Print-out(Penting Rujukan Masterku)

ioo

Information & Management 50 (2013) 322335

u

p

e

n

(N

re

Contents lists available at SciVerse ScienceDirect

Information &

.1. Introduction

Organizations have increased their use of and reliance uponcomputers and networks. With the increased availability ofcomputers and the Internet at work, however, employees alsohave increased opportunities to use these same devices forpersonal reasons [6,12,30,35]. We refer to such behavior aspersonal use of the Internet. The research ndings on personalInternet use are twofold. On the one hand, scholars have notedthat when employees use the Internet and related applications(e.g., messenger and email applications), the employees sufferfrom a decrease in work output efciency [21,26,35], with adecrease in productivity of 20% [19] or 24% [20]. This reduction inefciency is costly for the organization in terms of reducedemployee output and the costs associated with the potential forincreased spyware, viruses, security leaks, and use of IT resources(e.g., the use of bandwidth for Skype, video, or Internet radio)[43,78]. Estimates have placed this loss in the billions of dollarsannually [6]. On the other hand, research shows positive effectsassociated with personal Internet use, such as stress relief [7,45].With the contradictions in the current empirical ndings, it isimportant to continue studying this topic to determine whichmotivations lead to personal use of the Internet. By understandingthe motivations behind the personal use of organizationalresources, companies will be able to adopt practices and

procedures and develop training methods to reduce the amountof inappropriate personal use of the Internet, if needed. We seek tofurther previous understanding of personal use of the Internet byproposing and testing a novel theoretical model, namely,Triandiss theory of interpersonal behavior (TIB). Triandis [67]not only included new antecedents of personal use of the Internetthat had not yet been studied but also compared multiplemotivations for personal use of the Internet concurrently ratherthan separately, as done in previous work on this topic[21,26,34,35,40,77].

The TIB includes the theory of reasoned action (TRA) and theoryof planned behavior (TPB) concepts (i.e., attitudes, social inuence,and intentions). New factors are also included in this context,namely, emotional factors, habits, and different sources of socialinuence. In so doing, the TIB provides a broader understanding ofwhat may lead to personal use of the Internet in the workplace. As aresult of such a broad theoretical framework, we shed new light onthe inuence of habit, affect, role, and self-concept on Internet use.We also show that widely advocated methods of control ordeterrence in the IS literature are unable to reduce personal use ofthe Internet.

The remainder of this paper is structured as follows. We reviewthe literature and ndings regarding personal use of the Internet inthe Section 2. We then examine the theory of interpersonalbehavior in the third section. Using this theoretical basis, weexplicate our model and its hypotheses in Section 3. We thendescribe our study and the analysis of the subsequent data. Theresults are then provided, along with implications for research andpractice.

* Corresponding author. Tel.: +1 702 895 1365.

E-mail address: [email protected] (G.D. Moody).

0378-7206/$ see front matter 2013 Elsevier B.V. All rights reserved.http://dx.doi.org/10.1016/j.im.2013.04.005Using the theory of interpersonal behavrelated personal use of the Internet at w

Gregory D. Moody a,*, Mikko Siponen b

aUniversity of Nevada, Las Vegas, United StatesbUniversity of Jyvaskyla, Finland

A R T I C L E I N F O

Article history:

Received 7 July 2012

Received in revised form 6 April 2013

Accepted 13 April 2013

Available online 26 April 2013

Keywords:

Personal use of the Internet

Theory of interpersonal behavior

Computer abuse

Computer misuse

A B S T R A C T

Non-work-related personal

from scholars. We increase

based on the theory of interp

attitudes, social inuence, a

social inuence. Our results

for non-work purposes. Our

in the use of the Internet.

jo u rn al h om ep ag e: ww wr to explain non-work-rk

se of the Internet within organizations has received increased attention

revious understanding of this phenomenon by proposing a novel model

rsonal behavior (TIB). The TIB includes previous researched constructs (i.e.,

d intentions) as well as emotional factors, habits, and different sources of

= 238) suggest that the model well predicts the use of the Internet at work

sults shed new light on the inuence of habit, affect, role, and self-concept

2013 Elsevier B.V. All rights reserved.

Management

els evier .c o m/lo c ate / im

aney belovedHighlight






2. Literature review

This section will briey describe two main literature streamsthat are relevant for this study. First, we will review the literaturerelated to personal use of the Internet. Second, we will explain thetheory of interpersonal behavior. For the purposes of this study,we dene personal use of the Internet as any use of a computer thatdoes not involve completing work-related tasks [34]. Such usemay include, for example, web surng, chatting, or onlineshopping. This denition is more expansive than those of previouswork in this area, as this denition includes the use otherapplications (e.g., messengers, video conferencing, Skype, onlinegaming, online communities) rather than focusing only on the useof an Internet browser for personal use while at work[26,35,58,61].

2.1. Use of the Internet at work for personal reasons

With the increasing presence of the Internet in the workplace,researchers initially focused on the extent to which personal use ofthe Internet was prevalent in the workplace. This type of researchtypically emphasizes that businesses should be aware of employees

personal use of the Internet and the detrimental effects it has on theorganization [35,61]. Such studies have found that at least 20% ofemployees in an organization engage in such an action and haveposited that it lowers employees performance and productivity[19,20].

This line of research has led to the provision of advice forreducing personal use of the Internet. Researchers and practi-tioners have advised organizations to adopt Internet use policies[60] that would help employees understand what constitutessuitable use of the Internet, thus deterring unsuitable use withinthe organization. The other common types of advice of this earlierwork on personal use of the Internet prescribe the use ofmonitoring tools, reports, and devices to ensure compliance withsaid policies [47,61,69]. However, empirical work has found thatthe use of monitoring lowers the overall job satisfaction of theemployees being monitored [69].

Building on the initial studies measuring the prevalence ofpersonal use of the Internet, later studies began proling personalInternet users in an attempt to provide tools and procedures formanagers to use in identifying users. It was hoped that throughproling, management would be enabled to proactively address,train, and punish individuals who continued to be personal users of

Table 1Summary of research on personal use of the Internet.

Author Year Theoretical framework Methodology Findings

Cheung et al. 2000 TIB Survey The Internet is more likely to be used at work when it is not viewed as

complex, the individual perceives near-term benets from his or her

behavior, personal use of the Internet is socially accepted at the

organization, and the individual has access to the Internet at work.

Chang and Cheung 2001 TIB Survey Using the Internet at work is more likely to occur when the individual

perceives near-term benets, he or she feels good about using the Internet,

the organization provides access to the Internet, and use of the Internet is

socially acceptable.

Anandarajan 2002 Articial

neural networks

approach

Simulation. survey for

testing the simulation

AI-based behavior models can be used to prole employees Web use

behavior on an a priori basis.

Simmers 2002 None Case studies Internet policies and monitoring of employee behavior will lead to optimal

levels of employee freedom while minimizing costs and risks for the

organization.

Oravec 2002 Social capital theory None Workplace recreation using the Internet can increase employee morale and

creativity and, therefore, productivity.

Lim et al. 2002 None Survey and

focus groups

Personal use of the Internet is common in the workplace. Individuals are

more prone to this type of abuse if they perceive that the organization

overworks them or does not provide adequate compensation.

Stanton 2002 None Survey Frequent cyberloafers have higher levels of job and pay satisfaction,

G.D. Moody, M. Siponen / Information & Management 50 (2013) 322335 323Urbaczewski and Jessup 2002 Theory X/Y Observation

and experiment

Galletta and Polak 2003 TPB Survey

Woon and Pee 2004 TIB Survey

Seymour and Nadasen 2007 TPB Survey

Pee et al. 2008 TIB Survey

Blanchard and Henle 2008 None Survey

Henle and Blanchard 2008 Coping theory Survey satisfaction with promotion opportunities, and higher ratings of

organizational support than non-abusers.

Monitoring decreases personal use of the Internet but also decreases the

users level of satisfaction.

Personal use of the Internet is most predicted by employees attitudes (job

satisfaction and Internet addiction) toward personal use of the Internet and

norms within the workplace environment.

Personal use of the Internet decreased when employees had intentions to

cyberloaf, a habit of personal use of the Internet, and conditions that

increased the likelihood of personal use of the Internet. Furthermore, job

satisfaction, affect toward personal use of the Internet, and social factors all

increased these mediating constructs, while perceived consequences

reduced intentions to engage in personal use of the Internet.

Only managerial supervision affected the level of personal use of the

Internet; other hypothesized TPB constructs were not signicant.

Habit, intention, and facilitating conditions all increased personal use of the

Internet; these factors were increased by affect, social factors, and perceived

consequences.

Minor and serious types of non-work-related use of the Internet exist. Minor

non-work-related use of the Internet is correlated with norms regarding this

behavior among peers or supervisors, but serious non-work-related use of

Internet is not.

Employees are more likely to engage in non-work-related use of Internet in

response to role ambiguity and role conict but less likely to do so as a result

of role overload. Furthermore, employees are more likely to participate in

non-work-related use of the Internet to cope with role ambiguity or role

conict when they perceive minimal sanctions.



