Motivations of employees’ knowledge sharing behaviors A self-determination perspective

Motivations of employees ’knowledge sharing

behaviors:A self-determination perspective

Wei-Tsong Wang a,?,Ya-Pei Hou b,1

a

Department of Industrial and Information Management,National Cheng Kung University,1st University Road,Tainan 701,Taiwan b Institute of Information Management,National Cheng Kung University,1st University Road,Tainan 701,Taiwan

a r t i c l e i n f o a

b s t r a

c t

Article history:Received 13January 2014Received in revised form 18November 2014Accepted 25November 2014Available online 22December 2014Prior studies on knowledge-sharing motivations mostly concentrate on

discussing motivation in terms of level or amount,and thus,discussions

regarding the quality of motivations,in terms of their levels of autono-

my,are scarce.Additionally,while researchers have addressed the

signi ?cant relationships among different types of motivations,there is

still controversy concerning these relationships in a knowledge-sharing

context.With reference to self-determination theory,this study examines

a model that depicts the in ?uence of various types of motivations on em-

ployees ’knowledge sharing behaviors (KSBs).Based on the data collected

from 259employees in 34organizations,hard reward,soft reward,and

altruism for organizational bene ?ts are signi ?cant in ?uencing factors of

KSBs,while altruism for personal satisfaction is not.Additionally,soft re-

ward has a signi ?cant positive effect on both altruism for organizational

bene ?ts and altruism for personal satisfaction.The theoretical and

practical implications and suggestions for future research are discussed.

?2014Elsevier Ltd.All rights reserved.Keywords:Knowledge sharing behavior Self-determination theory Quality of motivations Autonomous motivation

1.Introduction

While knowledge sharing is vital to the development of competitive advantages of organizations and is integral to knowledge management (KM),managers often ?nd it to be the most dif ?cult KM practice to promote in various social contexts because knowledge sharing does not come naturally to most individuals (Bock,Kankanhalli,&Sharma,2006).For example,organizations ’?nancial-based incentives may discourage rather than encourage knowledge sharing if employees perceive that knowledge sharing will hinder their su-pervisors from distinguishing their personal efforts from those of their competing associates/coworkers Information and Organization 25(2015)1–26

?Corresponding author.Tel.:+88662757575x53122;fax:+88662362162.

E-mail addresses:wtwang@https://www.360docs.net/doc/768373305.html,.tw (W.-T.Wang),wheat_cure@https://www.360docs.net/doc/768373305.html, (Y.-P.Hou).

1Tel.:+88662757575x53120111;fax:+8866

2362162.

https://www.360docs.net/doc/768373305.html,/10.1016/https://www.360docs.net/doc/768373305.html,andorg.2014.11.001

1471-7727/?2014Elsevier Ltd.All rights

reserved.

Contents lists available at ScienceDirect

Information and Organization

j o u r n a l h o m e p a g e :w w w.e l s e vi e r.c o m /l o c a t e /i n f o a n d o r g

2W.-T.Wang,Y.-P.Hou/Information and Organization25(2015)1–26

(Bock,Zmud,Kim,&Lee,2005).Thus,organizations often?nd that it is dif?cult to facilitate effective knowl-edge sharing practices among employees due to human factors,institutional complexities,and the sticky na-ture of knowledge itself(Liu&Liu,2008;Yang&Wu,2008).There have been studies indicating the sticky nature of knowledge as a barrier to knowledge sharing because it causes the sharing processes among individ-uals to be slow,costly,and uncertain(Liu&Liu,2008).Achieving effective knowledge sharing practices thus depends on individuals’willingness to put signi?cant effort into the associated social processes(Chang& Chuang,2011).Additionally,from an economic perspective,personal behaviors are primarily motivated by self-interest,and people therefore tend to do their best to maximize individual utility(Bock&Kim,2002; Yang&Wu,2008).This implies that if an individual possesses particular knowledge that is valuable to him-or herself,the individual would be unlikely to share it with others to maximize personal bene?ts,such as in-creased job security and continued possession of a unique and strong position in the organization(Bock& Kim,2002;Yang&Wu,2008).In some cases,the costs instead of the bene?ts of knowledge sharing play an important role in determining knowledge sharing behaviors(Kankanhalli,Tan,&Wei,2005).

Prior research recognizes the importance of examining personal motivational factors that facilitate or restrain knowledge sharing behaviors,and many studies have intensively investigated these issues in various contexts(Bakker,Leenders,Gabbay,Kratzer,&van Engelen,2006;Chang&Chuang,2011;Cruz,Perez,& Cantero,2009;Foss,Minbaeva,Pedersen,&Reinholt,2009).Self-determination theory(SDT)is recognized as a well-established theory of motivation and has been widely adopted to investigate how and why a partic-ular human behavior is induced(Deci,Koestner,&Ryan,1999;Deci&Ryan,1985).SDT proposes that human behaviors may be encouraged not only by externally induced incentives,called controlled motivation,but also by internally evoked incentives,known as autonomous motivation.SDT proposes that a speci?c type of motivation can be considered as having a low level of quality if it is experienced by recipients as an attempt to control their behavior.In other words,SDT indicates that autonomy-oriented motivations have relatively higher levels of quality than control-oriented motivations in terms of facilitating a particular behavior.

Previous models of knowledge-sharing motivation mostly concentrate on discussing motivation in terms of level or amount,and thus,the discussions regarding the quality of motivations are scarce(Gagne,2009).To be speci?c,the autonomy-oriented motivations proposed by SDT tend to be either ignored or falsely integrat-ed with the control-oriented motivations in knowledge sharing studies that adopt cost-bene?t approaches (Kankanhalli,Lee,&Lim,2011;Kankanhalli et al.,2005;Lin&Huang,2010;Wasko&Faraj,2005).Therefore, our?rst objective is to identify and simultaneously examine key autonomy-oriented and control-oriented motivations of individuals’knowledge sharing behaviors by consulting the controlled-to-autonomous contin-uum of motivations proposed by SDT to better understand how various types of motivations with different degrees of autonomy may affect individuals’knowledge sharing behaviors.Additionally,it remains inconclu-sive whether some types of motivations having different levels of quality in?uence individuals’knowledge sharing behavior in the workplace more than the others.Consequently,our second objective is to examine the difference in the effects of various types of motivations on individuals’knowledge sharing behaviors. Finally,prior research mostly conceptualizes various types of motivations by adopting the perspective of the distinction between intrinsic(i.e.,autonomy-oriented)and extrinsic(i.e.,control-oriented)motivations, and it indicates the signi?cant effects of different types of extrinsic motivations on intrinsic motivations (e.g.,Deci et al.,1999;Roca&Gagne,2008;Sheldon,Turban,Brown,Barrick,&Judge,2003;Yoo,Han,& Huang,2012).However,there is still controversy concerning such effects in the area of knowledge sharing (e.g.,Cockrell&Stone,2010;Foss et al.,2009;Gagne,2009).A lack of a holistic understanding of how various knowledge-sharing motivations that have different quality levels,in terms of their degrees of autonomy,on employees’knowledge sharing behaviors in?uence one another would hinder an organization’s ability to plan for,evaluate,and justify their efforts to enhance employees’knowledge sharing behaviors.Thus,our third objective is to investigate the effects of the control-oriented motivations on the autonomy-oriented motivations regarding employees’knowledge sharing behaviors.

2.Theoretical background and hypothesis development

2.1.Self-determination theory

The core of SDT is that individuals may be motivated to perform certain behaviors both externally (i.e.,controlled motivation)and internally(i.e.,autonomous motivation)(Cockrell&Stone,2010).

