You are on page 1of 10

ARTICLE IN PRESS

International Journal of Information Management 28 (2008) 423–432


www.elsevier.com/locate/ijinfomgt

Sources of knowledge acquisition and patterns of knowledge-sharing


behaviors—An empirical study of Taiwanese high-tech firms
Min-Shi Liu, Nien-Chi Liu
Institute of Human Resource Management, National Central University, No. 300, Jhongda Rd., Jhongli City, Taoyuan Country 32001, Taiwan

Abstract

The new knowledge acquisition and sharing stage represents the start of the organization’s overall knowledge creation process. It is
especially important for contributing to the critical foundation of organizational knowledge creation. This study explores the relationship
between employees’ knowledge acquisition sources and the patterns of knowledge-sharing behaviors. We use structural equation
modeling to test a sample of R&D professionals from high-tech companies in Taiwan. Data analysis suggests that most employees prefer
to acquire knowledge from, and share knowledge with, their team members. This implies that greater familiarity between team members
and strengthened cooperative relationships foster productivity. Furthermore, employees should be encouraged to participate in
professional communities in order to acquire new knowledge. Knowledge acquired via these channels will facilitate the sharing of R&D
knowledge within the organization.
r 2008 Elsevier Ltd. All rights reserved.

Keywords: Knowledge innovation; Knowledge acquisition; Knowledge sharing; Structural equation modeling; Taiwan high-tech companies

1. Introduction suggests that there may exist ‘‘the edge of chaos,’’ which
is the place between excessive disorder and excessive order,
From a firm’s knowledge-based view, the uniqueness of in the knowledge creation process. The knowledge will be
knowledge plays an important role in maintaining a created under highly constrained and formal contexts, as
company’s competitive advantage (Grant, 1996). Never- well as under high uncertain and random ways (Koivuaho
theless, the essential dilemma within the firm is how to & Laihonen, 2006). Knowledge management in an evolu-
manage knowledge in a way that creates a competitive tionary perspective proposes that the knowledge creation
advantage (Barney, 1991). A company wishing to establish in an organization may start from a simple reaction to
and maintain the knowledge needed for a competitive information into higher levels of structural and functional
advantage must create and acquire new knowledge, complexity (Laszlo & Laszlo, 2002). Network theory then
transmit knowledge to appropriate parts of the company, suggests that the formality of network structure may result
interpret that knowledge and integrate it with existing in different types of knowledge diffusion within an
knowledge, and use knowledge to achieve better perfor- organization (Dyer & Nobeoka, 2000). It seems that every
mance (Cohen & Levinthal, 1990). Therefore, managing perspective has its own developmental path to explore
knowledge effectively is an immense challenge for firms. knowledge management issues, but still presents similar
In recent years, more theories also have been applied to trends. They seem to suggest that knowledge creation is a
knowledge management issues. For example, from com- dynamic and evolving process, in which kinds of patterns
plexity theory, to understand how the information flows is still emerge, and it is extremely imperative to explore what
the first step of understanding a complex system. The kinds of knowledge flows are created.
approach to complexity-based knowledge management Nonaka & Konno (1998) have argued that the knowl-
edge creation process is a continuous process through
Corresponding author. Tel.: +886 222157010. which one transcends the boundary between self and other,
E-mail address: kayliu0102@yahoo.com.tw (M.-S. Liu). as knowledge is created through the interactions between

0268-4012/$ - see front matter r 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijinfomgt.2008.01.005
ARTICLE IN PRESS
424 M.-S. Liu, N.-C. Liu / International Journal of Information Management 28 (2008) 423–432

