres- Table 5 presents the correlation matrix for all the measured
sion models to test the moderating effect of the timing of ESM variables, containing the dependent variable, the independent
usage on the relationship of ESM usage and job performance. variables, and the control variables. As shown in Table 5,
Tenure is highly, positively correlated with Age. It is rea-
V. RESULTS sonably well understood that only the older employees could
A. SUMMARY STATISTICS possibly have worked for more years in the company. In addi-
Table 2 presents the summary statistics of employee demo- tion, Seek and Use are highly correlated in the analysis of the
graphics used in the study. The employee sample was com- correlations among independent variables. The reason might
posed of 1,008 males (70.39%) and 424 females (29.61%). be that we used similar proxy measures for Seek and Use;
Their mean age was 33.6 years old, and a majority of the we used the number of clicks on post subjects to measure
employees (68.92%) were between 26 and 35 years old. Seek and the number of total clicks to measure Use. The
For educational level, 83 held a high school degree or less two proxy measures have a similar nature, as they are based
(5.80%), 361 held bachelor’s degrees (25.21%), 759 held on click activities. The total click activities will inevitably
master’s degrees (53.00%), and 229 held doctoral degrees increase with the number of clicks on post subjects, resulting
(15.99%), showing that employees with high levels of edu- in the highly positive correlation between Seek and Use. The
cation constitute a majority of the population. The employee correlations among other variables are not high, indicating no
sample has an average of 10.6 years of working experience in problem of high multicollinearity.
the company and has a balanced distribution of employees in In the correlation between the dependent variable and other
different job tenure groups. variables, the findings in Table 5 indicate that all control
Descriptive statistics of the independent variables measur- variables except education are significantly related to job
ing employee online behaviors are shown in Table 3. The performance and that all three independent variables are also
average number of posts written by one user was 30.12, significantly related to job performance.

VOLUME 7, 2019 137535

 Y. Lu, T. Pan: Effect of Employee Participation in ESM on Their Job Performance

TABLE 5. Correlation matrix for the variables.