Autonomous motivation refers to the incentives based on which individuals perform certain behaviors that do not contribute to their core-self needs and bene ?ts.Individuals feel autonomously motivated when they perceive self-determination in selecting their objectives freely based on self-interests,curiosity,care,or abiding values.The internal/autonomous motivating factors include self-interests,enjoyment,and sel ?ess care for other parties.In contrast,controlled motivation refers to the incentives based on which individuals perform certain behaviors that explicitly contribute to their core-self needs and bene ?ts.Individuals feel that they are being controlled with regard to performing certain behaviors when they sense the pressure or necessity of performing these behaviors to achieve desirable outcomes.The external/controlled motivating factors include reward systems,formal or informal evaluations from others,and status within signi ?cant groups.

The results of a literature review reveal four premises regarding the application of SDT (Cockrell &Stone,2010;Gagne &Deci,2005;Sheldon et al.,2003).First,SDT asserts the distinction between autonomous and controlled motivations in terms of their underlying regulatory processes and associated experience,and thus suggests that behaviors may be characterized to be self-determined versus non-self-determined in terms of the degree to which they are stimulated by autonomous versus controlled motivations.Second,the quality of a speci ?c type of motivation depends on how well it helps satisfy individuals ’universal psychological needs for autonomy (to be self-regulating regarding performing a behavior),competence (to be effective in what they do),and relatedness (to feel connected and in sympathy with signi ?cant others).Prior studies that adopt SDT indicate that autonomous motivations are more likely to result in more positive outcomes than controlled motivations in various contexts because autonomy-oriented motivations tend be more positively associated with the satisfaction of these three psychological needs,while control-oriented motivations tend to produce pressure (e.g.,Grolnick &Ryan,1987;Mitchell,Gagne,Beaudry,&Dyer,2012).Consequently,the more a speci ?c motivation is autonomy-oriented,the higher its quality is in terms of induc-ing a speci ?c behavior.Third,autonomous and controlled motivations may interact with each other regarding their in ?uences on individuals ’intention to perform a speci ?c behavior.To be speci ?c,one type of motivation may strengthen or undermine another type of motivation,depending on the manner in which the former is used with respect to ful ?lling the three psychological needs discussed above.

The last premise of SDT is the proposition of a controlled-to-autonomous continuum (i.e.,a self-determination continuum),as presented in Fig.1.This continuum is used to describe the degree to which an externally regulated motivation (i.e.,fully controlled motivation),which is initiated and maintained by entities external to an individual (e.g.,a ?nancial reward for an employee ’s performance),has been internalized (e.g.,the employee works hard to enhance performance even when the ?nancial reward does not exist).Greater internalization of externally regulated motivation leads to subsequent extrinsically motivated behavior that is more autonomous.Therefore,based on this controlled-to-autonomous continuum,motivations are further divided into three essential categories (Gagne &Deci,2005;Ryan,Lunch,Vansteenkiste,&Deci,2011).The ?rst category is termed amotivation,which represents a lack of motivation (i.e.,individuals have little or no perceived values,incentives,and competence for action).The second catego-ry is termed extrinsic motivation,which includes four subcategories.First,the controlled motivation/external regulation are associated with the situation in which individuals feel they are being controlled or pressured by forces/incentives from the outside (i.e.,external regulation)to perform the behavior.Second,the moderately controlled motivation/introjected regulation means that individuals feel they are being controlled or pressured to perform the behavior by the implicit consequences of performing or not performing the behavior (e.g.,pride,personal reputations,and relationship with others),even when they have not truly recognized the value of the behavior.Third,the moderately autonomous motivation/identi ?ed regulation indicates that individuals feel greater freedom regarding the regulation because they recognize the value of the behavior and willingly accept responsibility for regulating the behavior.The last subcategory is the autonomous motivation/integrated regulation.Individuals sense that they are integrated-regulated when they feel the regulation to be truly autonomous because they consider the value of performing the behavior as ?tting within their personal goals and values.Finally,the last category is termed intrinsic motivation,which has only one subcategory,inherently autonomous motivation.When individuals are autonomously motivated,they engage in a particular behavior because of its inherent satisfactions (i.e.,the behavior is itself interesting and enjoyable).

It is important to note that in SDT,the model of internalization is not a stage theory,meaning that individuals do not have to move through these stages of motivations that are included in the indicated

3

W.-T.Wang,Y.-P.Hou /Information and Organization 25(2015)1–26

Fig.1.The controlled-to-autonomous continuum of motivations in SDT(Deci&Ryan,2000;Gagne&Deci,2005).4 W.-T.

Wang,

Y.-P.

Hou

/

Information and Organization 25 (2015) 1

–26

controlled-to-autonomous continuum with regard to particular behaviors,and various types of motivation with different degrees of autonomy can take effect simultaneously regarding motivating a particular behavior (Gagne &Deci,2005;Ryan &Connell,1989).Additionally,SDT indicates that controlled-oriented motivations regarding a behavior can be internalized by the individuals,and thus be transformed into more autonomy-oriented motivations.For example,an employee may change from the state of working hard when my supe-rior is watching me to avoid punishments (i.e.,an external regulation)to the state of working hard even when the superior is not watching to maintain a positive reputation in my organization (i.e.,an introjected regulation).In this case,superior ’s surveillance on an employee is transformed from a controlled motivation to a moderately controlled motivation in the mind of the employee.

2.2.Motivations of knowledge sharing behaviors:An SDT perspective

2.2.1.Summary of motivations adopted in existing knowledge sharing studies

Prior studies indicate that motivations associated with individuals ’expectations of favorable out-comes are required to encourage individuals to engage in a particular behavior (Bandura,1986;Bandura,1997;Compeau &Higgins,1995a;Compeau,Higgins,&Huff,1999;Gagne &Deci,2005;Ryan et al.,2011).According to SDT,the motivation-related constructs adopted in prior knowledge sharing studies mostly refer to the motivations initiated and sustained by forces/entities external to individuals ’minds,such as external regulations/controlled motivations (e.g.,rewards,punishments,and promotion)and introjected regulations/moderately controlled motivations (e.g.,personal reputation and self-esteem contingent on performance).This observation implies that the concepts of autonomy-oriented motivations,namely the identi ?ed regulation and autonomous motivation,tend to be either ignored or falsely integrated with the concept of intangible bene ?ts/rewards in knowledge sharing studies (Compeau &Higgins,1995b;Compeau et al.,1999;Hsu,Ju,Yen,&Chang,2007;Lin &Huang,2008;Lin,Hung,&Chen,2009)or the cost-bene ?t approaches in general (Kankanhalli et al.,2005;Kankanhalli et al.,2011;Lin &Huang,2010;Wasko &Faraj,2005).Appendix 1presents a summary of the motivational factors examined in these studies.

In reference to SDT,the summary presented in Appendix 1indicates that although prior knowledge-sharing studies have examined various types of motivational factors to interpret and explain knowledge sharing-related behaviors/intentions,the conceptualization of the distinctions among these motivational factors is doubtful and incongruous,as discussed earlier.Additionally,to the best of our knowledge,none of these studies has simultaneously adopted the four primary types of motivations in SDT to examine their in ?uences on knowledge sharing behaviors.Moreover,in a similar vein to these knowledge sharing studies,several studies that adopt SDT indicate that autonomous motivations play a more important role than controlled motivations in facilitating knowledge sharing behaviors in various contexts (Cockrell &Stone,2010;Cruz et al.,2009;Grolnick &Ryan,1987;Mitchell et al.,2012).Autonomous motivations are based on the value that exists within the mind of an individual and may thus lead to sustained passions and efforts regarding speci ?c behaviors.However,the results of prior knowledge sharing studies (see Appendix 1)regarding the in ?uences of various types of motivations are inconsis-tent and inconclusive,contingent on the research contexts.In light of this discussion,this study will simultaneously adopt four primary types of motivations proposed by SDT,namely,controlled motivation,introjected regulation,identi ?ed regulation,and autonomous motivation,to investigate the motivational factors that signi ?cantly in ?uence individuals ’knowledge sharing behaviors.This approach enables us to incorporate the concepts of the quality of motivations depicted in SDT to examine the effects of both control-oriented and autonomy-oriented motivations on employees ’knowledge sharing behaviors and to investigate the differences among these effects.