or among individuals, groups, and in the organization Though organizations usually use a variety of mechan-
(Popadiuk & Choo, 2006). Looking through the whole isms (i.e., formal documents, training programs, group
knowledge creation process, the stages of knowledge meetings) to promote workplace learning, employees may
acquisition and sharing actually initiate the organization’s not accumulate their knowledge merely through inside
overall knowledge creation process. Thus, the significance sources. One of the most important ways for an organiza-
of individual knowledge acquisition and sharing behaviors tion to gain a competitive advantage is by obtaining
needs to be emphasized, since it will contribute to the information from external environments efficiently.
critical foundation of organizational knowledge creation Traditionally, the boundary spanners play a central
(Bock, Zmud, Kim, & Lee, 2005). However, in discussing role in introducing new knowledge into organizations
the knowledge creation process, existing research seldom (Burn, 2004). In today’s competitive landscape, it is
has integrated both sources of knowledge acquisition and impossible to rely merely on limited boundary spanners
knowledge-sharing behaviors. Thus, there is a need for to absorb knowledge. The traditional methods may not
research that explores an integrated model in which produce effective results, and the new changes encourage
acquiring knowledge from different sources will result most employees to acquire work knowledge from outside
in variations of the patterns of knowledge-sharing environments in order to ensure continued competitiveness
behaviors. and survival (Grant, 1996). For some individuals knowl-
In order to integrate and extend these existing views, we edge from both internal and external sources may be fresh,
developed a theoretical frame and undertook an empirical but for the organization as a whole the external sources
study to explore the relationship between the sources of usually offer more novel knowledge than internal sources.
knowledge acquisition and patterns of knowledge-sharing Some valuable information or know-how, which usually is
behaviors. The structure of the current article is as follows. tacit, is still conveyed through intensive interactions among
In the second section, we review the related literature on people from different organizations (Bontis, Crossan, &
the sources of knowledge acquisition and patterns of Hulland, 2002).
knowledge-sharing behaviors, and present an integrated Based on the previous theoretical discussion, knowledge
model as well as the hypothesized relationships of sources of individual workplace learning may be classified
these variables. The third section presents the research into two dimensions: knowledge storage formats and
methodology, and the fourth section includes a report knowledge storage locations. However, there are few
of the findings. The last section presents a discussion of the empirical studies to verify the previous theoretical argu-
findings and limitations of our study, as well as suggestions ments. Rulke, Zaheer, and Anderson (2000) found three
for further research. types of knowledge sources to exist for store managers in
food industries: internal relational, internal non-relational,
and external channels. The previous theoretical argument is
2. Sources of knowledge acquisition not supported fully by their findings. However, this does
not mean that external relational channels are not
Employees may obtain knowledge through a variety of important, or that the distinction between external
learning activities within an organization, such as training, relational and external non-relational channels is insignif-
formal education, experimentation, imitation, and self- icant. For example, researchers have pointed out that the
directed learning (Reio & Wiswell, 2000). Through exchange of ideas in professional communities in high-tech
continuous learning, employees can renew their basis for industries is very important for organizations to keep their
future organizational learning when knowledge sharing is technology future-oriented (Assimakopoulos & Yan,
generated among employees (Tsoukas & Vladimirou, 2006). Thus, we still need more empirical studies on
2001). knowledge sources in order to develop a more generalized
Knowledge generally is classified into two categories typology.
(Nonaka & Takeushi, 1995): explicit and tacit knowledge.
Explicit knowledge refers to knowledge that is codified as 3. Sources of knowledge acquisition and the patterns of
signs and symbols, or to formulated information that is knowledge-sharing behaviors
conveyed by formal systems. Books, manuals, and codes of
practice are popular forms of explicit knowledge. On the Once new knowledge has been acquired, it must be
other hand, it is more difficult to share implicit or tacit transferred to other parts of the organization where it is
knowledge only through codified signs and symbols. Since needed or can generate benefit. Incoming knowledge will
tacit knowledge is very personal and should be constructed have very little effect on the organization without this
socially, the interaction between knowledge transmitters stage. As a consequence, transferring knowledge across
and receivers, which allows for observations and real-time time and space is an extremely important part of knowl-
discussions, becomes critical for transmitting tacit knowl- edge accumulation within an organization. Here, knowl-
edge efficiently. Thus, it is apparent that individuals may edge sharing refers to the transfer of knowledge acquired
rely on different learning channels to obtain explicit and by an individual to other persons. We can distinguish the
tacit knowledge. patterns of knowledge-sharing behaviors in large-group
ARTICLE IN PRESS
M.-S. Liu, N.-C. Liu / International Journal of Information Management 28 (2008) 423–432 425

and small-group settings. Knowledge sharing in a settings. When the relative benefits of knowledge sharing
large-group setting refers to the transmission of useful in large- and small-group settings are compared, the strong
knowledge to a large number of people via the company’s ties that characterize small-group settings favor the sharing
internal and external education, training, and publications. of non-codified knowledge (Granovetter, 1982), because
Knowledge sharing in a small-group setting refers to the small-group settings typically provide knowledge transmit-
private sharing of useful knowledge with coworkers ters and receivers with opportunities for two-way interac-
belonging to the same group. Regardless of whether tion. Members of small groups usually have opportunities
knowledge is shared in a large or small-group setting, the to try out knowledge, make mistakes, and seek explanation
knowledge-sharing process always will include the trans- and feedback from receivers. The opportunities for two-
mission and reception of individual knowledge (Bock & way interaction provided by small-group settings facilitate
Kim, 2002), which is the most troublesome part of the the absorption of non-codified knowledge, because the
knowledge management process. receiver may not acquire complete knowledge during the
There are two reasons why knowledge sharing is the first instance of interaction, and may need several
most difficult part of the knowledge management process. opportunities for full absorption.
First is the inherently ‘‘sticky’’ nature of knowledge. Furthermore, even if the transmitter does not actively
Knowledge is embedded in individuals’ brains (Szulanski, aid in knowledge transfer or participate in two-way
2000), and its stickiness tends to make transfer between interactions, the mutual understanding between the two
individuals slow, costly, and uncertain (Kogut & Zander, parties eases the difficulty of transferring non-codified
1992). Second is that the knowledge-sharing process knowledge. When knowledge sharing takes place in a
includes the externalization of knowledge by the knowl- small-group setting, the two actors tend to establish special
edge transmitter and internalization of knowledge by the relationships and friendships as they work to resolve
receiver. In the process of externalizing and internalizing problems connected with the transfer of non-codified
knowledge, individuals obtain knowledge needed for their knowledge. For instance, Uzzi’s research has shown that
work not only from sources inside the organization, but the close personal ties that emerge in small-group settings
also from outside the organization. As a result, the failure play an important role in facilitating the communication of
to effectively link knowledge sharing with knowledge non-codified knowledge (Uzzi, 1997). According to the
acquisition channels will increase knowledge search and past research, internal non-codified sources increase
transmission costs and reduce knowledge-sharing effi- individual knowledge-sharing behaviors in small-group
ciency. settings, which leads to the following hypothesis.