C. RESULTS models, illustrating the robustness of the findings. Model 5

We conducted classical assumption tests used in the analysis added usage behavior of ESM Use, and the results show
of multiple linear regression econometric models, including that Use is positively associated with job performance; thus
tests for multicollinearity, heteroscedasticity, and endogene- Hypothesis 3 is supported.
ity. First, the variance inflation factors (VIF) of all coef- Finally, to examine whether the timing of ESM usage
ficient estimates in the model for the multicollinearity test might moderate the relationship between ESM usage and job
are between 1.81 and 6.36, well below the common cutoff performance, a variety of moderator variables were added to
of 10, indicating that multicollinearity does not contaminate the models. As shown in Table 6, model 6 added the mod-
the results. Second, the results of the heteroscedastic- erating variable Share_working and the interaction between
ity test show that there is significant heteroscedasticity Share_working and Share. The results show that the coef-
(Prob>chi2=0.0000) in the original models. Hence, robust ficient of Share is positive and significant which has the
standard error estimates for the model parameters are opposite sign compared with the Share in model 2, and the
employed in our study to make the models more robust. coefficient of interaction is negative and significant. These
Finally, we adopted the instrumental variable method to deal results indicate that employee information-sharing activities
with a possible endogeneity problem, and the results show on ESM during working hours have a significantly negative
there is no endogeneity problem in the models. effect on their job performance. As the proportion of work-
Table 6 presents the results of the hierarchical regres- ing time to total time spent in information-sharing activities
sion analysis. As the benchmark, Model 1 only contains the increases, the positive effect of online information-sharing
control variables. The results show that all four control vari- activities on job performance weakens and then becomes
ables, including age, gender, education, and tenure, signif- negative. Hence, Hypothesis 4-a is supported. Model 7 added
icantly affect job performance. Age is positively related to the moderating variable Seek_working and the interaction
job performance, indicating that the older employees have between Seek_working and Seek. The results show the coef-
better job performance. Gender is positively related to job ficient of interaction is negative and significant, indicating
performance, indicating that male employees are more likely that employee information-seeking activities on ESM during
to have better performance. Education is negatively related working hours has a significant and negative effect on their
to job performance, indicating that the employees with low job performance and that increasing the proportion of work-
education background have better job performance. Tenure is ing time spent in information-seeking activities will weaken
negatively related to job performance, a possible explanation the positive relationship between ESM usage and job perfor-
for which is that the new employees are willing to work hard mance. Hence, Hypothesis 4-b is supported. Model 8 added
to improve their performance for future promotion. the above two moderating variables and their interactions;
Next, independent variables were added to the model to the result is consistent with the above models, illustrating
estimate the effects of ESM usage behaviors on job perfor- the robustness of the findings. Model 9 added the mod-
mance. Model 2 added information-sharing behavior Share erating variable Use_working and the interaction between
to the benchmark model. Use_working and Use. The results show the coefficient of
As shown in Table 6, the coefficient for Share is negative interaction is negative and significant, indicating that the
and significant, indicating that information-sharing behaviors use of ESM during working hours negatively moderates the
on ESM have negative effects on job performance. Thus, influence of ESM on job performance. Hence, Hypothesis 4-c
Hypothesis 1 is not supported. Model 3 added information- is supported.
seeking behavior Seek to the benchmark model. The coef- We conducted further tests of the model results, shown
ficient for Seek is positive and significant, indicating that in Table 6. Goodness of fit is assessed through the adjusted
information-seeking behaviors on ESM have positive effects R-square value, indicating the percentage of the variabil-
on job performance. Thus, Hypothesis 2 is supported. Model ity in the dependent variable that can be accounted for by
4 added the above two distinct types of usage behaviors the predictors. The results show that adjusted R-square in
simultaneously, and the result is consistent with the above all models was significant, as indicated by the significant

137536 VOLUME 7, 2019

Y. Lu, T. Pan: Effect of Employee Participation in ESM on Their Job Performance

TABLE 6. Hierarchical regression for job performance.

F-Values (Adjusted R2 > 0.015, P < 0.01). The change evaluation used in our study is a reliable and valid measure
in R-square evaluates how much explanatory power was of job performance. To address this concern, we consider
added to the model by each new variable. We found that alternative measures of job performance to assess whether the
adding independent variables (Share, Seek, Use) to the bench- results are robust. Based on the obvious fact that the employ-
mark model increases the R-square by approximately 0.2% ees with high performance would probably be awarded
(Model 2), 1.4% (Model 3), 2.0% (Model 4) and 0.8% some honors, professional titles, or obtain some technical
(Model 5). The changes in R-square were all significant, qualifications and promotion, we aggregate these variables
as indicated by the significant F-Values (P<0.01), indicating and use weighted scores to measure job performance and
that all three independent variables have significant explana- re-estimate the alternative model. Table 7 reports the results
tory power in the models. Similarly, after adding moderating of the alternative specification. The statistical significance
variables and their interaction terms, the R-squared values of the coefficients on the three independent variables Share,
increase by 0.1% (Model 6), 1.0% (Model 7), 0.9% (Model 8) Seek, and Use is similar to the results in Table 5, indicating
and 1.5% (Model 9). The changes in R-square were all sig- that the inclusion of the additional variables measuring job
nificant, as indicated by the significant F-Values (P<0.01), performance does not have a significant impact on the rela-
indicating that all three moderating variables have significant tionship between ESM usage and job performance. However,
explanatory power in the models. the coefficients on the three interactions remain negative and
statistically significant, which further validates the robustness
D. ROBUSTNESS CHECKS of our results.
We conducted several tests to check the robustness of our In another robustness test, considering that information-
results. One concern about the results is whether annual sharing behaviors depend on not only the quantity but also

VOLUME 7, 2019 137537

 Y. Lu, T. Pan: Effect of Employee Participation in ESM on Their Job Performance

TABLE 7. Robustness test for the measure of job performance.