2.2.2.Control-oriented motivations and knowledge sharing behaviors

Social cognitive theory (SCT)implies that an individual ’s personal-and performance-related outcome expectations,as forms of personal motivations,regarding a given behavior will be developed with reference to anticipated consequences or incentives associated with the behavior (Compeau &Higgins,1995b;Compeau et al.,1999;Hsu et al.,2007;Lin &Huang,2008).The key factors in personal-and performance-related outcome expectations are the bene ?ts/rewards that accrue for individuals due to their knowledge exchange/sharing behaviors (Hsu et al.,2007).This argument implies that people ’s behaviors are dependent 5

W.-T.Wang,Y.-P.Hou /Information and Organization 25(2015)1–26

6W.-T.Wang,Y.-P.Hou/Information and Organization25(2015)1–26

on the controlled motivations considered in SDT,such as tangible and intangible rewards(Gagne&Deci, 2005).In addition to the concept of outcome expectations proposed by SCT,a prevalent perspective adopted for comprehending the motivation systems regarding knowledge sharing behaviors is the distinction between intrinsic/soft rewards(e.g.,personal reputation and relationships with signi?cant others)and extrinsic/hard rewards(e.g.,reciprocity,?nancial rewards/bene?ts,promotion,and other bene?ts in the workplace)(Cruz et al.,2009;Kankanhalli et al.,2005;Kankanhalli et al.,2011).Conse-quently,by consulting the concept of the controlled motivation in SDT and the existing motivation frameworks,the constructs of hard and soft rewards are adopted in this study to represent external regulations/controlled motivations and introjected regulations/moderately controlled motivations in SDT.

Hard rewards are de?ned as individuals’expectations of obtaining explicit outcomes(e.g.,promotion,?nancial rewards,reciprocity,and other explicit bene?ts in the workplace)in return for performing knowledge sharing behaviors,whereas soft rewards are de?ned as individuals’expectations of achieving implicit outcomes(e.g.,personal reputation and relationships with signi?cant others)in return for performing knowledge sharing behaviors(Hall&Graham,2004;Hummel et al.,2005;Kankanhalli et al., 2005;Kankanhalli et al.,2011).Based on SDT,the hard rewards are inherently externally initiated and sustained,and are thus forms of external regulations/controlled motivations.Foss et al.(2009)argue that hard rewards represent external contingencies administered by others based on an employee’s behaviors and thus regulate the employee’s behaviors.For example,employees’reception of positive feedback when others use their contributions(i.e.,reciprocity),can be an indication of the positive in?uence of the employees’work on the organizational performance,thus increasing their perceived ef?cacy(i.e.,competence) of their efforts(Cabrera&Cabrera,2002;Gagne,2009).

In addition,soft rewards may make individuals feel implicitly controlled or pressured to perform the behavior due to the implicit consequences related to the behavior,and are thus forms of introjected regulations/moderately controlled motivations(Gagne,2009).Gagne(2009)also indicates that soft rewards as a result of knowledge sharing behaviors,may satisfy employees’needs to be socially acceptable in an organizational context.For example,employees’outcome expectations,including improved work relation-ships with others(i.e.,relatedness)and self-image/reputation(i.e.,competence),can be considered forms of soft rewards that are positively associated with sharing intentions and behaviors(Bock&Kim,2002; Kankanhalli et al.,2011).

Prior studies have addressed the importance of control-oriented motivations in facilitating individuals’knowledge sharing behaviors(Cruz et al.,2009;Hall,2001;Hall&Graham,2004;He&Wei,2009; Kankanhalli et al.,2005;Staples&Webster,2008).For example,Brachos,Kostopoulos,Soderquist,and Prastacos(2007)propose that motivations such as awards,reciprocity,?nancial incentives(i.e.,hard rewards),and recognition/reputation(i.e.,soft rewards)are a powerful force that encourages interpersonal knowledge sharing activities.In addition,extrinsic motivations,such as?nancial rewards and promotion (i.e.,hard rewards)and improved self-image(i.e.,soft rewards),are integral components of a KM initiative and thus tend to have a positive impact on knowledge sharing under the condition that individuals perceive a high knowledge storage capability of their business units/organizations(Chae&Bloodgood,2006; Kankanhalli et al.,2011).Therefore,the following hypotheses are developed:

H1a.Hard rewards positively in?uence individuals’knowledge sharing behaviors.

H2a.Soft rewards positively in?uence individuals’knowledge sharing behaviors.

2.2.

3.Autonomy-oriented motivations and knowledge sharing behaviors

However,with reference to SDT,the usefulness of the perspective of soft and hard rewards is limited in terms of understanding knowledge sharing behaviors as responses to motivational factors related to autonomy-oriented motivations,such as personal interests/satisfactions and importance for sel?ess goals of individuals.This phenomenon occurs because the concepts of identi?ed regulation and autonomous motiva-tion are either missing or inappropriately aggregated into the category of intrinsic/soft rewards in most prior knowledge-sharing studies,whereas they deserve to be considered independently.In addition,there have been studies indicating that both expectations of personal bene?ts and of bene?ts for signi?cant third parties (e.g.,valued organizations,communities,and other individuals)may motivate knowledge sharing behaviors

(Chiu,Hsu,&Wang,2006;Chou &He,2011;He &Wei,2009;Hsu et al.,2007;Tsai &Cheng,2010).The various stimulants of these sel ?ess knowledge-sharing behaviors have also been related to the concept of altruism (Chang &Chuang,2011).

Altruistic knowledge sharing behaviors may be induced by the autonomous motivation that arises within the self (Cockrell &Stone,2010).Consequently,with reference to SDT,the two constructs altruism for organizational bene ?ts (e.g.,contributions to the organizations)(Chiu et al.,2006;Hsu et al.,2007;Hummel et al.,2005)and altruism for personal satisfaction (e.g.,enjoyment of helping others)(Chang &Chuang,2011;Hsu &Lin,2008;Lin,2007;Wasko &Faraj,2005)are identi ?ed and adopted in this study to represent the concepts of identi ?ed regulation and autonomous motivation in SDT,respectively.Altruism for organizational bene ?ts is de ?ned as individuals ’evaluations of the positive in ?uence of their efforts in performing altruistic knowledge sharing behaviors in their organizations (Chiu et al.,2006)and is considered a form of moderately autonomous motivations (i.e.,identi ?ed regulations).Additionally,altruism for personal satisfaction is de ?ned as individuals ’inherent pleasure in seeing the positive results of their efforts in performing altruistic knowledge sharing behaviors (Hall,2001)and is considered a form of inherently autonomous motivations.With reference to SDT,both altruism for organizational bene ?ts and altruism for personal satisfaction are similar,but distinct from the term intrinsic motivations ,which is frequently used in the existing knowledge sharing literature,such as the studies summarized in Appendix 1(e.g.,Cruz et al.,2009;Hung,Durcikova,Lai,&Lin,2011).To be speci ?c,the term intrinsic motivations is used in existing litera-ture as a general expression referring to motivational factors that are relatively more autonomy-oriented rather than control-oriented,regardless of their difference in the degrees of autonomy that are further classi ?ed as moderately autonomous motivations,autonomous motivations,and inherently autonomous motivations in SDT.