3.1. Internal sources of knowledge acquisition and the Hypothesis 3. Internal non-codified sources have a positive
patterns of knowledge-sharing behaviors influence on knowledge sharing in a small group setting.

Effectively linking knowledge acquisition channels and


In contrast, knowledge sharing in a large group setting
sharing patterns can lower sharing search costs and
tends to lack the interaction needed to transfer non-
enhance sharing efficiency. We will first discuss the
codified knowledge. Interaction between the transmitter
relationship between an organization’s internal knowledge
and receiver may be infrequent. The receiver must interpret
sources and knowledge sharing. The transfer of codified
and revise non-codified knowledge, and often cannot
knowledge within an organization is relatively easy, and
obtain a follow-up explanation. It may be impossible to
the transfer of codified knowledge can be done readily
communicate directly with the transmitter, and the receiver
between large and small groups (Hansen, 1999; Tsai, 2002).
may not be able to get an immediate solution from the
Thus, according to past research, internal codified sources
transmitter when problems or questions arise. Even if the
increase individual knowledge-sharing behaviors in large-
transmitter can provide assistance, the fact that there is no
and small-group settings. Therefore, we have developed the
means of communicating tacit knowledge between the two
following hypotheses.
parties makes it difficult to achieve good knowledge
Hypothesis 1. Internal codified sources have a positive transfer results. It takes a considerable amount of time to
influence on knowledge sharing in a small group setting. resolve these impediments, and knowledge sharing may
become a burden, possibly hindering project implementa-
Hypothesis 2. Internal codified sources have a positive tion (Hansen, 1999). According to past research, internal
influence on knowledge sharing in a large group setting. non-codified sources decrease individual knowledge-shar-
It is difficult, nonetheless, to transfer non-codified ing behaviors in large-group settings, which leads to the
knowledge. When the knowledge to be shared is non- following hypothesis.
codified, the best results may be obtained by conducting
knowledge sharing in small group settings. Transmitters Hypothesis 4. Internal non-codified sources have a nega-
may need to spend more time externalizing tacit knowledge tive influence on knowledge sharing in a large group
when knowledge sharing takes place in small-group setting.
ARTICLE IN PRESS
426 M.-S. Liu, N.-C. Liu / International Journal of Information Management 28 (2008) 423–432

3.2. External sources of knowledge acquisition and the Furthermore, when organizational knowledge is a public
patterns of knowledge-sharing behaviors good (Wasko & Faraj, 2000), the use of knowledge from
others to improve one’s own work performance will not
An organization’s external channels may be an impor- lessen the value of that knowledge to others. When this
tant means of knowledge acquisition. External sources of knowledge is shared among employees, boundary spanners
codified and non-codified knowledge often provide valu- are willing to share knowledge acquired from external
able information and facilitate knowledge innovation channels with the organization’s employees. Past research
(Assimakopoulos & Yan, 2006). Although knowledge from has clearly shown that there is a cause-and-effect relation-
outside the organization may be more abstract and difficult ship between non-codified external sources of knowledge
to acquire, it also may be of greater value. This external acquisition and knowledge sharing in a large-group setting
knowledge can inspire fresh thinking among employees and (Cabrera & Cabrera, 2005). We therefore propose the
provide a context for internal knowledge benchmarking. following hypothesis and the model is displayed in Fig. 1.
The union of internal knowledge and useful external non-
codified knowledge can generate unique insights and Hypothesis 8. External non-codified sources have a posi-
promote knowledge innovation. External knowledge ac- tive influence on knowledge sharing in a large-group
quisition channels include individuals’ acquisition of public setting.
information via the Internet or specialized books, as well as
acquisition of new external knowledge through personal 4. Research methods
networks and professional associations (Brown & Duguid,
2001). In addition, the organization can acquire external 4.1. Sampling and respondents
knowledge through external training, recruiting of other
companies’ employees, alliances (Grant, 1996), and so on. R&D professionals are widely acknowledged as knowl-
External codified knowledge constitutes public informa- edge workers. Their work is characterized by complex
tion, and the transfer of this knowledge may have the same system design, knowledge related to the rapid update of
effects in large-and small-group settings (Hansen, 1999). technology, and strong competition for sustaining innova-
Past research has shown that the external codified sources tion (Assimakopoulos & Yan, 2006). In addition, the
of knowledge may affect the patterns of knowledge-sharing knowledge acquisition and sharing patterns among R&D
behaviors (knowledge sharing in small and large group professionals are quite different from those of traditional
settings). We therefore propose the following hypotheses. workers (Assimakopoulos & Yan, 2006; Berends, Bij,
Debackere, & Weggeman, 2006; Cabrera, Collins, &
Hypothesis 5. External codified sources have a positive Salgado, 2006; Hsu, 2006). Since the R&D process is
influence on knowledge sharing in a small-group setting. extremely complex and an employee possesses expertise
only in a specific area, knowledge will not be exposed to
Hypothesis 6. External codified sources have a positive other individuals until knowledge owners make the objects
influence on knowledge sharing in a large-group setting. available. Therefore, it is valuable to investigate how R&D
professionals acquire and share knowledge among mem-
Non-codified knowledge, which tends to be abstract,
bers. Since knowledge acquisition and sharing behaviors
difficult to acquire, and beneficial to innovation, can be
are so critical for R&D professionals, the current sample
transferred via personal networks or professional commu-
consists of R&D employees employed in nine high-tech
nities (Brown & Duguid, 2001). In contrast to homogenous
companies in Taiwan. The companies studied are engaged
knowledge acquired from internal non-codified sources,
in applied research within the computer systems and other
knowledge acquired from external non-codified sources
industries that are pivotal for high-tech industries. These
tends to be heterogeneous and varied, and often serves to
nine companies were chosen in order to control for the
promote knowledge innovation. As a consequence, when
differences in firm size. Generally, companies with success-
employees cannot obtain needed information from within
ful experience in managing knowledge management activ-
the company, they may seek support from outside the
ities have a large operating scale, and the average total
organization’s boundaries. Small-group settings are char-
operating revenue of the nine companies is more than
acterized by strong ties (Granovetter, 1982), so the sharing
US$3 billion (Table 1).
of knowledge acquired from external non-codified sources
in small-group settings (whose members may have close
relationships and be in frequent contact with one other) 4.2. Measures
may better facilitate the acquisition of varied heteroge-
neous knowledge in such settings. Therefore, we propose 4.2.1. Independent variables
the following hypothesis. We used nine items based on the theoretical literature to
assess possible sources of knowledge acquisition in high-
Hypothesis 7. External non-codified sources have a posi- tech industries. The items were rated on a 5-point Likert
tive influence on knowledge sharing in a small-group scale regarding the importance of each source in searching
setting. for solutions to work-related problems. The nine items
ARTICLE IN PRESS
M.-S. Liu, N.-C. Liu / International Journal of Information Management 28 (2008) 423–432 427