the quality of posts, we redefine the independent variable E. DISCUSSION

Share as the number of words in the posts written by one user. First, the results of this study support the argument that the
Similarly, for the measure of information-seeking behav- usage of social media within enterprises could help improve
ior, we redefine the independent variable Seek as the num- employee job performance, which agrees with the previous
ber of words written in all the posts clicked by one user. studies. The finding will help enterprise managers to realize
Table 8 reports the results of this robustness analysis. The the importance of ESM and motivate them to make full use
coefficients on all independent variables and their interac- of ESM.
tion with moderating variables in the alternative models are We further divided the online behaviors on ESM
statistically significant, and their signs remain unchanged into information-seeking behavior and information-sharing
except for the coefficients of independent variable Share in behavior and found that the two online behaviors have
model 6. The coefficients of Share remain positive but fall opposite effects on job performance. On one hand, we con-
in significance with the inclusion of moderating variable clude that information-seeking behaviors on ESM have pos-
Share_working. One possible reason is that many posts were itive effects on job performance. The finding proves that
copied from other websites and not originally written by the ESM could provide an efficient information platform where
post authors, so the length of posts may less accurately mea- employees could gain much valuable work-related informa-
sure information-sharing behavior compared to the number tion and practical experience of others to make them become
of posts. This provides additional proof that the number of more productive and professional. Meanwhile, it usually
posts is a reliable and valid measure of information-sharing takes very little time and effort for information-seeking activ-
behaviors. Overall, the results from the alternative models ities; they would not result in an apparent decrease in total
appear to be broadly consistent with the results from the employee productivity. To a certain extent, the results could
original model. help ease employer worries that ESM is a time-waster and

137538 VOLUME 7, 2019

Y. Lu, T. Pan: Effect of Employee Participation in ESM on Their Job Performance

TABLE 8. Robustness test for the measure of online behaviors.

will weaken productivity. On the other hand, the results show information-sharing behaviors on job performance when the
that information-sharing behaviors on ESM have negative behaviors occur during working hours.
effects on job performance, which contradicts earlier studies.
One important reason is that it will take much more time VI. CONCLUSION AND IMPLICATIONS
and effort to write posts than to read posts. Much non-work- The rapid development of social media creates a dilemma for
related recreation and leisure information that employees managers; they must decide whether to deploy social media
share on ESM has little value for the improvement of their within the company. If used properly, ESM as an information
job performance. The findings support the idea that addictive communication platform could help employees share and
and excessive use of ESM, especially in non-work-related obtain knowledge and work experience. In contrast, inappro-
activities, can be blamed for reducing productivity. priate or excessive use of ESM may bring unfavorable con-
Finally, to find a practical management approach that how sequences. In this paper, we adopted an empirical approach
to effectively utilize ESM, this study considered whether to explore the effect of ESM on employee job performance
limited-use policies should be implemented. A valuable and have produced interesting results, shown in Table 9. The
finding is that the usage of ESM during working hours findings have significant implications for ESM managers,
negatively moderates the relationship between ESM usage as follows.
and job performance. Using information-sharing behaviors First, ESM should be not perceived as a time-
as an example, though this kind of online behavior could waster or rejected by enterprise managers because of
bring employees certain indirect benefits that may positively much time-consuming. Our studies provide empirical evi-
affect their job performance, it will inevitably take much dence that, overall, the usage of social media within enter-
work time and result in weakening the positive effect of prises could help improve employee job performance.

VOLUME 7, 2019 137539

 Y. Lu, T. Pan: Effect of Employee Participation in ESM on Their Job Performance

TABLE 9. Summary of results.