Deci and Ryan (2000)argue that when a type of behavior is not inherently interesting or enjoyable for an individual,the main reason the individual performs such a behavior is because this behavior is valued by and/or bene ?cial to signi ?cant external entities to whom he or she feels connected,such as his or her organization.Bock and Kim (2002)also indicate that making a signi ?cant contribution to organizational performance,regardless of the presence of tangible or intangible rewards,can contribute to the increased degree of competence of employees,and are thus positively related to the employees ’knowledge sharing intentions.In addition,altruism for organizational bene ?ts may reduce excessive interpersonal competition and facilitate collaboration that can lead to more positive interpersonal relationships (i.e.,need for relatedness),and create employees ’desire for self-improvement for supporting their organizations (i.e.,need for competence),thus promoting knowledge sharing behaviors (Cruz et al.,2009).Moreover,employees who are able to obtain feedback on previous instances of knowledge sharing behaviors are more likely to understand how such behaviors may positively in ?uence the works of others or the overall organizational performance,and they tend to develop a sense of self-worth (Bock et al.,2005;Kamhawi,2010).This understanding may then encourage them to continue to share their knowledge with others in the future.

Prior studies also argue that on some occasions,individuals are willing to share their knowledge because they feel satis ?ed or pleased by seeing the positive results of helping others (i.e.,altruism for personal satisfaction)(Hall,2001;Hall &Graham,2004;Lin,2007).Additionally,altruism for personal satisfaction involves individuals ’desire to perform self-learning for pursuing personal knowledge growth (i.e.,needs for autonomy),and is thus positively associated with their knowledge sharing intentions (Kankanhalli et al.,2011).Furthermore,Gagne (2009)also implies that when employees feel competent,autonomous,and related to their colleagues with whom they have opportunities to share knowledge,they are likely to value and enjoy (i.e.,be intrinsically motivated)to perform knowledge sharing behaviors that can bene ?t those colleagues.Finally,Chang and Chuang (2011)argue that when individuals develop a sense of belonging to a virtual community,they are likely to have shared visions or goals with other community members.Therefore,they tend to increase their altruistic knowledge sharing behaviors because they feel pleased by simply helping other members (i.e.,altruism for personal satisfaction)and/or by seeing the achievement of the shared vision or goals (i.e.,altruism for organizational bene ?ts).Therefore,the following hypotheses are developed:

H3.Altruism for organizational bene ?ts positively in ?uences individuals ’knowledge sharing behaviors.H4.Altruism for personal satisfaction positively in ?uences individuals ’knowledge sharing behaviors.7

W.-T.Wang,Y.-P.Hou /Information and Organization 25(2015)1–26

8W.-T.Wang,Y.-P.Hou/Information and Organization25(2015)1–26

2.3.The relationships between control-oriented and autonomy-oriented motivations

SDT research indicates the signi?cant effects of control-oriented motivations(i.e.,external regulations and introjected regulations)on autonomy-oriented motivations(i.e.,identi?ed regulations and autonomous motivations).For example,Newby and Alter(1989)indicate that extrinsic-driven motivators regarding employees’adoption of e-learning in organizations may hinder the employees from participating in e-learning activities that are intrinsically challenging and enjoyable.Vansteenkiste et al.(2007)also argue that extrinsically oriented people tend to be driven by obtaining external indicators of worth(e.g.,external rewards),and therefore they often ignore their personal wants and interests.Additionally,a number of studies indicate that external controls(i.e.,controlled motivations/external regulations)on a particular behavior,such as rewards,evaluation,and threats,can diminish people’s autonomy-oriented motivations because the former may thwart people’s sense of autonomy regarding the behavior(Cockrell&Stone, 2010;Deci&Ryan,2000;Deci et al.,1999;Lian,Ferris,&Brown,2012;Ryan&Connell,1989).It is clear that autonomy-oriented motivations may interact with control-oriented motivations in facilitating a particular behavior.However,as presented in Appendix1,these interactions tend to be ignored in existing knowledge sharing studies.This insuf?cient understanding of the relationships between control-oriented and autonomy-oriented motivations inevitably presents challenges for organizations to develop motivational strategies to promote employees’knowledge sharing behaviors.We speci?cally discuss these issues in this study to eliminate this research gap.

One can further comprehend the relationship between control-oriented and autonomy-oriented motivations with reference to satisfaction of the basic psychological needs for autonomy,relatedness,and competence proposed by SDT.Sheldon et al.(2003)argue that extrinsic rewards(e.g.,hard rewards)may undermine employees’autonomy-oriented motivations when they are used by the authorities of an organization as a means of over-controlling their behaviors(i.e.,controlled motivations),but rewards that are experienced as providing information and praise and thus satisfy individuals’psychological needs for competence,may positively impact autonomy-oriented motivations.Ryan et al.(2011)also state that individuals can simultaneously have multiple motivations that vary in their individual degrees of autonomy, resulting in these individuals feeling more or less autonomous overall with respect to a behavior.These authors further argue that external controls(e.g.,hard rewards)are negatively associated with individuals’sense of autonomy because they discourage individuals from basing their actions on their own reasons, while feeling connected with and being signi?cant to others(i.e.,soft rewards)increases individuals’autonomy-oriented motivations because they contribute to the satisfaction of the individuals’need for relatedness and competence.

The relationships between control-oriented and autonomy-oriented motivations mentioned above are evident in various research areas.For example,in the area of technology adoption,academics also indicate that intangible bene?ts,such as favorable interactions with signi?cant others and supervisor/organizational support(i.e.,soft rewards)can enhance autonomy-oriented motivations because they help in satisfying the basic psychological needs for autonomy,relatedness,and competence,while external controls(e.g.,hard rewards)can diminish individuals’autonomy-oriented motivations because control-oriented motivations do not positively contribute to the satisfaction of these basic psychological needs(Mitchell et al.,2012;Roca& Gagne,2008;Yoo et al.,2012).In a similar vein,prior knowledge sharing research that adopts SDT has proposed that performance appraisals regarding a particular behavior must have a developmental focus(e.g.,unexpected informal positive feedback from others)rather than an evaluative focus(e.g.,expected formal evaluation from supervisors)because developmental appraisals(introjected regulations)enhance autonomy-oriented motivations,while evaluative appraisals(controlled motivations)decrease autonomy-oriented motivations (Gagne,2009).This is because the more individuals sense that the appraisals regarding a behavior are delivered in a controlled manner,the more their perceived locus of causality regarding their behavior shifts from a self-perception of being the initiator of their own behavior to behaving for external reasons(i.e.,being controlled) (Foss et al.,2009).This observation indicates that while controlled motivations(e.g.,hard rewards)diminish autonomy-oriented motivations,introjected regulations(e.g.,soft rewards)may enhance autonomy-oriented motivations.Based on this discussion,the following hypotheses are developed:

H1b.Hard rewards negatively in?uence individuals’motive for altruism in achieving organizational bene?ts.

F i

g

.2

.T

h e r

e s e a r c h m

o d e l o f

k

n o w l

e d g e s

h a r i n g b

e h a v i o r s .9

W.-T.Wang,Y.-P.Hou /Information and Organization 25(2015)1–26

10W.-T.Wang,Y.-P.Hou/Information and Organization25(2015)1–26

H1c.Hard rewards negatively in?uence individuals’altruism in ful?lling personal satisfaction.

H2b.Soft rewards positively in?uence individuals’motives for altruism in achieving organizational bene?ts.

H2c.Soft rewards positively in?uence individuals’altruism in ful?lling personal satisfaction.

3.Research model and research method

Based on the discussion above,we develop a theoretical model that depicts the effects of various types of motivational factors on employees’knowledge sharing behaviors based on SDT.We also included our survey re-spondents’age,gender,and level of education as control variables for the latent construct of knowledge sharing behaviors to capture the potential effects of demographic factors on employees’knowledge sharing behaviors. Additionally,we use our survey respondents’departments as a control variable for the construct of knowledge sharing behaviors to examine the potential effects of the differences in interpersonal interaction orientation,in the rewarding policies regarding employees’work-related performance,and in the features(i.e.,voluntary versus mandatory)of employees’knowledge sharing behaviors of different departments(i.e.,functional areas)on the employees’actual knowledge sharing behaviors.The proposed research model is presented in Fig.2.