used to assess sources were based on the discussion of Confirmatory factor analysis (CFA) was used to examine
knowledge sources by Bird (1994), Sacks (1994), and Rulke the validity and reliability of the research instrument.
et al.(2000), including discussions of ‘‘internal codified Validity was addressed in two ways. First, operationaliza-
sources,’’ ‘‘internal non-codified sources,’’ ‘‘external codi- tion of the constructs drew upon an extensive literature
fied sources,’’ and ‘‘external non-codified sources.’’ review, and questionnaires were discussed with superiors in
The internal codified sources include internal formal actual organizations. Both processes were intended to
training and formal training from technology transfer. The enhance content validity. Second, the results of the CFA
internal non-codified sources include sources such as super- illustrated that six variables displayed convergent validity,
visors and colleagues, as well as formal meetings. External since the analysis yielded six factors with factor loading
codified sources refer to information from professional displaying the expected patterns. Table 2 shows the
journals, books, and Internet sources. Employees in high- correlations among the six factors of the CFA model.
tech industries also could acquire task knowledge through The diagonal of the matrix presents the internal-consis-
external non-codified sources, such as exchanging ideas with tency coefficients of the reliability of the unity-weighted
personal friends and professionals in other firms. sums of the item scores, and the a-coefficients vary between
0.62 and 0.87. The Cronbach a coefficients of the six
4.2.2. Dependent variables variables of this study all exceeded the 0.60 threshold
The patterns of knowledge-sharing behaviors were the recommended by Malhotra and Birks (2003), with the
major dependent variables of interest in this study. Rated reliability and internal consistency of the variables con-
by a 5-point Likert scale, two patterns of knowledge- firmed.
sharing behaviors were measured. First, knowledge sharing
in a large-group setting refers to the transmission of useful 5. Results
knowledge to a large number of people via internal and
external education, training, and publications. Three items 5.1. The measurement model
were used to measure this construct. The second, measured
by three questions, was knowledge sharing in a small-group We tested the factor structure of our survey measures
setting, which was defined as the private sharing of useful using CFA. The CFA model was developed with the
knowledge with coworkers belonging to the same group. purpose of testing whether the hypothesized latent
The measures of these two variables were adapted from the variables could be identified empirically, as well as for
theoretical literature (Senge, 1990). examining the validity and reliability of the measures.
First, according to the successive testing and modification
of different models on the same data, different variable sets
Table 1 were combined into the final model. The final CFA model
Respondent demographic information had a w2 value of 173.79, df ¼ 75 (p ¼ 0.00). The RMSEA
index for this model was 0.061, with a 90 percent
Variable Frequency Percentage (%)
confidence interval from 0.049 to 0.072, indicating
Gender acceptable fit of the model to the data. Other indices of
Male 318 85.7 fit also showed acceptable fit (GFI ¼ 0.94, AGFI ¼ 0.90,
Female 53 14.3 CFI ¼ 0.97, NFI ¼ 0.95, NNFI ¼ 0.96). Standardized
Age factor loadings along with the descriptive item data are
Under 30 148 39.9 shown in Table 3.
30–39 194 52.3 Two items measured the internal codified sources factor.
40–49 28 7.5
The items with loadings of 0.67 were ‘‘internal formal
Over 50 1 0.3
training’’ and ‘‘formal training from technology transfer.’’
Education The internal non-codified sources factor was related to three
High school or less 1 0.3
items. Higher loadings (0.75) were observed for two items:
Some college 27 7.3
Bachelor’s degree 117 31.5 ‘‘asking colleagues’’ and ‘‘internal meeting.’’ The lower
Graduate degree 218 58.8 loading (0.63) was ‘‘asking supervisors.’’ The external
Ph.D. degree 8 2.1 codified sources factor was related to two items. The highest
Tenure in present jobs loading (0.71) was observed for the ‘‘professional books’’
Under 1 105 28.3 item. The lower loading item (0.69) was ‘‘Internet.’’ Two
1–2.9 143 38.5 items measured the external non-codified sources factor,
3–4.9 77 20.8 ‘‘personal networks’’ and ‘‘professional communities,’’ and
5–9.9 33 8.95
had loadings of 0.73 and 0.74, respectively.
Over 10 13 3.5
Three items measured the knowledge sharing in a large
Manager group setting factor, and all items had substantial loadings
Yes 86 23.2
(0.70–0.92) on the factor. The highest loading (0.9) was
No 285 76.8
observed for the item ‘‘sharing with people in training
ARTICLE IN PRESS
428 M.-S. Liu, N.-C. Liu / International Journal of Information Management 28 (2008) 423–432