G.D. Moody, M. Siponen / Information & Management 50 (2013) 322335324the Internet. Most approaches focused on varying dimensions ofsatisfaction or attitudes held toward the organization [35,63].These studies found that various attitudes toward the organizationand social norms do, in fact, alter the levels of personal use of theInternet in the organization. Simmers [61] also reported a neuralnetwork approach based on genetic algorithms that would captureusage statistics to predict users inclinations toward personal useof the Internet.

More recent work has attempted to use theoretical approachesto predict and understand why individuals engage in personal useof the Internet. The main approaches depended on specifyingantecedents of attitudes and social norms from the theory ofplanned behavior [26,58]. These studies found that employeesattitudes toward personal use of the Internet and norms at theworkplace have strong predictive power concerning employeesintentions and actual personal use of the Internet. Building onconcepts from interpersonal behavior, Pee and colleagues [50,77],along with Cheung and colleagues [14,16], showed that con-sequences, habits, facilitating conditions, and employees emo-tions also strongly predict personal use of the Internet at work. A

Fig. 1. Summary of the theory of interpersonal behavior.Adapted from Triandis, 1977summary of this work is shown in Table 1.The later theoretical work on the use of the Internet at work for

personal reasons shows conicting empirical results. Despitereplications from two theoretical bases (i.e., the TPB [26,58] andthe TIB [14,16,50,77]), the results are inconsistent. In the rsttheoretical study of workplace Internet abuse, Galletta and Polak[26] found that only certain attitudes (i.e., job satisfaction andInternet addiction) and subjective norms can predict the use of theInternet at work for personal reasons. However, using the sametheoretical approach, Seymour and Nadasen [58] found that noattitudes or subjective norms can predict abuse. Instead, theauthors reported that only the perceived supervision of managersis able to reduce abuse. Furthermore, using portions of the TIB, Peeand his colleagues [50,77] reported conicting ndings. In theinitial study [77], the authors showed that habits, intentions, andfacilitating conditions all negatively affect behavior; the later study[50], in contrast, demonstrated the opposite. Meanwhile, Changand Cheung [14] and Cheung and Limayem [15] reported that near-term consequences and facilitating conditions are conducive to the

1 Note that these studies are not focused on personal use of the Internet but on

whether the Internet will be adopted by the individual to aid him or her in his or her

job functions.intention to use the Internet at work,1 and they obtained mixedresults regarding the complexity of the application, social factors,and affect felt toward Internet use. With these mixed empiricalndings, it is difcult to understand what motivates individuals toengage in personal use of the Internet.

The current study seeks to expand on these former studies andto provide a more complete view of the antecedents of personal useof the Internet in several ways. First, we use the TIB, which hasbeen shown to account for more variance in a model comparedwith the TRA and the TPB [9]. Second, we expand the scope ofpersonal use of the Internet to include not only personal userelated to using Internet browsers but also all applications that usenetworking or telecommunications abilities at a computer. Third,we specically expand upon the work of Pee and his colleagues[50,77] and Cheung and his colleagues [14,16] by including all theconstructs from the TIB, with their respective antecedents asspecied by Triandis [67].

3. Theoretical framework: the theory of interpersonal behaviorWe believe the TIB is highly appropriate for the context of thisstudy, as personal use of the Internet at work is a highly socialbehavior learned within the organization through the observationof such behaviors on the part of other employees. Althoughultimately the employee is the one who engages in using theInternet for personal reasons, he or she perceives signals and cuesupon entering the organization that determine whether suchbehaviors will be negatively or positively perceived, and thereforerewarded or punished. Only after the individual has learned therelatively informal rules regarding such behaviors will he or she bewilling to engage in them. This study is in accordance with work incriminology that explains the social learning that must occurbefore deviance [5,13,62].

The theory of interpersonal behavior was rst specied byTriandis [67] as a theoretical alternative to the TRA and the TPB(see Fig. 1). Triandis [67] argued that the TRA and the TPB sufferfrom several weaknesses that are overcome by this model. The TRAand the TPB focus on predicting behaviors related to intentions toperform a given behavior. These intentions are predicted by theindividuals beliefs regarding the behavior and subjective normsthat are relevant to the behavior [3]. The TIB builds on this workand proposes several additions to the underlying model proposedin the TRA and the TPB. Each addition is discussed in turn, and they








are adopted or extended regarding the personal use of the Internetat work.

First, by focusing on the cognitive aspects of behavior, neitherthe TRA nor the TPB account for the emotions involved in theeventual behavior. Triandis argued that individuals often arrive atdecisions not only through focusing solely on the cognitive aspectsof a situation but also by relying on their feelings. Thus, Triandisproposed that affect serves as an input in the decision-makingprocess. For this paper, we dene affect as the emotional responseto a particular situation that is based on instinctive andunconscious processes in the mind [67]. Specically, we refer tothe emotional response an individual has when using the Internetat work, which is likely a source of emotional coping to addressissues that may arise in the workplace [30].

Second, the TRA and the TPB assume that intentions lead tobehavior, without any consideration of the previous occurrence ofthe same behavior [59]. The TRA and the TPB posit that intentionspredict behavior, but these theories do not consider whether thisbehavior has been so often repeated by the individual that it has

performed. Thus, the TIB proposes that facilitating conditions willserve as a moderator of the relationship between intentions andbehavior [25,67].

Fourth, building on concepts from neoclassical criminology[28], we extend the TIB and propose that attitudes are formedbased on the beliefs individuals hold as well as the evaluations ofthese beliefs. Beliefs refer to internally held information that oneholds to be true [24], whereas evaluation denotes the internalcalculation of the individual that determines how relevant thebelief is when forming an attitude in a given circumstance [24,67].Although an individual may hold several beliefs regarding anattitude object, each belief may be of a different level ofimportance in a given situation. Thus, only beliefs evaluated asrelevant will have signicant impacts on the formation of theattitude toward the given object. Thus, we propose that anattitude is formed by an interaction of relevant beliefs and theirrespective evaluations [67]. In this study, we focus on two specicbeliefs regarding the benets and costs of personal use of theInternet. Thus, we ascertain the beliefs and the evaluation of

reti

G.D. Moody, M. Siponen / Information & Management 50 (2013) 322335 325become automatic and, thus, performed without the consciousdeliberation assumed by the TRA and the TPB. This type ofautomated response to a situation to behave in a given manner isreferred to as habit [72]. For example, when someone arrives at astop sign, deciding to slow down and stop does not requiredeliberate and conscious reasoning; rather, this action is dictatedand automated due to the numerous times the individual hasperformed the same action. Thus, it is not merely intention thatdictates whether a behavior will occur but also how habitual thisbehavior has become. It is proposed that the relationship betweenintention and behavior is altered by the level of habit characteriz-ing the persons enactment of the behavior. Specically, thebehavior will be more pronounced when intention and habit arepresent than when either one is evident in isolation, resulting in anadditive interaction effect [25]. In adopting the theory of habit[71,72] and the TIB [67], we thus propose that the habitual use ofthe Internet at work for personal reasons will have its own maineffect on the behavior and alter the intention to engage in the samebehavior.

Third, similar to the theory of planned behavior [3], the TIBproposes that the decision to engage in a behavior will be affectedby the ability of the individual to perform the behavior. Facilitatingconditions refers to the lack of environmental or situationalconstraints that may prevent the individual from performing thedesired behavior. Even if a person commonly performs a behaviorand has an intention to engage in the behavior, if it is not possibleto do so due to extenuating circumstances, the behavior cannot be

Fig. 2. Theobenets and penalties of using the Internet at work for personalreasons.

Fifth, the TIB proposes a more detailed explanation of howthe individuals environment will inuence intentions andbehaviors. The TIB expands upon the role of social inuencesthrough roles, self-concepts, and social norms. Similar to theTRA and the TPB, social norms are included in this model andrefer to the pressures and expectations of others that cause anindividual to behave in a given manner [67]. Similar to the TRAand the TPB, the TIB proposes that social norms increase theinclination of individuals to act in ways that will increaseconformity with the known social group. However, unlike theTRA and the TPB, the TIB also proposes that social inuencescome from sources beyond the norms of the group in which thebehavior is performed. Roles refers to the sets of actions deemedappropriate for individuals occupying a given position withinthe group, whereas self-concept refers to the idea thatindividuals have their own internal goals and values regardingwhich behaviors are appropriate [9,67]. For the present context,we have specically adopted roles, norms, and concepts thataddress employees in an organizational setting regarding theirpersonal use of the Internet. Whereas norms apply equally to allindividuals in a group, the insertion of roles into the modelallows for the variability groups experience due to the uniquepositions and functions of individuals within the group. Forexample, although a group may have a norm for individuals tobe silent unless called upon, it is considered appropriate for the

cal model.




G.D. Moody, M. Siponen / Information & Management 50 (2013) 322335326group leader to lead discussion and to speak during the majorityof the meeting without being called upon. Additionally, theinsertion of self-concept into the model accounts for individualdifferences due to the values of the individual, which may bemore important than desires for inclusion within a group. Forexample, consider a group that normally celebrates successes byattending a local bar for happy hour. One member, however,although a long-standing and reputable member of the group,does not attend these celebrations due to religious convictions.Personal convictions and central values may override socialpressures to conform to desired group behaviors when thesebehaviors challenge an individuals central and salient values[10].