3.1.Development of instruments

A survey of professionals in?nancial services organizations was conducted to test the proposed model.To develop an effective survey,measurement items found in the related studies were re?ned and then used.All the items were measured using a seven-point Likert scale ranging from(1)strongly disagree to(7)strongly agree.

The preliminary questionnaire had26items and was pilot-tested with30individuals who were working full-time for?nancial services companies.The questionnaire was further re?ned based on the results and feedback from the pilot test.The?nal questionnaire consisted of18items for measuring the?ve constructs in the research model.The survey items were considered highly reliable because the Cronbach’s alpha coef?cients of all the constructs(ranging from0.91to0.94)reached the recommended level of0.7.Please refer to Appendix2for the?nal questionnaire and references.

3.2.Data collection

The data for this study were collected using a survey administered in Taiwan in2010.A list of the Top 100?nancial services companies was acquired,and representatives of these companies were contacted by phone and/or personal visit to explain the purpose of the research and to inquire whether the compa-ny would be willing to participate.Finally,34companies agreed to participate in this study,including19 commercial banks,10insurance companies,and?ve securities/?nancial holding companies.The names of these companies are withheld due to a non-disclosure agreement with their executives.The liaisons from these participating companies were asked to randomly distribute the questionnaires.A total of626 questionnaires were distributed,and484(77.3%)questionnaires were received.Among the responses received, 225incomplete or problematic ones were later removed,yielding259valid responses and a valid return rate of 53.5%.

Potential nonresponse bias was assessed by comparing the early respondents with the later ones based on demographic variables,including gender,age,and level of education,using independent sample t tests.The re-sults indicated no statistically signi?cant differences between these two data sets in terms of gender(p=0.45), age(p=0.97),or level of education(p=0.82).We also performed a set of independent sample t tests to compare the valid responses(259samples)and incomplete/problematic responses(225samples)based on demographic variables,including gender,age,and level of education.The results indicated no statistically signi?cant differences between these two data sets in terms of gender(p=0.95),age(p=0.79),or level of education(p=0.68).Therefore,nonresponse bias was not determined to be a serious concern.In addition, because a nonrandom sampling method was used,tests of homogeneity on the corporation type and department/job characteristics of the respondents were performed using one-way ANOVA.The mean score of

all the survey items were indifferent (p N 0.05)among respondents across different corporation types (p values ranged from 0.11to 0.95)and departments/job characteristics (p values ranged from 0.17to 0.98).Consequent-ly,the 259valid survey responses were used as a single data set in the subsequent analysis.

To eliminate the concern of common method bias in the survey design,questionnaire items were arranged to counterbalance the order of the measurement of the dependent and independent variables (Lin &Huang,2010).Speci ?cally,items for dependent variables were arranged following rather than preceding the items for the independent ones,and the order of the items was arranged randomly in the questionnaire.In addition,the presence of common method variance (CMV)was assessed using statistical measures.A recent study summarizes the pros and cons of various popular methods for assessing CMV and implies that so far there is no universally viable means of addressing this issue (Chin,Thatcher,&Wright,2012).However,the methods discussed in the works of Podsakoff,MacKenzie,Lee,and Podsakoff (2003)and Williams,Edwards,and Vandenberg (2003)have been found to be the most widely used among information management researchers.Therefore,one of the approaches suggested by Podsakoff et al.(2003),which tests for CMV using a single unmeasured latent method factor,is adopted in this study.This method suggests adding a ?rst-order factor to the theoretical model with all the measures as indicators,which may be referred to as a latent method factor (LMF).By following the procedures of Liang,Saraf,Hu,and Xue (2007),the LMF modeling was conducted using the component-based structured equation modeling technique,namely the partial least squares (PLS).In this approach,each indicator is converted into a single-indicator construct,which makes all major constructs of interest second-order constructs.An LMF is added to the theoretical model as a second-order construct by linking it to all the ?rst-order single-indicator constructs.Although the construct of trust was treated as a formative construct when testing the research model of this study,it was modeled as a re ?ective construct when assessing CMV to ensure the interpretability of the results,as done in previous studies (Herath &Rao,2009).Next,each indicator ’s variance substantively explained by the constructs of interest and by the method was calculated.As shown in Appendix 3,the average substantively explained variance of all the indicators is 0.850,whereas the average method-based variance is 0.009.The ratio of substantive variance to method variance is approximately 94:1.In addition,most of the method factor loadings are not signi ?cant.Because the method variance identi ?ed was found to be insigni ?cant and small in magnitude,it is concluded that the common method bias is unlikely to be a serious concern for this study.

The demographic pro ?le of the sample population indicates that 52.5%of the respondents were males,the majority of whom were between 21and 40years old (80.0%)and had a college-level education or above Table 1

Convergent validity for the measurement model.

Construct

Indicator Factor Loading ?Composite Reliability (CR)Average Variance Extracted (AVE)Hard reward HR1

0.950.960.84HR2

0.93HR3

0.88HR4

0.91Soft reward SR1

0.840.940.81SR2

0.92SR3

0.92SR4

0.90Altruism –personal satisfaction ALPS1

0.940.970.91ALPS2

0.96ALPS3

0.95Altruism –organizational bene ?t ALOB1

0.960.970.91ALOB2

0.94ALOB3

0.95Knowledge sharing behavior KSB1

0.900.950.83

KSB2

0.92KSB3

0.93KSB40.90?All factor loadings of the individual items are statistically signi ?cant (p b 0.01).

11

W.-T.Wang,Y.-P.Hou /Information and Organization 25(2015)1–26

12W.-T.Wang,Y.-P.Hou/Information and Organization25(2015)1–26

(93.0%).Most of the respondents worked for banks and insurance companies(85.4%).In terms of job type, they were mostly sales/marketing specialists(45.6%),customer service representatives(24.3%),and adminis-trative personnel(21.2%),whereas a minority of them were Information Technology staff(5.4%)and?nancial analysts(3.5%).

3.3.Data analysis methods

To examine our proposed research model at both individual and organizational levels,the PLS technique was used for the data analysis due to its disregard for the constraint of multivariate normality of data distri-bution and its ability to appropriately estimate the error variances of the latent constructs measured by only one or two items(Hair,Black,Babin,Anderson,&Tatham,2006;Petter,Straub,&Rai,2007;Wetzels, Odekerken-Schroder,&van Oppen,2009).SmartPLS2.0was adopted for measurement validation and for testing the structural model based on the data collected from the259survey respondents in34organizations. Con?rmatory factor analysis was performed to examine the validity and reliability of the constructs.In addition,a bootstrapping procedure was conducted for the signi?cant tests of the research hypotheses.

4.Data analysis and results of individual-level examination

4.1.Measurement model

The reliability of the measures for the constructs of interest was?rst tested by examining the individual Cronbach’s alpha coef?cients,which were all greater than the recommended level of0.7(ranging from0.92 to0.95).The psychometric properties of the measures for the?rst-order re?ective constructs were then further assessed in terms of convergent and discriminant validity.Three primary measures were used to evaluate the convergent validity(Hair et al.,2006):(a)the factor loadings of the indicators,which must be statistically signi?cant with values greater than0.6;(b)composite reliability(CR),with values greater than 0.7;and(c)average variance extracted(AVE)estimates,with values greater than0.5.As shown in Table1, all factor loadings(ranging from0.84to0.96)were statistically signi?cant,and all were larger than0.6.In addition,all CR values(ranging from0.94to0.97)were higher than0.7.The AVE values ranged from0.81 to0.91.Overall,all the measures exhibited adequate convergent validity.

Finally,the discriminant validity of the measures was determined.As shown in Table2,the squared correlations between the factors were smaller than the corresponding AVE estimates.This?nding indicates that the constructs were more strongly related to their respective indicators than to the other constructs in the model.Table3presents the descriptive statistics for each of the constructs in the proposed research model.