Table 2
Correlations among the factors of the measurement model

Variable 1 2 3 4 5 6

1. Internal codified sources (0.62)


2. Internal non-codified sources 0.65 (0.75)
3. External codified sources 0.69 0.44 (0.66)
4. External non-codified sources 0.58 0.58 0.58 (0.70)
5. Knowledge sharing in a large-group setting 0.49 0.27 0.52 0.55 (0.87)
6. Knowledge sharing in a small-group setting 0.39 0.46 0.31 0.39 0.21 (0.85)

N ¼ 371. Values in parentheses are Cronbach’s.

Table 3
Means, standard deviations, and factor loadings for the items

Item Item no. Mean SD Standard factor loading

Internal codified sources


Internal formal training IC1 3.26 0.85 0.67
Formal training from technology transfer IC2 3.30 0.88 0.67
Internal non-codified sources
Asking supervisors INC1 3.47 0.92 0.63
Asking colleagues INC2 3.83 0.69 0.75
Internal meeting INC3 3.72 0.78 0.75
External codified sources
Professional books EC1 3.62 0.78 0.71
Internet EC2 3.18 0.99 0.69
External non-codified sources
Personal networks ENC1 3.33 0.90 0.73
Professional communities ENC2 3.27 0.96 0.74

Knowledge sharing in a large-group setting


Sharing with people in training outside the firm LG1 1.84 1.03 0.92
Publishing in professional journals LG2 1.65 0.96 0.85
Sharing with people in training inside the firm LG3 2.43 1.12 0.70
Knowledge sharing in a small-group setting
Private sharing in small groups SG1 3.76 0.69 0.87
Sharing in the meeting SG2 3.68 0.80 0.77
Coaching my subordinate SG3 3.75 0.77 0.78

outside the firm’’. Another loading (0.85) was observed for with a 90 percent confidence interval between 0.054 and
the item ‘‘publishing in professional journals’’ item. The 0.077, indicating an acceptable fit of the model to the data.
lower loading item (0.70) was ‘‘sharing with people in Other indices of fit also showed the fit to be acceptable
training inside the firm.’’ Three items measured the (GFI ¼ 0.93, AGFI ¼ 0.90, CFI ¼ 0.97, NFI ¼ 0.95,
knowledge sharing in a small-group setting factor, and three NNFI ¼ 0.96), so we cannot reject the hypothesis that
items had substantial loadings (0.77–0.87): ‘‘private sharing the path model correctly reproduced the correlations
in small groups,’’ ‘‘sharing in the meetings,’’ and ‘‘coaching among the latent variables. The model is displayed in
my subordinate.’’ Fig. 2.
The pattern of direct effects revealed by the path model
5.2. The structural model seems to provide somewhat mixed evidence for the study’s
hypotheses. According to Hypotheses 1 and 2, we expected
In the next step, path models were fitted to test the positive effects of internal codified sources on knowledge
proposed model. The hypotheses do not specify explicitly sharing in small- and large-group settings. However,
the relations among all of the factors, and some unexpected according to the path model, there was no effect of the
relations were found, so a sequence of models was fitted. internal codified sources factor on any other latent
Selection criteria for the final model were (1) fit to data and variable. Internal codified sources had a 0.39 correlation
(2) interpretability of the estimated relations. with knowledge sharing in a small group setting and a 0.49
The path model had a w2 test-statistic of 194.66, df ¼ 77 correlation with knowledge sharing in a large group setting
(p ¼ 0.00). The RMSEA index was 0.065 for this model, in the CFA correlation matrix. Had the other factors not
ARTICLE IN PRESS
M.-S. Liu, N.-C. Liu / International Journal of Information Management 28 (2008) 423–432 429

Internal codified
sources H1
Knowledge sharing in
H2 a small group setting
Internal H3
non-codified
sources
H4
H5 Knowledge sharing in a
External codified large group setting
sources H6

H7

External H8
non-codified
sources

Fig. 1. Hypothesized integrated Model.