By incorporating these additional constructs into predictingintentions, our extension of the TIB to the personal use of theInternet at work provides a more comprehensive theoreticalmodel for predicting behaviors [67] than previous studies basedon the TRA or the TPB. Due to the conicting ndings in theprevious literature regarding the antecedents or motivations foremployees to engage in personal use of the Internet [26,50,77],our extension of the TIB is an ideal theory for examining theantecedents of personal use of the Internet. The TIB examinescognitive, affective, social, and habitual factors that mayinuence personal use of the Internet rather than only a subsetof this list.

4. Model development

Having reviewed the relevant literature and provided thebackground of the theory of interpersonal behavior, we nowexplain our theoretical model (see Fig. 2). We rst explain the twoantecedents of attitude, followed by the antecedents of socialfactors. Next, we explain how attitudes, affect, and social factorsinuence intentions. Last, we propose how habit, intentions, andfacilitating conditions impact actual behavior related to personaluse of the Internet.

4.1. Attitude

The neoclassical view in criminology is that people selectengagement in behaviors that go against the establishedprocedures, rules, and guidelines (i.e., personal use of the Internetin the context of this study) when it pays off, i.e., people engage inthese behaviors when the benets are high and the risk ofsanctions is low [44,57]. Thus, this study adopts benets andpenalties as positive and negative antecedents of attitude,respectively. We propose that the expectation of positiveoutcomes due to personal use of the Internet should result inmore favorable attitudes toward such behavior. First, beliefsregarding future potential outcomes serve as a motivation toengage in a given behavior [11]. This motivation is based on theattitude that the benets are relevant and possible and will helpthe individual achieve their desired goals. If an individual believesbenets are possible and likely, his or her positive attitude towardthe behavior should increase in an effort to achieve the benets[18].

The connection between perceived beliefs regarding potentialbenets from a behavior and the attitude toward engaging in thebehavior has long been proposed and supported in prior literaturein different applications of neoclassical theory of crime andmotivation research [4,18,22,65,66]. We extend these ndings topersonal use of the Internet and propose the following:

H1. The perceived benets regarding personal use of the Internetare positively related to the attitude toward the use of the Internetat work for personal reasons.As individuals attempt to maximize benets and minimizepenalties [66], the perception of penalties associated with personaluse of the Internet should also alter an individuals attitude towardfuture use of the Internet at work for personal reasons. Prior workhas already indicated that individuals tend to be more averse toloss and punishment than attracted by rewards [31]. Althoughbenets should increase an individuals attitudes toward engagingin the behavior, perceived penalties associated with this behaviorwill have a stronger, negative impact on the individuals attitudetoward the same behavior. In line with the neoclassical theory ofcrime [57], we propose that individuals will avoid negativeoutcomes and punishments by forming negative attitudes towardengaging in a punishable behavior. Therefore, we propose thefollowing:

H2. The perceived penalties regarding personal use of the Internetwill be negatively related to attitudes toward the use of theInternet at work for personal reasons.

4.2. Social factors

Triandis [67] dened social factors as an individuals assessmentof the reference groups culture and the specic interpersonalagreements the individual has made with others in specic socialsituations [77]. The TIB is more socially oriented than either theTRA or the TPB, and it proposes several sources of social inuencebeyond those of social norms [9]. Although other sources of socialinuence could be proposed and tested, this paper tests thoseproposed by Triandis [67].

First, Triandis [67] suggested that the rst major source of socialinuence is the presence of social norms within referent groups.Social norms inuence individuals by increasing the desire andpressure from a reference group to conform to the expectedbehavior. Individuals within the group, or being observed by agroup, follow these unwritten rules to conform to the pressures ofthe group and thus act in accordance with the norms of therelevant referent group [3]. Although individuals may break normsand behavior contrary to norms, the presence of norms serves as acue or pressure that increases the likelihood that the individualwill behave in accordance with the given norm [38].

Norms are known to inuence behavioral intentions ofindividuals in groups, which is also a central tenet of the TRAand the TPB [3,24,33,38]. However, social norms are only onesource of inuence that can occur within the social context of agiven situation. Thus, in accordance with the TIB, we propose thatthe social factors should be considered jointly rather thanseparately. Thus, although social norms have been shown todirectly impact behavioral intentions from research based on theTRA and the TPB, in accordance with the TIB, we propose that thejoint effects of all social factors will mediate the relationshipsbetween each source of social inuence and abuse intentions.Social norms produce an effect upon the inuence the individualfeels from social sources for a specic situation, which are thenused to create behavioral intentions. Thus, we propose thefollowing:

H3. Social norms regarding non-work-related Internet use withinthe organization will be positively related to the social factorsregarding personal use of the Internet.

The term roles refers to the idea of what is normal and properbehavior, as determined by the position of the individual withinthe relevant social group [67]. Like social norms, this idea can beconsidered only within a social situation in which the individual isdeciding how to behave. Roles are socially construed andunderstood [1,2,9,73]. When an individual is deciding how to





G.D. Moody, M. Siponen / Information & Management 50 (2013) 322335 327behave, the consideration of the individuals roles can be understoodonly by looking at social roles of relevance within the group. Forexample, an individuals role as sister may not be relevant indeciding whether to use the Internet at work for personal reasons,unless her sister happens to be her boss at the company.

Because individuals have numerous roles and functions, it ispossible to understand the impact of roles on eventual behavioronly by considering the relevant group involved in the given role[8,68]. Individuals will consider the various roles deemed relevantfor the given context and determine what type of inuence thisrole has on that behavior [9,67]. Thus, an individuals roles withinthe organization should signicantly affect the overall socialinuence an individual feels toward personal use of the Internet.Therefore, we propose the following:

H4. Organizational roles within the organization will be positivelyrelated to the overall social factors regarding use of the Internet atwork for personal reasons.

Finally, Triandis [67] proposed that an individuals self-conceptregarding the behavior should also affect the amount of socialinuence perceived by the individual. Self-concept is proposed toimpact social factors due to the ability of signicant and knownothers to observe behaviors. For example, if an individual stronglybelieves it is important to reduce his or her own carbon footprinton the world, he or she will experience strong pressure fromknown others to engage in behavior that supports this known self-concept. He or she will have an increased likelihood of riding a biketo work or using a hybrid car to publicly conform to his or her owninternal value system. Individuals often assess themselves basedon the opinions and feedback they receive from others [48,56].Thus, in deciding how to behave in a given situation, the potentialsocial consequences and the impact of relevant social others on theindividual and his or her self-concept will alter how an individualdecides to behave. Thus, we propose the following:

H5. Self-concepts regarding personal use of the Internet will bepositively related to the overall social factors regarding use of theInternet at work for personal reasons.

4.3. Antecedents of intention

Previous work has long proposed and found that attitudes,emotions, and social factors inuence behavioral intentions[9,24,33,50,53,67]. Because the relationship between attitudeand emotions on intentions is well-known and specied in priorliterature, we will briey explain the relationship between socialfactors and intentions, as this relationship is mentioned only in theTIB. As previously stated, the TIB expands upon the socialinuences that may alter an individuals behavioral decisions byincluding other social factors beyond social norms. In the previoussection, we described these sources of social factors and explainedhow they would affect behavior by altering the level of socialinuence felt by an individual in a given situation. Thus, theconnection between social factors and behavioral intentions isbased on the same reasoning given for each factor, or thosecommon to social norms in the TRA and the TPB. Essentially,individuals are inuenced by pressures the individuals perceivefrom relevant social groups where the behavior would beperformed [33,59]. These sources of social inuence increase thelikelihood that an individual will desire to conform to known groupnorms, internal value structures, or roles to which the individual isexpected to adhere.

Because these relationships have all been found in priorliterature, we merely extend these previous ndings to t thecontext of our study, and we propose the following:H6. Attitudes regarding use of the Internet at work for personalreasons will be positively related to the intention to engage inpersonal use of the Internet.

H7. Affect regarding use of the Internet at work for personalreasons will be positively related to the intention to engage inpersonal use of the Internet.

H8. Social factors regarding use of the Internet at work for per-sonal reasons will be positively related to the intention to engagein personal use of the Internet.

4.4. Predicting personal Internet use behavior

Previous research has long proposed and found that theintention to engage in a behavior and a habit of performing abehavior are strong predictors of behavior [2,9,25,33,39,59,70,73,72]. However, the theory of interpersonal behavior builds uponthis research by proposing an interaction of these constructs oneventual behavior [10,9,25,67]. Triandis [67] originally explainedthat the intention to behave in a given fashion would be stronglyinuenced by the previous frequency of the behavior. For example,if an individual is not in the habit of checking his or her email atwork, it is unlikely that the individual will check his or her email,despite a strong intention to do so at a given point in time.Likewise, if the individual has a strong habit of checking his or heremail every hour, it is very likely that he or she will continue tocheck email at work, despite an intention not to do so. However,the joint effects of habit and intention will be magnied when theyare in the same direction. If the individual has an intention to checkhis or her email, and he or she usually does this, it is even morelikely that he or she will do so than if he or she had neither the habitnor the intention. In other words, the TIB proposes that futurebehavior is a function not only of what the individual intends to dobut also of what the individual typically does.