Therefore, enterprise managers should make full use of social human subjects. According to the US federal regulation for
media technology and consider implementing an integrated the protection of human subjects, research involves human
social media platform within the enterprise that is appropriate subjects only if it collects ‘‘identifiable private information.’’
to their specific conditions. As suggested by Kraut et al. [50], we believe that people
Second, ESM managers should have a better under- who post in some social media platforms should have no
standing of different usage patterns of ESM and their reasonable expectation of privacy, and researchers should
effects on job performance. Our experiments provided be able to treat online communication as public behavior.
insights into how online behaviors affect job performance. In the context of the enterprise social media investigated in
The results demonstrate that information-seeking behav- our study, we consider employee engagement in the ESM
iors on ESM have positive effects on job performance, platform as public behavior inside the enterprise. Accord-
but information-sharing behaviors on ESM have negative ingly, we concluded that there was little risk of breach of
effects on job performance. Therefore, ESM managers do confidentiality when gathering employees’ online behavioral
not have to worry too much about information-seeking data. Furthermore, given that ethical considerations in human
activities and should encourage employees to search and subjects also include undue harm as a consequence of par-
browse valuable information and the practical experience ticipation in research, we think that our practice could mini-
of others through ESM platforms. In contrast, ESM man- mize the harm to human subjects compared to other methods
agers should be aware of the negative effect of information- because our research does not involve the direct involvement
sharing behaviors on job performance and take measures of users and therefore does not harm users.
to restrict the amount of time spent on information-sharing Another key ethical issue in our study is whether gather-
activities. ing data related to job performance through an HRM sys-
Third, the findings have valuable managerial implications tem may pose a potentially high risk to human subjects.
for the use of ESM. The results show that the usage of ESM We approached the issue from two perspectives. First, we col-
during working hours negatively moderates the relationship lected annual performance evaluation data instead of pay-
between ESM usage and job performance. Therefore, ESM for-performance data to measure employee job performance,
managers may consider limiting the use of ESM during work- considering that pay-for-performance is sensitive data but that
ing hours but instead encourage employees to participate in performance evaluation data are much less sensitive because
ESM in their spare time or off-work hours. Corresponding employee performance evaluation was conducted by their
control policies could be implemented to effectively utilize supervisors and peers, and the evaluation results were open to
ESM to help employees improve their job performance while all employees of the enterprise. The performance evaluation
avoiding unfavorable consequences. data are not regarded as protected private information in the
context of the enterprise. Second, we think that a greater
VII. ETHICS risk to confidentiality might result from possible disclosure
Collecting social media data produced in social media of performance evaluation data outside of the research con-
platforms has several strengths compared to traditional text. For example, researchers may mistakenly distribute the
approaches that mainly use surveys and interviews to collect data outside the enterprise, while performance evaluation
data, including the benefit that using large amounts of data data are accessible only within the enterprise but are not
can mitigate the effect of spurious information or extreme open to the public. Considering the potential risk to privacy
views, which makes the research more accurate. However, and confidentiality, researchers are prohibited from access-
social media scholars propose that of particular importance ing personal sensitive information. Instead, in the course of
to the online data collection approach is the need to deter- data processing, the IT department is responsible for sepa-
mine best practices around ethical considerations involving rating identifying information from research data, integrating