4.2.Structural model and hypothesis testing

By adopting the PLS technique using the bootstrapping procedure,the structural model was evaluated for hypothesis testing.The?t of the structural model was assessed by the explained variances(R2)for endoge-nous constructs and a global?t measure,namely,the goodness-of-?t index(GoF),speci?cally for PLS path modeling(Tenenhaus,Vinzi,Chatelin,&Lauro,2005;Wetzels et al.,2009).The proposed model explained a considerable proportion of the variance of endogenous latent constructs(ranging from0.52to0.54),as

Table2

Discriminant validity of the constructs of interest.

Construct 1. 2. 3. 4. 5.

1.Hard reward0.84

2.Soft reward0.430.81

3.Altruism–personal satisfaction0.210.540.91

4.Altruism–organizational bene?t0.410.460.540.91

5.Knowledge sharing behavior0.330.430.280.390.83 Note:Diagonals represent the average variance extracted,and the other matrix entries represent the squared factor correlations.

depicted in Fig.3.In addition,the geometric mean of the average communality (for exogenous constructs)and the average R 2(for endogenous constructs)were calculated as 0.83and 0.53,respectively.A GoF value (the square root of the product of the average communality and average R 2)of 0.66was then obtained,which is larger than the cutoff value of 0.36for large effect sizes of R 2(Wetzels et al.,2009).These results indicate support for the ?t of the structural model.

Given an adequate structural model,bootstrapping of the 259cases (the same as the original sample size)was conducted with 2000samples to evaluate the signi ?cance of the proposed research hypotheses.Fig.3presents the standardized path coef ?cients (β)and t values,the signi ?cance of the paths,and the R 2for each endogenous construct.Hypothesis H1a was supported,while hypotheses H1b and H1c were rejected.These ?ndings indicate that hard reward has a signi ?cant positive in ?uence on knowledge sharing behavior,but it does not have a signi ?cant negative in ?uence on altruism for organizational bene ?ts and altruism for personal satisfaction as we hypothesized.Additionally,hypotheses H2a,H2b,and H2c were all supported,indicating that soft reward has a signi ?cant positive in ?uence on altruism for organizational bene ?ts,altruism for personal satisfaction,and knowledge sharing behavior.Hard reward and soft reward accounted for 54%and 53%of the variance of altruism for organizational bene ?ts and altruism for personal satisfaction,respectively.Furthermore,hypothesis H3was con ?rmed,while hypothesis H4was rejected,indicating that altruism for organizational bene ?ts has a signi ?cant positive in ?uence on knowledge sharing behavior,but altruism for personal satisfaction does not.Finally,we found that none of the four control variables,namely age,gender,level of education,and department of the respondents,had a signi ?cant effect on knowledge sharing behavior.All the constructs that directly and/or indirectly in ?uence knowledge sharing behavior accounted for 54%of its variance.

4.3.Hierarchical moderated regression analysis

To compare the effects of the four types of motivations on employees ’knowledge sharing behaviors,we performed a hierarchical moderated regression analysis using a multiple regression technique (Bell,Auh,&Smalley,2005).The results indicate that soft reward is the most important determinant of employees ’knowledge sharing intentions.In addition to soft reward,hard reward is a key motivational factor in ?uencing employees ’knowledge sharing intentions (see Appendix 3).Finally,the results of the hierarchical moderated regression analysis indicated that none of the four control variables had a signi ?cant effect on the construct of knowledge sharing behavior.

5.Data analysis and results of organizational-level examination

5.1.Measurement model of organizational-level analysis

To further assess the relationships among various types of motivational factors and knowledge sharing behaviors in the workplace,we examined the proposed research model using an organizational-level analysis.To examine the proposed research model at the organizational level,we averaged all the scales of our survey items to form composites for these items based on the 34companies of our respondents.To assess the appropriateness of using this data aggregation approach to form the composites for the survey items based on the companies of our respondents,we ?rst examined the intraclass correlation coef ?cients for group mean ratings (ICC(K))for all the aggregated variables (LeBreton &Senter,2008).The results show that the

Table 3

Descriptive statistics of the investigated constructs.

Construct

Mean Standard Deviation Hard reward

4.73 1.23Soft reward

5.300.99Altruism –personal satisfaction

5.32 1.03Altruism –organizational bene ?t

5.02 1.07Knowledge sharing behavior 4.95 1.00

13W.-T.Wang,Y.-P.Hou /Information and Organization 25(2015)1–26

14

W.-T.

Wang,

Y.-P.

Hou

/

Information

and

Organization

25

(2015)

1

–

26

Fig.3.Hypotheses testing results based on the individual-level analysis.

ICC(K)indices of all the 18survey items were higher than 0.8(ranging from 0.84to 0.97).These ?ndings indicate that there is little variation between the scores given to each of the survey items by the raters (i.e.,survey respondents)within the same company,and thus provide support to the aggregation of individual-level data to form organizational-level data in this study.Additionally,the multi-item within-group reliability statistics (r WG(J))for all the aggregated variables of each of the 34participating companies were calculated.The results show that among the 34participating companies,23of them had a r WG(J)statistic that was larger than 0.7(ranging from 0.71to 0.89),indicating a high level of interrater agreement,eight of them had a r WG(J)statistic that was between 0.5and 0.7(ranging from 0.55to 0.69),indicating a moderate level of interrater agreement,and three of them had a r WG(J)statistic that was smaller than 0.5,indicating a low level of interrater agreement (ranging from 0.01to 0.10)(LeBreton &Senter,2008).Based on the results of the examination of ICC(K)indices and r WG(J)statistics,we excluded the data of the three companies with a r WG(J)statistic that was smaller than 0.5,and eventually obtained a sample of 31cases (representing the 31participating companies)for our organizational-level analysis.

Based on the sample for organizational-level analysis,the reliability of the organizational-level measures for the constructs of interest was ?rst tested by examining the individual Cronbach ’s alpha coef ?cients,which were all greater than the recommended level of 0.7(ranging from 0.94to 0.97).The psychometric properties of the measures for the constructs of interest were then further assessed in terms of convergent and discriminant validity.Here,we adopt the three primary measures previously used for the individual-level analysis to evaluate the convergent validity for measures of the organizational-level analysis.As shown in Table 4,all factor loadings (ranging from 0.90to 0.98)were statistically signi ?cant,and all were larger than 0.6.In addition,all CR values (ranging from 0.96to 0.98)were higher than 0.7.The AVE values were higher than 0.5(ranging from 0.89to 0.95).Overall,all the measures exhibited adequate convergent validity.

The discriminant validity of the measures was also determined.As shown in Table 5,the squared correla-tions between the factors were smaller than the corresponding AVE estimates.This ?nding indicates that the constructs were more strongly related to their respective indicators than to the other constructs in the model.

5.2.Structural model and hypothesis testing of organizational-level analysis

Then,by adopting the PLS technique using the bootstrapping procedure,the structural model was evaluated for hypothesis testing.The ?t of the structural model was assessed by the explained variances (R 2)for endogenous constructs and a global ?t measure,namely,the goodness-of-?t index (GoF),speci ?cally Table 4

Convergent validity for the measurement model of organizational-level analysis.

Construct

Indicator Factor Loading ?Composite Reliability (CR)Average Variance Extracted (AVE)Hard reward HR1

0.980.980.92HR2

0.98HR3

0.91HR4

0.97Soft reward SR1

0.940.970.90SR2

0.95SR3

0.97SR4

0.94Altruism –personal satisfaction ALPS1

0.940.960.89ALPS2

0.98ALPS3

0.90Altruism –organizational bene ?t ALOB1

0.970.980.95ALOB2

0.96ALOB3

0.98Knowledge sharing behavior KSB1

0.960.980.94

KSB2

0.96KSB3

0.97KSB40.98?All factor loadings of the individual items are statistically signi ?cant (p b 0.01).

15

W.-T.Wang,Y.-P.Hou /Information and Organization 25(2015)1–26

16W.-T.Wang,Y.-P.Hou/Information and Organization25(2015)1–26

Table5

Discriminant validity of the constructs of interest.