Although this interaction was proposed by Triandis [67], laterempirical tests of the theory within this context have not consideredthe interaction of these constructs on behavior [14,16,50,77]. Thus,we extend the proposed interaction of intention and habit onbehavior from the TIB, as found in prior research [9], to the context ofpersonal use of the Internet as follows:

H9a. The intention to use the Internet at work for personal reasonswill be positively related to actual personal use of the Internet.

H9b. The habit of using the Internet at work for personal reasonswill be positively related to actual personal use of the Internet.

H9c. The interactive effect of intentions and habit will stronglypredict the use of the Internet at work for personal reasons.

In accordance with neoclassical theories of crime [57], such asrational choice models, facilitating conditions should moderate therelationship between intentions and habit on behavior [28,32,65]. Ifan individual has an intention to engage in use of the Internet atwork for personal reasons but does not have the means oropportunity to easily perform the behavior, it is not likely thatpersonal use of the Internet will take place [76]. By controlling andmonitoring workstations and network trafc, organizations candecrease the facility with which individuals may engage in personaluse of the Internet and avoid punishment. Thus, the presence ofcontrols and monitoring functions within the network would likelydeter abuse behavior by individuals. This has been proposed in priorwork [22,32,66], and we likewise replicate the following prediction:

H10a. Having ready access to the Internet at work (facilitatingconditions) is positively related to actual personal use of theInternet.

(i.e., benets or penalties) as described by the literature on the TIB[25,67].

5.3. Data analysis

5.3.1. Establishing factorial validity

Before the hypotheses were assessed, several steps were takento ensure the reliability and accuracy of the collected data. First, weascertained the types of constructs used in this study. UsingDiamantopoulos and Winklhofer [23] and the sources of the

G.D. Moody, M. Siponen / Information & Management 50 (2013) 322335328H10b. Having ready access to the Internet at work (facilitatingconditions) negatively affects the relationship between intentionsand actual personal use of the Internet.

5. Methodology and data analysis

5.1. Method and data collection

This study used a survey methodology. Data were collectedduring a two-week period from a Finnish private service companythat offers electrical and related services. Following Westlands[74] guidelines, we determined that at least 100 responses shouldbe collected to test the model. The company employs 1,150employees, of whom 238 submitted completed surveys, for ausable response rate of 21%. The survey was anonymous; noidentifying information of any type was gathered from theparticipants. It was also clearly communicated to the respondentsthat independent university researchers would analyze the results.The organization monitors Internet use at the organizational andgroup levels but not at the individual level, which is illegal inFinland due to European Union (EU) privacy laws. The companyhas a non-work-related Internet use policy.

The reliability of constructs can be improved by usingpreviously validated and tested questions [63]. Accordingly, weused items taken from previously validated and reported instru-ments (with some minor wording adjustment to t the context ofthis study). Appendix A provides a detailed list of the scales usedfor this study. Participants were asked to report their personal useof the Internet at work for non-work purposes. The participantswere then asked to provide answers for the remaining constructsin the theory. These constructs were the following: attitude [51],beliefs about and evaluations of the outcome of their behaviors[50], norms [25], roles [9], self-concept [25], social factors [50],affect [50], habit [71], facilitating conditions [50], intentions [50],and behavior [50].

Because we used the TIB in a new context, we used a pilot test toensure the readability and validity of the questions. The pilotpopulation consisted of staff members at a public university inFinland. The pilot respondents included IT support staff, lecturers,secretaries, administrative staff, and educational planners. Weobtained 43 usable responses. Our pilot study involved a paper-based questionnaire that consisted of 65 questions, includingspace in which respondents could leave remarks and feedbackabout the questions. We used these responses to ascertain thevalidity of the questions and to identify any points of confusionwithin the survey. Based on the feedback and initial statisticalanalysis, several questions were modied slightly before the naldata collection.

5.2. Construction of benets and penalties

This section will briey emphasize how the benets andpenalties were created from subjects respective beliefs andmatching evaluations. The TIB proposes that weighted beliefsform initial attitudes that serve as antecedents for intentions. Foreach relevant belief in this context, participants were asked tojudge the likelihood of the benet or penalty occurring and themagnitude of the impact on the participant. The two scores (i.e.,belief and evaluation) for each item were then multiplied to forman evaluated belief score for each respective item. For example,suppose a participant gave a score of 6 (fairly likely) that she wouldreceive a warning for using the Internet at work for non-work-related purposes and that she rated the severity of this warning as a2 (lenient). Her evaluated belief score would then be 12 (6 2).These formed scores were loaded onto their respective constructinstruments, we determined whether the constructs were forma-tive or reective. The remainder of this section reports ourprocedures for establishing factorial validity tests for reective andformative constructs using the respective tests.

5.3.1.1. Reective constructs. To analyze the factorial validity of theconstructs, we applied partial least squares (PLS) using SmartPLSversion 2.0 [55]. To establish the validity of our reectiveindicators, we followed the procedures outlined in [27]. Toestablish convergent validity, we generated a bootstrap with200 resamples and examined the t-values of the outer modelloadings. All retained items were signicant at the .05 a level (seeTable A2.1 in Appendix B). This demonstrates strong convergentvalidity for the reective constructs.

We then used two conventional methods for establishingdiscriminant validity: correlating the latent variable scoresagainst the indicators, also known as a conrmatory factoranalysis (see Table A2.2), and calculating the square root of theaverage variance extracted for each reective construct (see TableA2.3). Both demonstrated strong convergent validity for allretained items.

Finally, to establish reliability, PLS computes a compositereliability score as part of the model analysis (see Table 1). Thisscore is a more accurate assessment of reliability than Cronbachsalpha because it does not assume that the loadings or error terms ofthe items are equal [17]. Each reective construct in our researchmodel demonstrates high composite reliability that exceedsstandard thresholds (Table 2).

Convergent and discriminant validities were also assessedusing STATAs (version STATA/SE 12.1) conrmatory factor analysis(CFA) (see Table A2.5). The model t was good, with values thatwere are all within acceptable parameters: x21266 = 4372.511,p < .001; x2/df = 3.45; CFI = 0.947; TLI = 0.931; RMSEA = 0.062;SRMR = 0.039; CD = 1.000, [29,75]. Convergent validity wassupported by large and standardized loadings for all constructs(p < .001) and t-values that exceeded statistical signicance.Convergent validity was also supported by calculating the ratioof factor loadings to their respective standard errors, which exceedj10.0j (p < .001).

Discriminant validity was tested by showing that the measure-ment model had a signicantly better model t to a competingmodel with a single latent construct and to all other competingmodels in which pairs of latent constructs were joined. The x2

differences between the competing models (omitted for the sake ofbrevity) were signicantly larger than those of the original model,

Table 2Composite reliability.

Construct Composite reliability

Affect 0.962

Attitude 0.910

Behavior 0.952

Habit 0.959

Intention 0.970

Roles 0.911

Self-concept 0.895


as also suggested by factor loadings, modication indices, andresiduals [42]. In sum, these tests conrm convergent anddiscriminant validities.

5.3.1.2. Formative constructs. Before discussing the validation ofour formative measures, we explain the operationalization ofbenets and penalties. The TIB proposes that attitudes are createdbased on the joint beliefs and evaluations of outcomes in the givencircumstance, such as use of the Internet at work for personalreasons. For each item regarding benets and penalties, thesubjects were asked about their beliefs regarding that item and foran evaluation of that item. We then interacted the belief andevaluation scores to form an evaluated belief item score for eachrespective item. These scores then served as formative indicatorsfor the respective benets and penalties constructs.

Validating formative indicators is more challenging thanvalidating reective indicators because the established proceduresthat exist to determine the validity of reective measures do notapply to formative measures [52], and the procedures validatingformative measures are less known and established [23]. Research-ers have generally used theoretical reasoning to support the validity

Finally, we used another approach to assess formative validity,as suggested by Petter et al. [52], which involved testing themulticollinearity among the indicators. This step is particularlyimportant with formative indicators because multicollinearityposes a much greater problem than with reective indicators.Hence, low levels of multicollinearity are usually indicated withlevels of the variance ination factor (VIF) below 10, but in the caseof formative indicators, the VIF levels need to be below 3.3 as amore stringent test [52]. In our case, the VIFs for ve indicators (anitem from benets, from penalties, two from norms, and one fromsocial factors) were above 3.3, and these were all subsequentlydropped from the nal analysis.

In sum, using MTMM analysis and assessing the VIF levels, weconcluded that there was reasonable discriminant validity withour formative constructs. Finally, because of the nature offormative measures, reliability checks could not reasonably becarried out [23].

5.4. Testing for common methods bias

Because the data were collected using one method, we used two

*

**

**

uare

G.D. Moody, M. Siponen / Information & Management 50 (2013) 322335 329of formative constructs [23], although approaches can be usedbeyond theoretical reasoning alone [41,52]. Although no techniquehas been widely established for validating formative measures, weused the modied multitrait-multimethod (MTMM) approach, aspresented in [36,37,41], which is a promising solution.