137540 VOLUME 7, 2019

Y. Lu, T. Pan: Effect of Employee Participation in ESM on Their Job Performance

ESM usage data and job performance data to produce the final we could have access to ESM data. However, employees may
data available for analysis, and then providing them to the participate in other external social media; we could not collect
researchers. Thus, the researchers are never in possession of such data. We consider collecting such information through
identifying information that could harm the subjects, ensuring surveys and questionnaires to extend the model for a more
that there is no risk of sensitive information disclosure. robust analysis. Finally, the dataset used in our study is only
In addition, we provide a detailed overview of our study from one large equipment manufacturing company, which
in the informed consent document sent to each participating may limit the generalizability of the results. Future studies
employee in order to avoid possible misunderstanding. First, may need to collect more data from other enterprises and
We explained that the purpose of the study is to investigate then compare the effects of ESM on job performance between
whether employee engagement in ESM will affect their job samples from different enterprises to generalize the results.
performance, rather than take their ESM usage as a criterion
for judging their job performance. Second, we made it clear REFERENCES
that the approach used in our study is based on statistical [1] S. Dreher, ‘‘Social media and the world of work: A strategic approach to
hypothesis testing and thus does not require identifiable infor- employees’ participation in social media,’’ Corporate Commun., Int. J.,
vol. 19, no. 4, pp. 344–356, Sep. 2014.
mation. Employees’ personal information including ESM [2] A. P. Mcafee, ‘‘Enterprise 2.0: The dawn of emergent collaboration,’’ MIT
usage and job performance was not used to identify partici- Sloan Manage. Rev., vol. 47, no. 3, pp. 21–28, 2006.
pants and was also protected from being misused. Therefore, [3] P. M. Leonardi, M. Huysman, and C. Steinfield, ‘‘Enterprise social media:
Definition, history, and prospects for the study of social technologies in
it is not necessary for participating employees to worry that organizations,’’ J. Comput.-Mediated Commun., vol. 19, no. 1, pp. 1–19,
we will use their ESM behavior data to assess their perfor- Oct. 2013.
mance. Third, some possible benefits have been described in [4] M. H. Jarrahi, ‘‘Social networking, social network technologies, and the
enterprise,’’ in Proc. iConference, Feb. 2011, pp. 688–690.
the informed consent document to make the employees feel [5] M. Brandel, ‘‘Social networking behind the firewall,’’ Computer World,
that they are likely to benefit from the research findings, such Tech. Rep., Aug. 2008. [Online]. Available: https://www.computerworld.
as better understanding the advantages and disadvantages com/article/2551401/the-new-employee-connection–social-networking-
behind-the-firewall.html
of ESM, eliminating the worry about the use of ESM is a [6] J. DiMicco, D. R. Millen, M. Geyer, C. Dugan, B. Brownholtz, and
waste of time, and using it in the right way to improve their M. Müller, ‘‘Motivations for social networking at work,’’ in Proc. ACM
job performance. Conf. Comput. Supported Cooperat. Work, Nov. 2008, pp. 711–720.
[7] C. Steinfield, J. M. Dimicco, N. B. Ellison, and C. Lampe, ‘‘Bowling
In summary, as suggested by Kraut et al. [50], our data col- Online: Social networking and social capital within the organization,’’ in
lection approach is not inherently riskier to human subjects Proc. 4th Int. Conf. Commun. Technol., Jun. 2009, pp. 245–254.
than more traditional research styles; in some respects, it may [8] M. J. Brzozowski, ‘‘WaterCooler: Exploring an organization through enter-