Construct 1. 2. 3. 4. 5.

1.Hard reward0.92

2.Soft reward0.720.90

3.Altruism–personal satisfaction0.560.780.89

4.Altruism–organizational bene?t0.790.660.630.95

5.Knowledge sharing behavior0.640.750.560.680.94 Note:Diagonals represent the average variance extracted,and the other matrix entries represent the squared factor correlations.

for PLS path modeling(Tenenhaus et al.,2005;Wetzels et al.,2009).The proposed model explained a consid-erable proportion of the variance of endogenous latent constructs(ranging from0.78to0.81),as depicted in Fig.4.In addition,the geometric mean of the average communality(for exogenous constructs)and the average R2(for endogenous constructs)were calculated,which were0.91and0.80,respectively.A GoF value(the square root of the product of the average communality and average R2)of0.85was then obtained, which is larger than the cutoff value of0.36for large effect sizes of R2(Wetzels et al.,2009).These results indicate support for the?t of the structural model.

Given an adequate structural model,bootstrapping of the31cases(representing the31participating companies)was conducted with5000samples to evaluate the signi?cance of the proposed research hypotheses.Fig.4presents the standardized path coef?cients(β)and t values,the signi?cance of the paths, and the R2for each endogenous construct.Overall,the hypothesis testing results of the analysis at the organi-zational level were mostly consistent with those of the analysis at the individual level presented in Fig.3,ex-cept for hypotheses H1a and H2b.To be speci?c,the positive effect of hard reward on knowledge sharing behavior and the positive effect of soft reward on altruism for organizational bene?ts were con?rmed at the individual level but were discon?rmed at the organizational level.

6.Discussion and implications

6.1.Discussion

The examination of the hypotheses developed in this study yields valuable insights.First,this examination ?nds a lack of a systematic and thorough framework for understanding how various types of motivational fac-tors may cause problems or biases in evaluating and understanding the effects of personal motivations on in-dividuals’knowledge sharing behaviors.By consulting SDT and the?ndings of previous studies,four key constructs of motivations,namely,hard rewards,soft rewards,altruism for personal satisfaction,and altruism for organizational bene?ts,are adopted in this study.The research results also con?rm the positive direct ef-fects of hard reward,soft rewards and altruism for organizational bene?ts on employees’knowledge sharing behaviors.Although the discon?rmation of the positive effect of altruism for personal satisfaction is inconsis-tent with some existing studies(Chang&Chuang,2011;Kankanhalli et al.,2005),others report?ndings sim-ilar to those in this work(Wasko&Faraj,2005).A plausible explanation for these?ndings is that because the employees’knowledge sharing behaviors investigated in this study are intended to support professional ac-tivities,achieving overall bene?ts for the organization may be a more important concern for employees than pursuing personal satisfaction by voluntarily helping others.

Second,we found that hard reward does not have a signi?cant negative effect on altruism for organization-al bene?ts and altruism for personal satisfaction,as we hypothesized based on SDT.Moreover,we found that hard reward has a signi?cant positive effect on altruism for organizational bene?ts.A possible reason for these ?ndings is that based on the?ndings of prior SDT studies,control-oriented motivations do not necessarily have a negative impact on autonomy-oriented motivations.To be speci?c,when these control-oriented mo-tivations are used by the superiors in organizations to offer important competence information and/or posi-tive feedback to their subordinates,they may strengthen rather than undermine the autonomous motivations of the subordinates by helping satisfy the subordinates’need for autonomy and competence (Gagne&Deci,2005;Sheldon et al.,2003).

Fig.4.Hypothesis testing results based on the organizational-level analysis.17

W.-T.Wang,Y.-P.Hou /Information and Organization 25(2015)1–26

18W.-T.Wang,Y.-P.Hou/Information and Organization25(2015)1–26

Finally,the con?rmation of the positive in?uences of soft reward on altruism for organizational bene?ts and altruism for personal satisfaction is consistent with the?ndings of prior studies(e.g.,Roca&Gagne, 2008;Sheldon et al.,2003;Yoo et al.,2012).These results indicate that control-oriented motivations are not invariantly controlled.While controlled motivations(e.g.,hard rewards)are external control mechanisms of human behaviors and thus negatively impact individuals’autonomy-oriented motivations,introjected regulations(e.g.,soft rewards)are likely to be internalized by individuals(i.e.,motivations have embedded within the individuals but has not been accepted by the individuals as their own),and are thus likely to promote autonomy-oriented motivations by facilitating individuals’perception of autonomy,being related to others,and self-competence(e.g.,Levesque,Stanek,Zuehlke,&Ryan,2004;Vansteenkiste,Lens,&Deci, 2006).

With regard to the hypothesis testing results based on the organizational-level analysis,we found that the results of all of the direct-effect hypotheses were consistent with those of our individual-level analysis,except for those that depict the positive relationships between hard reward and knowledge sharing behaviors and between soft reward and altruism for organizational bene?ts,which were supported based on the individual-level analysis but were rejected based on the organizational-level analysis.A possible explanation for these?ndings is that,compared with other business functions in organizations,the adoption of KM is still in its infancy.Consequently,it is likely that some organizations do not have a well-designed system for rewarding individuals’knowledge sharing efforts compared to other organizations.Thus,the differences in the rewarding systems of individual organizations may have somehow contributed to the insigni?cant relationship between hard reward and employees’knowledge sharing behaviors.The ability to leverage soft rewards and to facilitate employees’internal-ization of organizational values and norms may thus be more important than hard rewards in motivating employees’knowledge sharing behaviors.Additionally,in cultures that value collectivism,such as Asian cultures,people tend to resonate with group/organizational norms and thus,acting based on these collective norms may lead them to experience relatedness and autonomy and be more satis?ed with the associated collective outcomes(i.e.,altruism for organizational bene?ts),regardless of the tangible or intangible rewards they may receive for performing those actions(Deci&Ryan,2000).Another reason for the insigni?cant relationship between soft reward and altruism for organizational bene?ts at the organizational level is that a high level of interpersonal trust in an organization can enhance the effects of the shared organizational values and core beliefs of the employees on their intentions to share knowledge because it enables the employees to consider these shared organizational values and beliefs as an effective way to improve organizational outcomes(Peralta&Saldanha,2014).

6.2.Implications for research

The results of this work have a number of implications for research with reference to SDT.First,prior research indicates that it is dif?cult to fully understand the occurrence of individual knowledge sharing behaviors because these are affected by various personal motivational factors,including concerns related to self-interest(i.e.,costs versus bene?ts),altruistic orientations,and other human elements(Bock&Kim, 2002;Kankanhalli et al.,2005;Yang&Wu,2008).Although there have been studies that examine the in?uences of various types of motivations on individuals’knowledge sharing behaviors,they mostly focus on discussing various motivations in terms of level or amount rather than the quality of motivations.In other words,the concepts of autonomy-oriented motivations,namely the identi?ed regulation and autonomous motivation,tend to be either ignored or falsely integrated with the concept of intangible bene?ts/rewards(i.e.,control-oriented motivations)in these studies.From this perspective,the proposed research model is considered comprehensive,as it identi?es the critical control-oriented and autonomy-oriented motivations that have different levels of quality,in terms of their degrees of auton-omy,and it simultaneously examines their effects on individuals’knowledge sharing behaviors in the workplace based on the controlled-to-autonomous continuum of motivations proposed by SDT.It is thus believed that the current research?ndings have shed light on the use of SDT to investigate the key motivational factors,both control-oriented and autonomy-oriented,of individuals’knowledge sharing behaviors in various contexts.