Internal codified
sources 0.07
Knowledge sharing in
0.21 a small group setting

Internal 0.33
non-codified
sources
-0.22

0.04 Knowledge sharing in a


External codified large group setting
sources 0.18

0.13

External 0.41
non-codified
sources

Fig. 2. Path model for relations Among the latent variables. Note: Path coefficients are standardized partial regression coefficients Estimated by maximum
likelihood. Ellipses represent latent variables. Unidirectional arrows Express direct effects. T-values above 1.96 (po0.05) are shown in bold. n ¼ 371.

been included in the model, this would have been the knowledge sharing in a small-group setting (0.34; po0.01)
estimate of the total effect. However, in the presence of and there was a negative direct effect on knowledge sharing
internal non-codified sources, external codified sources, in a large-group setting (0.22; po0.05). According to
and external non-codified sources, with which the internal Hypotheses 5 and 6, we expected positive effects of external
codified sources factor was correlated, no effect could be codified sources on knowledge sharing in small- and large-
seen. Further research is needed to clarify the relations group settings. However, according to the path model,
among the internal codified sources, internal non-codified there was no effect from the external codified sources factor
sources, external codified sources, and external non- on any other latent variable. According to Hypotheses 7
codified sources. and 8, we expected positive effects of external non-codified
According to Hypotheses 3 and 4, we expected a positive sources on knowledge sharing in small- and large-group
effect of internal non-codified sources on knowledge settings, but we observed only a direct effect of 0.41
sharing in a small-group setting and a negative effect of (po.001) on knowledge sharing in a large-group setting
internal non-codified sources on knowledge sharing in a (H8), and there was no effect from the external non-
large-group setting. The results of the path model support codified sources factor on knowledge sharing in a small-
Hypotheses 3 and 4: There was a positive direct effect on group setting (H7).
ARTICLE IN PRESS
430 M.-S. Liu, N.-C. Liu / International Journal of Information Management 28 (2008) 423–432

6. Discussion and implications studies. Knowledge sharing in a small-group setting within


the organization provides knowledge transmitters and
6.1. Discussion receivers with opportunities for two-way interaction,
facilitating the sharing of non-codified knowledge. In
The analysis above explains how different knowledge contrast, knowledge sharing in a large-group setting lacks
acquisition sources can influence the patterns of knowl- the level of interaction needed for the transfer of non-
edge-sharing behaviors. Employees use several channels to codified knowledge. In this situation, the transmitter
acquire and search for knowledge within the organization. cannot resolve the issue easily or quickly when the receiver
First, within the organization, employee knowledge acqui- has some question or problem, which tends to make
sition and sharing activities involve large amounts of tacit knowledge sharing problematic. The acquisition of knowl-
knowledge. Employees receive instructions from their edge from internal non-codified sources consequently
managers, privately ask their colleagues questions, and favors knowledge sharing in small-group settings, but does
receive information at internal meetings. Employees are not favor knowledge sharing in large-group settings.
able to obtain needed and useful knowledge, most of which When an organization cannot internally provide the
constitutes tacit knowledge, through these internal non- information needed to resolve a problem, employees may
codified channels. When employees acquire knowledge in turn to external sources to acquire knowledge. Our findings
these ways, they typically only are willing to share the suggest that personal networks and professional commu-
knowledge privately in small groups. However, when nities are important sources of knowledge for employees.
employees acquire valuable knowledge, most of which In contrast, professional books and online public informa-
constitutes tacit knowledge, from outside the organization tion are not commonly used as sources of knowledge.
via personal networks or professional communities, they External personal networks are important channels for the
often are more willing to share the information openly transfer of innovation knowledge (Assimakopoulos & Yan,
within the organization or with other employees (who may 2006). Apart from providing friendship and support,
not be members of the employee’s team or department). personal networks also facilitate the ongoing sharing of
Our research results show that neither internal nor information among individuals. In contrast to the two-way
external codified sources are very significant sources of relationships characterizing personal networks, profes-
knowledge for individuals. This may be due to the sional communities provide assistance via multiple channel
circumstances of R&D professionals, whose knowledge relationships.
acquisition and sharing behaviors differ from those of Employees may tend to use professional communities for
conventional workers (Assimakopoulos & Yan, 2006; two reasons. First, professional communities can provide a
Berends et al., 2006; Cabrera et al., 2006; Hsu, 2006). wide range of support. Large and diversified networks
R&D professionals are heavy users of tacit knowledge, and typically provide a higher level of support than small and
their knowledge-sharing behavior includes the acquisition homogeneous groups. It is not easy for most employees to
and sharing of large amounts of tacit knowledge. In maintain a large number of direct relationships with other
addition, the role of the R&D professional includes personnel, and it is also almost impossible to find out who
innovation, so the knowledge needed by these individuals in the interpersonal network possesses knowledge and
is not obtained readily from the organization’s formal experience connected with one’s job. In contrast to sources
documents. The knowledge needed for innovation emerges within the organization, professional communities bring
from combinations of tacit knowledge from different places together large numbers of professional personnel in the
within the organization. The knowledge acquisition and same area of specialization. The varied and important
sharing behavior of R&D professionals consequently backgrounds of these groups’ participants make the groups
involves mainly non-codified knowledge, which explains useful sources of technical knowledge and forums for
why neither internal nor external codified sources are continued discussion. The second reason is that the groups
significant sources of knowledge. promote efficient communication and high-quality discus-
In accordance with our findings, when employees sion. Employees tend to use telephone or e-mail to seek
encounter difficult technical or R&D problems, they first immediate help and interaction through their external
will seek assistance and answers from their team members. personal networks. These employees typically make one-to-
As a consequence, the acquisition of knowledge through one calls to their friends in what usually is a trial and error
communication channels among team members will facil- process, making the process quite time-consuming. Pro-
itate the private sharing of knowledge within the team. In blems can be disseminated to hundreds or even thousands
addition, employees seldom rely on internal formal training of professional personnel at the same time via professional
from technology transfer or other internal sources of communities, and responses can be obtained in a very short
codified knowledge to obtain the knowledge they need. It is time. Respondents may use a variety of formats to
worth noting that researchers have found a negative communicate with others, such as asking questions,
relationship between internal non-codified sources and discussions, trials of potential solutions, and the recom-
knowledge sharing in large-group settings. Our research mendation of persons who may possess relevant knowledge
results are consistent with the inferences made in past and experience or who may be interested. Thus, the quality
ARTICLE IN PRESS
M.-S. Liu, N.-C. Liu / International Journal of Information Management 28 (2008) 423–432 431