For each formative item, we created new values that were theproduct of the original item values by their respective PLS weights(representing each items weighted score). We then created acomposite score for each construct by summing all the weightedscores for a construct. Next, we produced correlations of thesevalues, providing inter-measure and item-to-construct correla-tions (see Table A2.4).

To test the convergent validity, we checked whether all theitems within a construct correlated highly with each other andwhether the items within a construct correlated with theirconstruct value; this was mostly true in all cases, implyingconvergent validity. Although we would ideally want inter-itemcorrelations to be higher within a given construct, this cannot bestrictly enforced, as there are exceptions depending on thetheoretical nature of the formative measure [23,36]. Thus, webelieve that the most meaningful discriminant validity check withformative measures involves looking at the degree to which itemswithin a construct correlate to the given construct.

Benefits

Penalties

Attitude(.393)

Norms

Roles

Self Concept

Social Factors(.466)

Affect

0.614***

-0.126*

0.414**

0.356*

0.320*0.301***

0.229*

0.568***

Fig. 3. Model results (the R-sqmethods to test for the presence of common methods bias. First,we used Harmans single factor test [54]. This required that we runan exploratory unrotated factor analysis on all the rst-orderconstructs. The aim of this test is to determine whether a singlefactor emerges that explains the majority of the variance in themodel. If a factor emerges, then common methods bias likely existsat a signicant level. The result of our factor analysis for our studyproduced 35 distinct factors, the largest of which accounted foronly 15.8% of the variance of the model.

Second, we examined a correlation matrix of our latentconstructs to determine whether any of the correlations wereabove .90, which is strong evidence that common methods biasexists [49]. None of the correlations were near this threshold.Because our data passed both tests for common methods bias, weconcluded that there was little reason to believe that our dataexhibit any of the negative effects from common methods bias.

6. Results of hypotheses testing

Because our data displayed factorial validity and did not displaycommon methods bias, we continued to test our model. We alsovalidated these results through CB-SEM analysis using STATA/SE12.1. As an additional verication of our theory, we ran the model

Intention(.401)

Habit

Facilitating Conditions

Behavior(.867)

0.601***

0.281***

0.726***

-0.023-0.017

d values are in parentheses).

consistent with the TIB [67]. Related studies on personal use of theInternet have reported mixed ndings regarding social factors.Galletta and Polak [26] reported that subjective norms, operatio-

Table 3Summary of hypotheses testing.

# Hypothesis Coef. Supported?

1 Benets ! attitude .614 *** Yes2 Penalties ! attitude .126 * Yes3 Norms ! social factors .301 *** Yes4 Roles ! social factors .229 * Yes5 Self-concept ! social factors .568 *** Yes6 Attitude ! intention .414 *** Yes7 Affect ! intention .356 *** Yes8 Social factors ! intention .320 *** Yes9a Intention ! behavior .281 *** Yes9b Habit ! behavior .726 *** Yes9c Intention habit ! behavior .601 *** Yes10a Facilitating conditions ! behavior .017 ns No10b Facilitating conditions

intention ! behavior.023 ns No

ns not signicant.

Fig. 4. Interactive effects of habit on the relationship between intentions andbehaviors.

G.D. Moody, M. Siponen / Information & Management 50 (2013) 322335330using a PLS-based and CB-SEM-based analysis. The model resultsfrom both tools were equivalent, and we report the results from theCB-SEM analysis for brevity.

The common t indices show that the model t is acceptable,with values that are all within acceptable parameters:x21266 = 4372.511, p < .001; x

2/df = 3.45; CFI = 0.947;TLI = 0.931; RMSEA = 0.062; SRMR = 0.039; CD = 1.000, [29,75].Because numerous indices provide moderate to adequate results,we report the results of our CB-SEM-based model in Fig. 3. Theresults of our hypotheses, as based on the model testing, are shownin Table 3.

Because the hypothesized relationship of intention-predictingbehavior is interacted by habit, we now graphically depict thenature of this interaction (see Fig. 4). To depict this interaction, weextrapolated from our model results, and holding all othervariables constant, predicted the behavioral score given a standarddeviation change in both interacting variables as depicted above.

6.1. Results of hypothesis tests

We found that the majority of our hypotheses were supported.First, we report that the benets (H1) and penalties (H2) associatedwith using the Internet at work for personal reasons weresignicant predictors of attitudes toward engaging in thisbehavior.

Second, we found that all antecedents of social factors weresignicant in predicting social factors. Specically, we determinedthat the most signicant predictor of social factors was ones self-concept associated with personal use of the Internet (H5;b = 0.568). Norms (H3) and roles (H4) only moderately predictedsocial factors (bnorms = 0.301; broles = 0.229).

Third, we found that the main structure of the TIB wassupported. That is, attitude (H6), affect (H7), and social factors(H8) of personal Internet use all signicantly affected onesintention to use the Internet for personal purposes at work. Wealso found that intention to engage in this behavior (H9a) and thehabits based on this behavior (H9b) were signicant in predictingactual personal Internet use at work. Moreover, we found that theproposed interaction of intentions and habits was highlysignicant (H9c; b = 0.726). Further exploration of the interactionrevealed that intentions have more pronounced effects onbehavior while the effect of habit was small, with high intentionsleading to using the Internet at work for personal reasonscorrelating to such behaviors, whereas lower intentions were lesslikely to lead to those behaviors at work. However, we found thatbecause the habit of using the Internet at work for personalreasons was average compared with the sample, low and highintentions resulted in increased behaviors toward this type of use.Finally, we observed that as habits increased, compared with therest of the sample, the difference between low or high intentionsdecreased, which resulted in smaller differences betweenbehaviors at that range.

However, we found no support for either hypothesis involvingthe facilitating conditions around the personal use of the Internetat work. The hypothesized main effect of this construct on actualbehavior was nonsignicant (H10a; b = 0.017), and the hypoth-esized interaction between facilitating conditions and the inten-tion to engage in personal Internet use was insignicantly relatedto the behavior (H10b; b = 0.023).

7. Discussion

7.1. Summary

Here, we briey emphasize a number of ndings based on ourempirical study. First, our results indicate that all antecedents ofthe intention to engage in personal use of the Internet are ofrelatively equal strength. Individuals who have emotions, atti-tudes, and social inuences that positively regard the use of theInternet are more likely to use it at work for non-work purposes.This is consistent with the TIB [67] as well as previous studies onpersonal use of the Internet by Pee et al. [50] and Chang andCheung [14].

Second, our results show that ones attitude toward personaluse of the Internet is differentially inuenced by the benetsperceived in personal use of the Internet as opposed to theperceived penalties for engaging in this behavior. The importanceof benets, penalties, and attitude support the TIB [9,50,67].However, previous research by Seymour and Nadasen [58]demonstrated that attitudinal variables do not promote personaluse of the Internet. We explain these differences through thedifferent operationalization of the attitude construct. For Seymourand Nadasen [58], attitude includes low job satisfaction, inade-quate rewards, and long working hours. We used a measure forattitude, and attitude was predicted by the perceived benets andpenalties associated with personal use of the Internet. Becauseattitude was measured directly, we believe this construct is moreaccurately operationalized as specied by the TIB.

Third, we found that social factors also represent a signicantantecedent of the intention to engage in personal Internet usage atwork. Further, we found that all three predicted antecedents ofsocial factors are signicant in predicting this construct, which is

*** p < .001.* p < .05.




G.D. Moody, M. Siponen / Information & Management 50 (2013) 322335 331nalized in terms of peer culture and supervisor culture, increasepersonal use of the Internet. However, Seymour and Nadasen [58]found that supportive peer and supervisor cultures do not lead toan increased intention to engage in personal use of the Internet. Weexpanded on this previous work by exploring not only the normsassociated with personal use of the Internet but also an individualsrole within the organization and the individuals concept regardingpersonal use of the Internet. Furthermore, we reported the rst testto show that ones self-concept regarding ones personal use of theInternet at work is the strongest predictor of social factorsregarding this intention.

Fourth, our results show that, as predicted by the TIB, TRA, andTPB, the intention to engage in personal use of the Internet stronglypredicts actual behavior related to personal use of the Internet.Previous work on personal use of the Internet also found supportfor this relationship [50,77]. However, unlike those predicted bythe TIB, the facilitating conditions that would more easily enablean individual to engage in personal use of the Internet show nosignicant effects on actual personal use of the Internet behaviorsor an interaction with the intention to use the Internet at work forpersonal reasons.

Finally, our results indicate that an individuals habitualpersonal use of the Internet is a very important and strongindicator to consider when predicting actual behaviors relatedto personal use of the Internet. An individuals habit of engagingin personal use of the Internet in the past is the strongestantecedent of current, actual personal use of the Internet, whichis in alignment with previous research on habits [73,72].Additionally, we found that the interactive effect of intentionsand habit strongly predicted actual personal use of the Internet.Our nding is consistent with the TIB [67]. We found no studieslooking at the interaction effect between intention and habit inrelation to personal use of the Internet. Previous relatedresearch that built on the TIB either did not investigate thisinteraction [50,77] or ignored the effect of habit. Cheung andcolleagues [14,16], for example, focused on the adoption of theInternet in the function of the employees role within theorganization.