Second,we found that there is controversy regarding whether some types of motivations that have different levels of quality may in?uence individuals’knowledge sharing behavior in the workplace more

than others.Consequently,we examine the difference in the effects of various types of motivations on individ-uals ’knowledge sharing behaviors.Our ?ndings indicate that of all of the motivational factors that in ?uence individuals ’knowledge sharing behaviors,introjected regulations/moderately controlled motivations (e.g.,personal reputation and self-esteem contingent on performance),namely the construct of soft reward in this study,and external regulations/controlled motivations (e.g.,rewards,punishments,and promotion),which are re ?ected by the construct of hard reward in this study,have relatively greater in ?uences.Although these ?ndings are somewhat inconsistent with the proposition of SDT,which indicates that autonomy-oriented motivations have relatively greater power to in ?uence individuals ’behaviors,they imply that in some cases,control-oriented motivations may play a key role in determining individuals ’behaviors in situations in which intense interpersonal competition is present (e.g.,in an organization that concentrates on observing personal contributions rather than group contributions)or in which they are internalized by the individuals (i.e.,the motivations have been transformed to be a part of the individuals ’essential code of conduct).Our research ?ndings may thus provide researchers with insights into the investigation of how var-ious types of motivations may affect individuals ’knowledge sharing behaviors differently in various contexts.

Finally,SDT argues that control-oriented motivations may have a negative in ?uence on autonomy-oriented motivations when they are used by the management of organizations as mechanisms for over-controlling em-ployees ’behaviors and performance,but may have a positive in ?uence on autonomy-oriented motivations when they are perceived as being helpful in terms of satisfying individuals ’basic psychological needs for auton-omy,relatedness,and competence (Sheldon et al.,2003).However,very few studies have speci ?cally investigat-ed the indicated relationships regarding employees ’knowledge sharing behaviors.Consequently,by empirically examining the in ?uences of control-oriented motivations (i.e.,hard reward and soft reward)on autonomy-oriented motivations (i.e.,altruism for organizational bene ?ts and altruism for personal satisfaction),we have advanced the current understanding of this issue addressed in SDT within the context of interpersonal knowl-edge sharing in workplaces.

6.3.Implications for practice

The research ?ndings indicate a number of practical implications.First,the ?ndings indicate that soft rewards and hard rewards have relatively greater in ?uences on employees ’knowledge sharing behaviors.Additionally,soft reward is relatively more important than hard reward in terms of facilitating employees ’knowledge sharing behaviors.These ?ndings indicate that managers should not rely on hard rewards as the primary incentives to facilitate these behaviors,as they may have only limited and temporary effects (Lin,2007).In contrast,managers are encouraged to explicitly offer soft reward mechanisms (e.g.,awards for the best knowledge contributors)to boost employees ’positive outcome expectations and thus increase knowledge sharing behaviors.

Second,it is found that altruism for organizational bene ?ts rather than for personal satisfaction has a signi ?cant positive effect on knowledge sharing behaviors.It is thus suggested that managers promote the concept of working for the greater good (e.g.,the collective bene ?ts of the employees in an organization and the achievement of organizational goals in the market).One way of doing so may be to explicitly include knowledge sharing as part of the corporate culture and to promote this inclusion at corporate meetings,conferences,and employee training sessions.

Finally,we found that the performance of employees ’knowledge sharing behaviors is associated with the level of the employees ’perceived autonomy-oriented motivation (i.e.,altruism for organizational bene ?ts)regarding these behaviors,which also mediates the in ?uence of soft rewards on these behaviors.These ?ndings imply the importance of creating an autonomy-supportive work https://www.360docs.net/doc/768373305.html,anizations ’sup-port of autonomy in the workplace can help ful ?ll the psychological needs of employees for having choices,rationale for corporate-valued norms or actions,and the organizations ’acknowledgement of personal feelings regarding sharing knowledge with others (Deci,Eghrarl,Patrick,&Leone,1994).These workplace-related characteristics may help encourage the employees to internalize the externally regulated motivations (e.g.,soft reward)to increase their sense of autonomy for carrying out knowledge sharing behaviors,thus leading to an increase in these behaviors.One way to create an autonomy-supportive work environment is for organizations to explicitly promote a corporate culture that is based on managers taking employees ’per-spectives when they make decisions,to provide greater choices for employees at work and to more actively 19

W.-T.Wang,Y.-P.Hou /Information and Organization 25(2015)1–26

20W.-T.Wang,Y.-P.Hou/Information and Organization25(2015)1–26

appreciate employee achievements that are derived from their self-initiated actions(Gagne&Deci,2005), such as autonomous knowledge sharing behaviors.

7.Conclusion

Based on SDT,a theoretical model that incorporates key motivational factors of employees’knowledge sharing behaviors is presented and empirically examined in this study.The research results provide managers with signi?cant insights into effective methods to encourage their employees to voluntarily share knowledge with one another.

This work has some limitations.One is related to the generalizability of our research?ndings.We recruited our survey respondents from?nancial service companies.The reason for focusing on this particular industry is that the products/services of the?nancial services industry tend to be abstract and intangible,and it is rela-tively more dif?cult for consumers to communicate their needs to the product/service providers.To ensure that there is a close match between the customers’primary needs and a company’s abilities to develop prod-ucts/services that can appropriately meet these needs,it is critical for the?rm to perform continued knowl-edge sharing practices.Consequently,our research results may not be generalizable to organizational contexts of other industries.Future research that uses samples collected from employees of organizations in other industries to examine the proposed research model or its variations and extensions is recommended.

In addition,with reference to the controlled-to-autonomous continuum proposed by the SDT,motivations that belong to the category of autonomous motivation/integrated regulation are missing from this study.To promote a particular behavior of individuals,it is critical to enable the individuals to perceive the regulations regarding that behavior to be truly autonomous because they consider the value of performing the behavior as ?tting within their personal goals and values.Consequently,the consideration of the individual effects of dif-ferent types of employees’autonomous motivations/integrated regulations on their knowledge sharing be-haviors is encouraged in subsequent research.Furthermore,many knowledge-sharing studies have demonstrated the merits of considering the distinction between knowledge contribution/donating behaviors and knowledge seeking/collecting behaviors(He&Wei,2009;Lin,2007).Therefore,further studies may con-sider the effects of various types of motivations that have different degrees of autonomy in individuals’knowl-edge seeking and knowledge sharing behaviors.

Finally,employees’knowledge sharing behaviors were investigated regardless of the characteristics of the knowledge shared,such as the classic distinction between implicit and explicit knowledge.Knowledge with different characteristics may result in different costs and dif?culties for individuals with regard to organizing and sharing knowledge and may thus require different means of facilitating knowledge sharing behaviors. Consequently,consideration of the nature of the knowledge shared is encouraged in subsequent research.Fi-nally,there have been knowledge-sharing studies that demonstrate the merits of considering the distinction between knowledge contribution/donating behaviors and knowledge seeking/collecting behaviors(He& Wei,2009).Therefore,further studies may be conducted that consider this distinction.

Acknowledgements

The authors thank the participating companies and the survey respondents for providing valuable data. The authors also thank the Editor and anonymous reviewers for their valuable feedback on this paper.This study was funded by the National Science Council,Taiwan[grant number:NSC99-2410-H-006-071-MY3].

The scales for measuring a speci?c construct in the research model were averaged to form a composite to represent the construct.The potential threat of multicollinearity was checked by examining whether the variance in?ation factors(VIFs)of the constructs were smaller than the recommended cutoff value of10 (Hair et al.,2006).As presented in Table A1,the VIFs of all regression models were smaller than10;thus, multicollinearity was not considered a problem.

Comparing the R2of Model1to those of Models2,3,4,and5indicates that the exclusion of soft reward from the regression model(see Model3)signi?cantly decreases the explained variance of knowledge sharing behavior.These results indicate that soft reward is the most important motivational factor of employees’knowledge sharing intentions.Additionally,comparing the R2of Model1to those of Models6,7,and8,all of which had soft reward excluded from the regression model,indicates that individually excluding hard reward from the regression model(see Model6)signi?cantly decreases the variance explained of knowledge