of discussion is superior to that in personal networks, in from SMEs. Second, this study employed a self-assessment
which friends may lack experience or interest. All of the of knowledge-sharing frequency. Finally, the subjective
participants in professional group discussions can accu- assessment of knowledge-sharing behavior may not express
mulate tacit knowledge not found in books. real knowledge-sharing performance objectively. Future
This study has addressed that individuals’ knowledge research should employ subjective and objective assessment
acquisition sources consist chiefly of internal and external standards. This research, for example, may evaluate self-
non-codified sources, and these sources affect knowledge assessment scores and R&D patent results simultaneously.
sharing in both small- and large-group settings. Three
major conclusions are summarized. First, what our findings
have sought to put forth is the view that the attributes of Acknowledgment
knowledge will influence individuals’ knowledge-sharing
behaviors. It seems that we start the work on depicting ‘the Data using in the early version of this paper were
edge of chaos’ of knowledge creation process in a firm. founded by the National Science Council, Taiwan (project
Second, external knowledge sources provide a wider range no.: NSC 90-2416-H-008-021).
of knowledge than internal sources, and a broader range of
knowledge facilitates innovation and R&D work. Due to
their need for abundant innovation-related knowledge, References
professional R&D personnel, in particular, benefit from the
Assimakopoulos, D., & Yan, J. (2006). Source of knowledge acquisition
clash of different ideas and points of view. Our findings
for Chinese software engineers. R&D Management, 31(1), 97–105.
explain that these personnel acquire useful knowledge Barney, J. B. (1991). Firm resources and sustained competitive advantage.
solely from highly homogenous internal sources. Third, the Journal of Management, 17(1), 99–120.
findings of this study indicate that employees who acquire Berends, H., Bij, H., Debackere, K., & Weggeman, M. (2006). Knowledge
knowledge from external sources are relatively willing to sharing mechanism in industrial research. R&D Management, 31(1),
share it openly within the organization, and there also will 85–95.
Bird, A. (1994). Careers as repositories of knowledge: A new perspective
be a relatively large number of receivers. As far as the on boundaryless careers. Journal of Organizational Behavior, 15(4),
organization is concerned, the sharing of external non- 325–344.
codified knowledge in a large-group setting offers more Bock, G. W., & Kim, Y. G. (2002). Breaking the myths of rewards: An
benefit than the sharing of internal non-codified knowledge exploratory study of attitudes about knowledge sharing. Information
in a small-group setting. Resource Management Journal, 15(2), 14–21.
Bock, G. W., Zmud, R. W., Kim, Y. G., & Lee, J. N. (2005). Behavioral
intention formation in knowledge sharing: Examining the role of
6.2. Implications extrinsic motivators, social–psychological force, and organizational
climate. MIS Quarterly, 29(1), 87–112.
If we take seriously the notion that firm concentrates on Bontis, N., Crossan, M., & Hulland, J. (2002). Managing an organiza-
tional learning system by aligning stocks and flows. Journal of
the innovation and creation, the important management
Management Studies, 39(4), 437–469.
issues become how managers can promote knowledge- Brown, J. S., & Duguid, P. (2001). Structure and spontaneity: Knowledge
sharing behaviors effectively upon considering the main and organization. In I. Nonaka, & D. J. Teece (Eds.), Managing
knowledge sources of the firm. When R&D knowledge is industrial knowledge: Creation, transfer and utilisation. London: Sage
generated and shared via everyday activities, R&D Publication.
managers must work hard to promote informal daily Burn, R. (2004). Structural holes and good ideals. American Journal of
Sociology, 110(2), 349–400.
discussion, which is more beneficial than directly over- Cabrera, E. F., & Cabrera, A. (2005). Fostering knowledge sharing
seeing the knowledge acquisition and sharing process. We through people management practices. The International Journal of
found that most employees prefer to acquire knowledge Human Resource Management, 16(5), 720–735.
from, and share knowledge with, their team members. This Cabrera, A., Collins, W. C., & Salgado, J. F. (2006). Determinants of
individual engagement in knowledge sharing. The International Journal
implies that R&D managers should promote greater
of Human Resource Management, 17(2), 245–264.
familiarity among team members and strengthen coopera- Cohen, W. M., & Levinthal, D. A. (1990). Absorptive capacity: A new
tive relationships fostering productivity. Furthermore, perspective on learning and innovation. Administrative Science
employees should be encouraged to participate in profes- Quarterly, 35, 128–152.
sional communities in order to acquire new knowledge. Dyer, J. H., & Nobeoka, K. (2000). Creating and managing a high-
Knowledge acquired via these channels will facilitate the performance knowledge-sharing network: The Toyota case. Strategic
Management Journal, 21, 345–367.
sharing of R&D knowledge. Granovetter, M. S. (1982). The strength of weak ties: A network theory
revisited. In P. Marsden, & N. Lin (Eds.), Social structure and network
7. Limitations analysis (pp. 105–130). Beverly Hills, CA: Sage.
Grant, R. M. (1996). Prospering in dynamically competitive environ-
This research was subjected to the following limitations. ments: Organization capability as knowledge integration. Organization
Science, 7, 375–387.
First, the sample used in this study consisted of large Hansen, M. T. (1999). The search-transfer problem: The role of weak ties
enterprises (with annual operating revenues in excess of in sharing knowledge across organization subunits. Administrative
US$3 billion). Future research may consider using data Science Quarterly, 44, 88–111.
ARTICLE IN PRESS
432 M.-S. Liu, N.-C. Liu / International Journal of Information Management 28 (2008) 423–432