7.2. Contributions

Our study makes several important contributions to theliterature on personal use of the Internet. First, our model indicatesthat the largest predictor of this behavior is based on theinteraction of an individuals habits regarding personal use ofthe Internet and the individuals intention to use the Internet atwork for personal reasons. Specically, individuals who have thisintention and have a habit of such behavior are even more prone toengaging in personal use of the Internet than when they exhibiteither indicator on its own. This multiplicative effect is animportant contribution to this research stream for several reasons.

First, because this is the rst study to include the effect of habitson personal use of the Internet at work, we have also shown whythe inclusion of this variable is an important consideration forfuture research. This interaction is important in that managementand researchers are unable to directly alter the habits individualsmay have concerning personal use of the Internet, but it is possibleto alter the organizational environment and thus reduce theintention to engage in personal use of the Internet. This interactionincreases the importance of habit and intention constructs whenattempting to predict or control behaviors related to personal useof the Internet. Thus, future research should continue to emphasizefactors that can be used to attenuate the intention to engage in thisbehavior, especially with the differing perspectives affordedthrough the TIB that are based on emotional or socially relatedconstructs.Second, this study also emphasizes the advantage of adoptingeither the TIB or the theory of habit when studying the personaluse of the Internet at work. The predominant theories of the TRAand the TPB do not account for habit, which was shown in thisstudy to have a signicant impact on this behavior. This ndingalone indicates the importance of adopting the alternativeviewpoint afforded by the TIB. Future research on the personaluse of the Internet at work should account for, or at least controlfor, the habits regarding this behavior, which would then allow fora more accurate measurement of other factors being studied.Habits are strong predictors of behaviors and should beconsidered in future work on this phenomenon. Given the strongeffect of habit on personal use of the Internet, it would stand as aprimary candidate for interventions to reduce personal use of theInternet. However, habit research [73,72] has found thatdeprogramming habits is very difcult and that interventionsthat attempt to reduce habit strength are likely to fail. Due to thedifculty involved in such an endeavor, it is more important toprevent habits from forming regarding personal use of theInternet.

Third, our model indicates that certain interventions mayhave limited ability to reduce personal use of the Internet in anorganization. First, we see that the effect of penalties is minorcompared with the relevant benets, indicating that deterrentmethods of controlling personal use of the Internet have verymodest impacts on eventual personal use of the Internetcompared with the benets of doing so. This nding is interestingbecause it is contrary to the general approach to reducingundesired behaviors as proposed by control and deterrencetheories [28,46]. Second, an individuals affect toward personaluse of the Internet is largely outside the ability of organizationalinterventions to alter. Affect is an internal construct formed fromprior experience and is based on the individuals emotionalmake-up [64]. However, organizational interventions are gener-ally unable to alter these types of emotions. Rather, emotion-based training must be implemented that could alter affect overtime.

Furthermore, most extant research in IS usually proposes orsupports the notion that negative factors outweigh positive factors,as proposed in prospect theory [31]. However, we found that,contrary to many results in IS research, this behavior occurs in theopposite direction. We propose that given the main outcome ofpersonal use of the Internet, it is counterproductive rather thanuseful, from the managerial perspective, for the methods ofcontrolling this behavior to also be reversed. Future research intoother similar negative behaviors should thus consider that thesame general ndings may hold true and that carrots rather thansticks would produce more benecial or motivational outcomes.One reason for the non-signicant inuence of penalties on non-work-related Internet use is that individual-level monitoring ofweb trafc is illegal in Finland due to EU privacy laws. As a result,the risk of getting caught is low.

Finally, our model reports the highest reported R squaredregarding personal use of the Internet to date in an IS journal. Forexample, Blanchard and Henles [12] most predictive modelreported only an adjusted score of .21, while Manrique de Lara[40] reported .142 and Galletta and Polak [26] .192. Extant researchhas had limited predictive power regarding the intention to use theInternet at work for personal reasons, much less the actualbehavior. Much of the previous work on personal use of theInternet has stopped at the intention level, and when actualbehavior has been documented, lower R squared values werereported than in this study. By building on previous research andincluding many of the antecedents reported in these researchersstudies, we report the most powerful model for predicting personaluse of the Internet to date.






G.D. Moody, M. Siponen / Information & Management 50 (2013) 3223353327.3. Implications for practice

Here, we discuss ve strategic implications for practice basedon our ndings. First, given that our results show that penalties andcontrols have limited effect on personal use of the Internet, wesuggest that organizations need to establish other means foraddressing non-work-related Internet use other than penalties andcontrol-based means. These means are related to habitualbehavior, norms, roles, and benets and are discussed next.

Second, our nding on the role of habitual behavior behindInternet use implies that if employees Internet use becomeshabitual, the behavior will be difcult to unlearn. This suggeststhat if organizations want to intervene in their employees Internetuse behavior, it is easier to do so if the non-work-related Internetuse has not become habitual. Thus, for organizations preferring totake a stance on Internet use, early intervention is advised.

Third, given the signicant role of norms behind Internet use,for organizations preferring to intervene in their employeesInternet use it is important that top management, superiors, andthe IT department take a visible stand on Internet use. Here, thecommunication channel may not matter; instead, the key is tomake employees believe that this is the expectation for their workrole as determined by management and IT. It is important that sucha stand should not be mere lip service; rather, it should be visible inthe daily actions of the management and IT department. Ourndings regarding norms suggest that management and IT shouldmake a normative statement and create normative expectationsfor appropriate use of the Internet at work. Again, the key is not themedium through which this is communicated; the key is that thenormative stand be believable as normative expectations set bymanagement and IT.

Fourth, our ndings on the importance of roles imply thatorganizations that want to limit Internet use by their employeesneed to explain why the non-work-related use of the Internet is notnecessary in terms of carrying out work tasks.

Fifth, benets represent another challenge for organizationsaiming at reducing employees non-work-related use of Internet.Because our results suggest that employees want to save personaltime and expense through non-work-related Internet use and todevelop a more interesting work life through the non-work-relateduse of the Internet, organizations need to ensure that theiremployees work motivation is supported by appropriate leader-ship approaches. This also has implications for recruitment: it isimportant to recruit people who are motivated by their work.

7.4. Implications for future research

Based on our ndings, we emphasize ve topics for futureresearch. Future research needs to examine how a habit developsover time. Previous studies on habitual IT use have been based onvariance models. These models are important in that they help usmeasure whether IT users have a habit or not and whether thehabit predicts behavior in the given context. However, thesemodels do not provide any information on how to change thehabit or how the habit develops, in other words, how employeesacquire the habit of non-work-related Internet use and thenecessary steps employees need to take if they want to eliminatethe habit. We observe that examining habit from such aperspective requires understanding of the dynamic nature ofhabits, perhaps something that can be provided by applying aprocess theory viewpoint. To close this gap in the research, wesuggest that process theories be used to unveil the steps of habitdevelopment and the conditions of transformation from onestage to another. Then, we suggest that interventions beimplemented that aim to move people through these stages,for example, by unlearning the habit.Second, certain issues for future research stem from ourndings on roles. Future researchers need to study why employeesthink it is appropriate to engage in non-work-related Internet usewhen working. We suggest the use of qualitative researchmethods, especially interview techniques, to uncover thesereasons. In the same vein, given the signicant role of self-concept,the third issue that future research could examine is whyemployees feel that non-work-related use of Internet does notcontradict the individuals moral principles.

For the fourth future research issue, our nding regardingbenets related to employees non-work-related use of theInternet calls for more research. Future studies need to examinewhat aspects of non-work-related Internet use make employeeswork life more interesting and productive, according to theirperception. A similar examination is required for attitude andaffect. In particular, future researchers need to conductqualitative interviews to generate a deeper understanding ofwhy Internet use is a good idea (attitude) and why it is pleasant(affect) for employees. Here, we call for a grounded theorytype of research, which also reveals the dynamic aspects ofattitude and affect, including how they develop and change overtime.

The fth issue for future research is the role of national culturein non-work-related Internet use. One aspect of cultural research isthe role of sanctions. In Finland, where our data are from,individual-level monitoring and respective sanctioning are re-stricted due to privacy laws. Thus, future research could study theinuence of monitoring and sanctions in other countries, especiallythose with less-strict privacy laws, allowing monitoring ofindividuals web trafc at work.

8. Conclusions

Using the Internet at work for personal reasons is becomingincreasingly common. On the one hand, scholars argue thatpersonal use of the Internet entails a number of problems,including decreased efciency of work output, increased risk ofgetting viruses and spyware, and waste of IT resources. On theother hand, other researchers have shown a positive effectassociated with personal Internet use.

Studies have focused on understanding the type of individuallikely to engage in this behavior and what leads to these acts[6,14,16,26,35,50,58,63]. However, these studies have reportedconicting results. In the present work, we built on previousresearch by including these antecedents in one model in anattempt to compare their effects on personal use of the Internet. Iforganizations do not understand why employees engage inpersonal use of the Internet, then the organizations cannot modifytheir practices to reduce the likelihood of this behavior and itssubsequent cost to the organization.