prise social media,’’ in Proc. Int. Conf. Supporting Group Work, May 2009,
even be less risky. Therefore, we believe that our research has pp. 219–228.
minimal risk to human subjects and have followed the core [9] Z. Mo and Y. Yu, ‘‘The mechanism of empowering leadership’s effect on
ethical principles; we thus should be able to fulfill the ethical employee creativity in dingtalk alibaba context,’’ in Proc. 4th Int. Forum
Decision Sciences. Singapore: Springer, 2017, pp. 99–117.
requirements. [10] K. Le-Nguyen, Y. Guo, and J. Qiong, ‘‘Developing enterprise social soft-
ware in China: Insights from Mingdao,’’ J. Inf. Technol. Teaching Cases,
VIII. LIMITATIONS AND FUTURE RESEARCH vol. 7, no. 2, pp. 62–78, 2017.
[11] J. Bughin, M. Chui, A. Miller, ‘‘How companies are benefiting from Web
This study also has limitations. First, the measurement of 2.0,’’ McKinsey Quart., vol. 9, no. 4, pp. 84–85, Jun. 2009.
some variables used in the study may not be accurate enough. [12] M. Quarterly. Business and Web 2.0: An Interactive Feature.
For example, we used the number of total click activities as Accessed: Apr. 2010. [Online]. Available: https://www.mckinsey.
com/business-functions/digital-mckinsey/our-insights/business-and-web-
the measure of user involvement in ESM. However, some 20-an-interactive-feature
users probably write or read long posts that take much time [13] R. N. Landers and R. C. Callan, ‘‘Validation of the beneficial and harm-
but require few clicks; we could not collect such information ful work-related social media behavioral taxonomies: Development of
the work-related social media questionnaire,’’ Social Sci. Comput. Rev.,
from users’ online behavioral data. Therefore, we hope to
vol. 32, no. 5, pp. 628–646, Mar. 2014.
use multiple data sources and methods of data collection in [14] J. R. Carlson, R. B. Harris, K. J. Harris, and D. S. Carlson, ‘‘Social media
future research, such as combining online behavioral data use in the workplace: A study of dual effects,’’ J. Org. End User Comput.,
with self-report survey data, to improve the validity and vol. 28, no. 1, pp. 15–31, Jan. 2016.
[15] J. C. Y. Ng, I. Y. T. Shao, and Y. Liu, ‘‘This is not what I wanted: The effect
reliability of the study and gain an in-depth understanding of avoidance coping strategy on non-work-related social media use at the
of the research phenomenon. Second, we only divided the workplace,’’ Employee Relations, vol. 38, no. 4, pp. 466–486, Jun. 2016.
user online behaviors on ESM into information-sharing and [16] R. M. Davison, C. X. J. Ou, and M. G. Martinsons, ‘‘Information technol-
ogy to support informal knowledge sharing,’’ Inf. Syst. J., vol. 23, no. 1,
information-seeking behaviors according to their different pp. 89–109, Jan. 2013.
motivations for involvement in ESM. However, this division [17] C. S. Lee, M. B. Watson-Manheim, K. M. Chudoba, and C. H. Lee, ‘‘Use of
is somewhat rough, since users have a wide variety of motiva- social media in the workplace,’’ Personnel J., vol. 71, no. 6, pp. 1–16, 2013.
[18] C. X. J. Ou and R. M. Davison, ‘‘Interactive or interruptive? Instant
tions. Another frequently used division is between utilitarian messaging at work,’’ Decision Support Syst., vol. 52, no. 1, pp. 61–72,
and hedonic motivations. Accordingly, user online behav- Dec. 2011.
iors could be divided into work-related behaviors and non- [19] A. H. Pitafi, S. Kanwal, and A. Pitafi, ‘‘Effect of enterprise social media
and psychological safety on employee’s agility: Mediating role of com-
work-related behaviors. However, it is difficult to find the munication quality,’’ Int. J. Agile Syst. Manage., vol. 12, no. 1, pp. 1–26,
appropriate indicators to measure these two behaviors. Third, Mar. 2019.