Hsu, I. C. (2006). Enhancing employee tendencies to share knowledge— Senge, P. (1990). The fifth discipline: The art and practice of the learning
Case studies of nine companies in Taiwan. International Journal of organization. New York: Doubleday.
Information Management, 26, 326–338. Szulanski, G. (2000). The process of knowledge transfer: A diachronic
Kogut, B., & Zander, U. (1992). Knowledge of the firm, combinative analysis of stickiness. Organizational Behavior and Human Decision
capabilities, and the replication of technology. Organization Science, 3, Processes, 82, 9–27.
383–397. Tsai, W. (2002). Social structure of ‘‘coopetition’’ within a multiunit
Koivuaho, M., & Laihonen, H. (2006). A complexity theory approach to organization: Coordination, competition and intraorganizational
knowledge management—towards a better understanding of commu- knowledge sharing. Organization Science, 13(2), 179–190.
nication and knowledge flows in software development. The Electronic Tsoukas, H., & Vladimirou, E. (2001). What is organizational knowledge?
Journal of Knowledge Management Volume, 4(1), 49–58. Journal of Management Studies, 38(7), 973–993.
Laszlo, K. C., & Laszlo, A. (2002). Evolving knowledge for development: Uzzi, B. (1997). Social structure and competition in interfirm networks: The
The role of knowledge management in a changed world. Journal of paradox of embeddedness. Administrative Science Quarterly, 42, 35–67.
Knowledge Management, 6(4), 400–412. Wasko, M. M., & Faraj, S. (2000). It is what one does: Why people
Malhotra, N. K., & Birks, D. F. (2003). Marketing research: An applied participate and help others in electronic communities of practice.
approach. New York: Prentice-Hall. Joumal of Strategic Information Systems, 9(1), 155–173.
Nonaka, I., & Takeuchi, H. (1995). The knowledge-creating company.
New York: Oxford University Press.
Nonaka, I., & Konno, N. (1998). The concept of ‘ba’: Building a Min-Shi Liu is a doctoral candidate in the Institute of Human Resource
foundation for knowledge creation. California Management Review, Management at National Central University, Taiwan. She received her MS
40(3), 40–54. degree in Labor Research from National Chengchi University. Her current
Popadiuk, S., & Choo, C. W. (2006). Innovation and knowledge creation: research interests include knowledge management, organizational learning,
How are these concepts related? International Journal of Information and the impacts of knowledge flow on organizations and individuals.
Management, 26, 302–312.
Reio, T. G., & Wiswell, A. (2000). Field investigation of the relationship
among adult curiosity, workplace learning, and job performance. Nien-Chi Liu is an associate professor and director of the Institute of
Human Resource Development Quarterly, 11, 5–30. Human Resource Management at National Central University, Taiwan.
Rulke, D. L., Zaheer, S., & Anderson, M. H. (2000). Sources of managers’ Her research interests include high-performance work practices, human
knowledge of organizational capabilities. Organizational Behavior and resource management in knowledge intensive firms, and employee
Human Decision Processes, 82, 134–149. ownership. Her articles have been accepted for publication in Management
Sacks, M. (1994). On-the-job learning in the software industry: Corporate and Organization Review, IEEE Transactions on Engineering Management,
culture and the acquisition of knowledge. Westport, Conn.: Quorum The Nordic Journal of Political Economy, and other refereed journals and
Books. conference proceedings.

You might also like