This study used a theoretical approach to explore the variousmotivations that may lead to personal use of the Internet. Wefound that organizations need to consider several factors if theyare attempting to reduce this behavior. The key factors includethe perceived benets involved with personal use of theInternet, the emotions attached to engaging in this act, andthe habit that these employees have developed by which theycontinue to behave in this fashion. More practically, organiza-tions should ensure through recruitment that they have highlymotivated and committed employees in each work role. Inaddition, organizations that want to inuence their employeesInternet behavior should establish educational sessions andcampaigns that stress the negative implications of personal useof the Internet for the organization. Interestingly, our resultsshow that penalties and controls have limited effects onpersonal use of the Internet.




Future research should use interviews to obtain a deeper

G.D. Moody, M. Siponen / Information & Management 50 (2013) 322335 333understanding of why employees feel that they can save time andincrease work productivity through personal use of the Internet. Inaddition, the interviews should examine why employees believethat personal use of the Internet ts their work roles.

Appendix A. Instruments

All items are measured on a standard 7-point Likert scale(strongly disagree to strongly agree), unless otherwise indicated.

A.1. Attitude [51]

1. Using the Internet at work for non-work reasons is a bad idea2. Using the Internet at work for non-work reasons is a good idea3. Using the Internet at work for non-work reasons is a ___ idea

(foolish idea to wise idea)*2

A.2. Beliefs about outcomes [50]

Using the Internet at work for non-work-related purposes willresult in . . . (7-point scale: very unlikely 50% chance very likely)

A.3. Penalties FORMATIVE

1. Warnings2. Reprimands3. My Internet access privileges being restricted by the organiza-

tion

A.4. Benets FORMATIVE

1. Saving personal time using private Internet access2. Saving personal expense for using private Internet access3. Convenience4. More interesting work life5. Increase in my work productivity

A.5. Evaluation of outcomes [50]

Evaluate each of the items in the list below as a penalty for usingthe Internet at work for non-work purposes: (7-point scale: verylenient just right very harsh)

A.6. Penalties FORMATIVE

1. Warnings2. Reprimands3. My Internet access privileges being restricted by the organiza-

tion

A.7. Benets FORMATIVE

1. Saving personal time using private Internet access2. Saving personal expense for using private Internet access3. Convenience4. More interesting work life5. Increase in my work productivity

2 An * indicates that the item was reverse coded.A.8. Social factors [50] FORMATIVE

Evaluate each item in the list below as pertaining to his/her/their approval of you using the Internet at work for non-work-related purposes: (7-point scale: very low moderate veryhigh)

1. My familys2. My friends (outside of work)3. My co-workers4. My immediate supervisors5. My IT departments6. My top managements

A.9. Norms [25] FORMATIVE

1. My family would expect that I use the Internet at work for non-work purposes

2. My friends outside of work would expect that I use the Internetat work for non-work purposes

3. My clients would expect that I use the Internet at work for non-work purposes

4. My co-workers would expect that I use the Internet at work fornon-work purposes

5. The IT department at work would expect that I use the Internetat work for non-work purposes

6. Top-level management would expect me to use the Internet atwork for non-work purposes

A.10. Roles [9]

1. For me, as a employee of X, it is (appropriate/not appropriate) touse the Internet at work for non-work purposes

2. Using the Internet at work for non-work purposes is (tting/nottting) for my position as an employee of X

3. Due to my role at work, it is (justied/not justied) to use theInternet for non-work-related purposes

A.11. Self concept [25]

1. I would feel bad if I was not using the Internet at work for non-work purposes

2. Using the Internet at work for non-work purposes wouldconform to my principles

3. It would be unacceptable to not use the Internet at work for non-work purposes

A.12. Affect [50]

I feel that using the Internet provided by the organization fornon-work related purposes is . . .

1. Pleasantunpleasant2. Boringinteresting*3. Gratifyingdispleasing

A.13. Habit [72]

In regard to using the Internet at work for non-work relatedreasons, answer the following questions.


G.D. Moody, M. Siponen / Information & Management 50 (2013) 3223353341. I do it frequently.2. I do it automatically.3. I do it without having to consciously remember.4. It makes me feel weird if I do not do it.5. I do it without thinking.6. It would require effort not to do it.7. It belongs to my (daily, weekly, monthly) routine.8. I start doing it before I realize Im doing it.9. I would nd it hard not to do.

10. I have no need to think about doing it.11. Its typically for me.12. I have been doing it for a long time.

A.14. Facilitating conditions [50] FORMATIVE

In my organization (7-point scale: never sometimes veryoften)

1. My ability to use the Internet at work is high2. My access to the Internet at work is high3. The Internet connection at work is fast

A.15. Intention [50]

1. I intend to use the Internet at work for non-work-relatedpurposes in the future (strongly disagree neutral stronglyagree)

2. I will use the Internet at work for non-work-related purposes inthe future (7-point scale: very unlikely 50% chance verylikely)

3. I expect to use the Internet at work for non-work-related purposesin the future (strongly disagree neutral strongly agree)

A.16. Behavior [50]

1. In general, I use the Internet at work for non-work-relatedpurposes

2. I access the Internet at work for non-work-related purposesseveral times each day (7-point scale: very unlikely 50%chance very likely)

3. I do not spend a signicant amount of time on the Internet atwork for non-work-related purposes*

Appendix B. Supplementary data

Supplementary data associated with this article can be found,in the online version, at http://dx.doi.org/10.1016/j.im.2013.04.005.

References

[1] H. Aarts, T. Paulussen, H. Schaalma, Physical exercise habit: on the conceptualiza-tion and formation of habitual health behaviours, Health Education Research 12,1997, pp. 363374.

[2] H. Aarts, B. Verplanken, A. van Knippenberg, Predicting behavior from actions inthe past: repeated decision making or a matter of habit? Journal of AppliedPsychology 28, 1998, pp. 13551374.

[3] I. Ajzen, From intentions to actions: a theory of planned behavior, Action-Control:From Cognitions to Behavior 11, 1985, pp. 1139.

[4] I. Ajzen, M. Fishbein, Attitude-behavior relations: a theoretical analysis and reviewof empirical research, Psychological Bulletin 84, 1977, pp. 888918.

[5] R.L. Akers, M.D. Krohn, L. Lanza-Kaduce, M. Radosevich, Social learning and deviantbehavior: a specic test of a general theory, American Sociological Review 44, 1979,pp. 636655.

[6] M. Anandarajan, Proling web usage in the workplace: a behavior-based articialintelligence approach, Journal of Management Information Systems 19, 2002, pp.243266.[7] M. Anandarajan, N. Paravastu, C.A. Simmers, Perceptions of personal web usage inthe workplace: a Q-methodology approach, CyberPsychology and Behavior 9, 2006,pp. 325335.

[8] B.E. Ashforth, F. Mael, Social identity theory and the organization, Academy ofManagement Review 14, 1989, pp. 2039.

[9] S. Bamberg, P. Schmidt, Incentives, morality, or habit? predicting students car usefor University routes with the models of Ajzen, Schwartz, and Triandis Environ-ment and Behavior 35, 2003, pp. 264285.

[10] S. Bamberg, I. Ajzen, P. Schmidt, Choice of travel mode in the theory of plannedbehavior: the roles of past behavior, habit, and reasoned action, Basic and AppliedSocial Psychology 25, 2003, pp. 175187.

[11] A. Bandura, Self-efcacy: toward a unifying theory of behavioral change, Psycho-logical Review 84, 1977, pp. 191215.

[12] A.L. Blanchard, C.A. Henle, Correlates of different forms of cyberloang: the role ofnorms and external locus of control, Computers in Human Behavior 24, 2008, pp.10671084.

[13] R.L. Burgess, R.L. Akers, A differential association-reinforcement theory of crimi-nal behavior, Social Problems 14, 1966, pp. 128147.

[14] M.K. Chang, W. Cheung, Determinants of the Intention to use Internet/www at work: a conrmatory study, Information and Management 39,2001, pp. 114.

[15] M.K.C. Cheung, M. Limayem, The role of habit in information systems continu-ance: examining the evolving relationship between intention and usage, in:Proceedings of ICIS, 2005.

[16] W. Cheung, M.K. Chang, V.S. Lai, Prediction of Internet and world wide web usageat work: a test of an extended triandis model, Decision Support Systems 30, 2000,pp. 83100.

[17] W.W. Chin, B.L. Marcolin, P.R. Newsted, A Partial least squares latent variablemodeling approach for measuring interaction effects: results from a Monte Carlosimulation study and an electronic-mail emotion/adoption study, InformationSystems Research 14, 2003, pp. 189217.

[18] C.N. Cofer, M.H. Appley, Motivation: Theory and Research, John Wiley, Oxford,UK, 1967.

[19] Cyberloang, Intrinsic Health. http://myintrinsichealth.com/2012/03/12/cyber-loang/, 2012 (last accessed 9.12.12.).