VOLUME 7, 2019 137541

 Y. Lu, T. Pan: Effect of Employee Participation in ESM on Their Job Performance

[20] A. H. Pitafi, H. Liu, and Z. Cai, ‘‘Investigating the relationship between [43] G. von Krogh, ‘‘How does social software change knowledge manage-
workplace conflict and employee agility: The role of enterprise social ment? Toward a strategic research agenda,’’ J. Strategic Inf. Syst., vol. 21,
media,’’ Telematics Informat., vol. 35, no. 8, pp. 2157–2172, Dec. 2018. no. 2, pp. 154–164, Jun. 2012.
[21] A. H. Pitafi, S. Kanwal, A. Ali, A. N. Khan, and M. W. Ameen, ‘‘Mod- [44] J. T. Tsay, L. Dabbish, and J. Herbsleb, ‘‘Social media and success in
erating roles of IT competency and work cooperation on employee work open source projects,’’ in Proc. Conf. Comput. Supported Cooperat. Work
performance in an ESM environment,’’ Technol. Soc., vol. 55, pp. 199–208, Companion, Feb. 2012, pp. 223–226.
Nov. 2018. [45] L. Yingjie, S. Deng, and T. Pan, ‘‘Does usage of enterprise social media
[22] J. W. Treem and P. M. Leonardi, ‘‘Social media use in organizations: affect employee turnover? Empirical evidence from Chinese companies,’’
Exploring the affordances of visibility, editability, persistence, and asso- Internet Res., vol. 29, no. 4, pp. 970–992, Aug. 2019. doi: 10.1108/INTR-
ciation,’’ Ann. Int. Commun. Assoc., vol. 36, no. 1, pp. 143–189, 2013. 03-2018-0140.
[23] W. He and K.-K. Wei, ‘‘What drives continued knowledge sharing? An [46] R. Aggarwal, R. Gopal, R. Sankaranarayanan, and P. V. Singh , ‘‘1 blog,
investigation of knowledge-contribution and -seeking beliefs,’’ Decision blogger, and the firm: Can negative posts by employees lead to positive
support Syst., vol. 46, no. 4, pp. 826–838, Mar. 2009. outcomes?’’ Social Sci. Electron., vol. 23, no. 2, pp. 306–322, 2010.
[24] Y. Huang, P. V. Singh, and A. Ghose, ‘‘Show me the incentives: [47] W. van Zoonen, J. W. M. Verhoeven, and R. Vliegenthart, ‘‘How employ-
A dynamic structural model of employee blogging behavior,’’ in Proc. ees use Twitter to talk about work: A typology of work-related tweets,’’
ICIS, Dec. 2010, p. 200. Comput. Hum. Behav., vol. 55, no. 1, pp. 329–339, Feb. 2016.
[25] C.-M. Chiu, M.-H. Hsu, and E. T. G. Wang, ‘‘Understanding knowledge [48] N. Luo, X. Guo, J. Zhang, G. Chen, and N. Zhang, ‘‘Understanding
sharing in virtual communities: An integration of social capital and social the continued use of intra-organizational blogs: An adaptive habituation
cognitive theories,’’ Decision support Syst., vol. 42, no. 3, pp. 1872–1888, model,’’ Comput. Hum. Behav., vol. 50, pp. 57–65, Sep. 2015.
Dec. 2006. [49] M. Burke, C. Marlow, and T. M. Lento, ‘‘Social network activity and social
[26] P. V. Singh, N. Sahoo, and T. Mukhopadhyay, ‘‘How to attract and retain well-being,’’ in Proc. 28th Int. Conf. Hum. Factors Comput. Syst., Atlanta,
readers in enterprise blogging?’’ Inf. Syst. Res., vol. 25, no. 1, pp. 35–52, GA, USA, Apr. 2010, pp. 1909–1912.
Mar. 2014. [50] R. Kraut, J. Olson, M. Banaji, A. Bruckman, J. Cohen, and M. Couper,
‘‘Psychological research Online: Report of board of scientific affairs’
[27] Y. Huang, P. V. Singh, and A. Ghose, ‘‘A structural model of employee
advisory group on the conduct of research on the Internet,’’ Amer. Psychol.,
behavioral dynamics in enterprise social media,’’ Manage. Sci., vol. 61,
vol. 59, no. 2, pp. 105–114, Feb. 2004.
no. 12, pp. 2825–2844, May 2015.
[28] M. M. Wasko and S. Faraj, ‘‘Why should i share? examining social cap-
ital and knowledge contribution in electronic networks of practice,’’ MIS
Quart., vol. 29, no. 1, pp. 35–57, 2005.
[29] Y. Yan and R. M. Davison, ‘‘Exploring behavioral transfer from knowl-
edge seeking to knowledge contributing: The mediating role of intrinsic
motivation,’’ J. Amer. Soc. Inf. Sci. Technol., vol. 64, no. 6, pp. 1144–1157,
Jun. 2013.
[30] Y. Yan, X. Zha, and M. Yan, ‘‘Exploring employee perceptions of Web
2.0 virtual communities from the perspective of knowledge sharing,’’ Aslib
J. Inf. Manage., vol. 66, no. 4, pp. 381–400, Jul. 2014.
[31] S.-M. Pi, C.-H. Chou, and H.-L. Liao, ‘‘A study of Facebook Groups YINGJIE LU received the B.S. degree in informa-
members’ knowledge sharing,’’ Comput. Human Behav., vol. 29, no. 5, tion management and information system and the
pp. 1971–1979, Sep. 2013.
M.S. degree in management information system
[32] S. Yardi, S. A. Golder, and M. J. Brzozowski, ‘‘Blogging at work and
from the Harbin Institute of Technology, Harbin,
the corporate attention economy,’’ in Proc. SIGCHI Conf. Hum. Factors
China, in 2005 and 2007, respectively, and the
Comput. Syst., Apr. 2009, pp. 2071–2080.
[33] X. Cao, D. R. Vogel, X. Guo, H. Liu, and J. Gu, ‘‘Understanding the
Ph.D. degree in management science and technol-
influence of social media in the workplace: An integration of media ogy from Shanghai Jiao Tong University, Shang-
synchronicity and social capital theories,’’ in Proc. 45th Hawaii Int. Conf. hai, China, in 2013. He was a Visiting Scholar
Syst. Sci., Jan. 2012, pp. 3938–3947. with the Artificial Intelligence Lab, Eller School
[34] M. M. Skeels and J. Grudin, ‘‘When social networks cross boundaries: of Business, The University of Arizona, Tucson,
A case study of workplace use of Facebook and linkedin,’’ in Proc. ACM AZ, USA, in 2012. He is currently an Associate Professor with the Beijing
Int. Conf. Supporting Group Work, May 2009, pp. 95–104. University of Chemical Technology, Beijing, China. His research inter-
[35] E. Turban, N. Bolloju, and T.-P. Liang, ‘‘Enterprise social networking: ests include data mining, social media, electronic commerce, and big data
Opportunities, adoption, and risk mitigation,’’ J. Org. Comput. Electron. analysis.
Commerce, vol. 21, no. 3, pp. 202–220, Jul. 2011.
[36] B. B. Aguenza and A. P. M. Som, ‘‘A conceptual analysis of social net-
working and its impact on employee productivity,’’ IOSR J. Bus. Manage.,
vol. 1, no. 2, pp. 48–52, 2012.
[37] A. Ferreira and P. T. Du, ‘‘Effect of Online social networking on employee
productivity,’’ South African J. Inf. Manage., vol. 11, no. 1, pp. 1–11,
Mar. 2009.
[38] C. J. König and M. E. C. de la Guardia, ‘‘Exploring the positive side
of personal Internet use at work: Does it help in managing the bor-
der between work and nonwork?’’ Comput. Hum. Behavior, vol. 30,
pp. 355–360, Jan. 2014.
TAOTAO PAN received the B.S. degree in
[39] R. Benzie. Facebook Banned for Ontario Staffers. Accessed:
Jul. 2007. [Online]. Available: http://www.thestar.com/news/2007/05/03/ information management and information sys-
facebook_banned_for_ontario_staffers.html tem from Shandong Technology and Business
[40] J. Wilson, ‘‘Social networking: The business case,’’ Eng. Technol., vol. 4, University, Shandong, China, in 2017. She is
no. 10, pp. 54–56, Jun. 2009. currently pursuing the master’s degree with the
[41] P. Charoensukmongkol, ‘‘Effects of support and job demands on social School of Economics and Management, Beijing
media use and work outcomes,’’ Comput. Hum. Behav., vol. 36, University of Chemical Technology, Beijing,
pp. 340–349, Jul. 2014. China. Her research interests include information
[42] M. Moqbel, S. Nevo, and N. Kock, ‘‘Organizational members’ use of social retrieval, data mining, and management informa-
networking sites and job performance,’’ Inf. Technol. People, vol. 26, no. 3, tion systems.
pp. 240–264, Aug. 2013.

137542 VOLUME 7, 2019