[20] Cyberloang, Society for Industrial & Organizational Psychology, Inc. http://www.siop.org/Media/News/loang.aspx, 2012 (last accessed 9.12.12.).

[21] J. DArcy, A. Hovav, Deterring internal information systems misuse, Communica-tions of the ACM 50, 2007, pp. 113117.

[22] J. DArcy, A. Hovav, D.F. Galletta, User awareness of security countermeasures andits impact on information systems misuse: a deterrence approach, InformationSystems Research 23, 2009, pp. 7998.

[23] A. Diamantopoulos, H.M. Winklhofer, Index construction with formative indica-tors: an alternative to scale development, Journal of Marketing Research 38, 2001,pp. 269277.

[24] M. Fishbein, I. Ajzen, Belief, Attitude, Intention and Behavior: An Introduction toTheory and Research, Addison-Wesley, Reading, MA, 1975.

[25] M.-P. Gagnon, G. Godin, C. Gane, J.-P. Fortin, L. Lamothe, D. Reinharz, A. Cloutier,An adaptation of the theory of interpersonal behavior to the study of telemedicineadoption by physicians, International Journal of Medical Informatics 71, 2003, pp.103115.

[26] D.F. Galletta, P. Polak, An empirical investigation of antecedents of Internet abusein the workplace, SIG Workshop on HCI, 20034751.

[27] D. Gefen, D.W. Straub, A practical guide to factorial validity using PLS-graph: tutorial and annotated example, Communications of the AIS 16,2005, pp. 91109.

[28] J.P. Gibbs, Crime, Punishment, and Deterrence, Elsevier, New York, NY, 1975.[29] J.F. Hair, W.C. Black, B.J. Babin, R.E. Anderson, Multivariate Data Analysis,

Prentice-Hall, 2006.[30] C.A. Henle, A.L. Blanchard, The interaction of work stressors and organiza-

tional sanctions on cyberloang, Journal of Managerial Issues 20, 2008, pp.383400.

[31] D. Kahneman, A. Tversky, Prospect theory: an analysis of decision under risk,Econometrica 47, 1979, pp. 263291.

[32] S.M. Lee, S.-G. Lee, S. Yoo, An integrative model of computer abuse based on socialcontrol and general deterrence theories, Information and Management 41, 2004,pp. 707718.

[33] L. Leone, M. Perugini, A.P. Ercolani, A comparison of three models of attitude-behavior relationships in the studying behavior domain, European Journal ofSocial Psychology 29, 1999, pp. 161189.

[34] V.K.G. Lim, The IT way of loang on the job: cyberloang, neutralizing andorganizational justice, Journal of Organizational Behavior 23, 2002, pp. 675694.

[35] V.K.G. Lim, T.S.H. Teo, G.L. Loo, How do I loaf here? Let me count the waysCommunications of the ACM 45, 2002, pp. 6670.

[36] K. Loch, D.W. Straub, S. Kamel, Diffusing the Internet in the Arab world: the role ofsocial norms and technological culturation, IEEE Transactions on EngineeringManagement 50, 2003, pp. 4563.

[37] P.B. Lowry, N.C.J. Romano, J.L. Jenkins, R.W. Guthrie, The CMC interactivity model:how interactivity enhances communication quality and process satisfaction inlean-media groups, Journal of Management Information Systems 26, 2009, pp.155195.

[38] T.J. Madden, P.S. Ellen, I. Ajzen, A comparison of the theory of planned behaviorand the theory of reasoned action, Personality and Social Psychology Bulletin 18,1992, pp. 39.

[39] B.F. Malle, How people explain behavior: a new theoretical framework, Person-ality and Social Psychology Review 3, 1999, pp. 2348.

[40] P.Z. Manrique de Lara, Relationship between organizational justice and cyberloaf-ing in the workplace: has anomia a say in the matter? CyberPsychology andBehavior 10, 2007, pp. 464470.

[41] G.M. Marakas, R.D. Johnson, P.F. Clay, The evolving nature of the computer self-efcacy construct: an empirical investigation of measurement construction,validity, reliability, and stability over time, Journal of the Association for Infor-mation Systems 8, 2007, pp. 1646.

[42] H.W. Marsh, D. Hocevar, Application of conrmatory factor analysis to the studyof self-concept: rst-and higher order factors models and their invariance acrossgroups, Psychological Bulletin 97, 1985, pp. 562582.

[43] C. Mesa, The high cost of cyberslacking, Newsweek 134, 1999, pp. 6265.[44] L.L. Miller, The Perils of Federalism: Race, Poverty, and the Politics of Crime

Control, Oxford University Press, Oxford, UK, 2008.[45] J.A. Oravec, Constructive approaches to Internet recreation in the workplace,

Communications of the ACM 45, 2002, pp. 6063.[46] W.G. Ouchi, M.A. Maguire, Organizational control: two functions, Administrative

Science Quarterly 20, 1975, pp. 559569.[47] R.R. Panko, H.G. Beh, Monitoring for pornography and sexual harassment, Com-

munications of the ACM 45, 2002, pp. 8487.[48] A.M. Passos, A. Caetano, Exploring the effects of intragroup conict and past

[68] J.C. Turner, Social categorization and the self-concept: a social cognitive theory ofgroup, Advances in Group Processes 2, 1985, pp. 77121.

[69] A. Urbaczewski, L.M. Jessup, Does electronic monitoring of employee Internetusage work? Communications of the ACM 45, 2002, pp. 8083.

[70] P. Valois, R. Desharnais, G. Godin, A comparison of the Fishbein and Ajzen andTriandis attiudinal model for the prediction of exercise intention and behavior,Journal of Behavioral Medicine 11, 1988, pp. 459472.

[71] B. Verplanken, H. Aarts, Habit, attitude, and planned behaviour: is habit an emptyconstruct or an interesting case of goal-directed automaticity? European Reviewof Social Psychology 10, 1999, pp. 101134.

[72] B. Verplanken, S. Orbell, Reections on past behavior: a self-report index of habitstrength, Journal of Applied Social Psychology 33, 2003, pp. 13131330.

[73] B. Verplanken, H. Aarts, A. van Knippenberg, Habit, information acquisition, andthe process of making travel mode choices, European Journal of Social Psychology27, 1997, pp. 539560.

[74] J.C. Westland, Lower bounds on sample size in structural equation modeling,Electronic Commerce Research and Applications. 9, 2010, pp. 476487.

[75] L.J. Williams, R.J. Vandenberg, J. Edwards, Structural equation modeling in man-agement research: a guide for improved analysis, The Academy of ManagementAnnals 3, 2009, pp. 543605.

[76] R. Willison, M. Siponen, Overcoming the insider: reducing employee computer

G.D. Moody, M. Siponen / Information & Management 50 (2013) 322335 335performance feedback on team effectiveness, Journal of Managerial Psychology30, 2005, pp. 231244.

[49] P.A. Pavlou, H. Liang, Y. Xue, Understanding and mitigating uncertainty in onlineexchange relationships: a principal-agent perspective, MIS Quarterly 31, 2007,pp. 105136.

[50] L.G. Pee, I.M.Y. Woon, A. Kankanhalli, Explaining non-work-related computing inthe workplace: a comparison of alternative models, Information and Manage-ment 45, 2008, pp. 120130.

[51] R. Pennington, H.D. Wilcox, V. Grover, The role of system trust in business-to-consumer transactions, Journal of Management Information Systems 20, 2004,pp. 197226.

[52] S. Petter, D.W. Straub, A. Rai, Specifying formative constructs in informationsystems research, MIS Quarterly 31, 2007, pp. 623656.

[53] R.E. Petty, D.T. Wegener, Attitude change: multiple roles for persuasion variables,The Handbook of Social Psychology 1, 1998, pp. 323390.

[54] P.M. Podsakoff, S.B. MacKenzie, J.-Y. Lee, N.P. Podsakoff, Common method biasesin behavioral research: a critical review of the literature and recommendedremedies, Journal of Applied Psychology 88, 2003, pp. 879903.

[55] C.M. Ringle, S. Wende, S. Will, SmartPLS 2. 0 (M3) Beta, 2005.[56] S.L. Robinson, E. Weldon, Feedback seeking in groups: a theoretical perspective,

British Journal of Social Psychology 32, 1993, pp. 7186.[57] B. Roshier, Controlling Crime, Open University Press, Chicago, IL, 1989.[58] L. Seymour, K. Nadasen, Web access for IT staff: a developing world perspective on

web abuse, The Electronic Library 25, 2007, pp. 543557.[59] B.H. Sheppard, J. Hartwick, P.R. Warshaw, The theory of reasoned action: a meta-

analysis of past research with recommendations for modications and futureresearch, Journal of Consumer Research 15, 1988, pp. 325343.

[60] K. Siau, F. Fui-Hoon, L. Teng, Acceptable Internet use policy, Communications ofthe ACM 45, 2002, pp. 7579.

[61] C.A. Simmers, Aligning Internet usage with business priorities, Communicationsof the ACM 45, 2002, pp. 7174.

[62] W.F. Skinner, A.M. Fream, A social learning theory analy

Print-out(Penting Rujukan Masterku)

Documents

Transcript of Print-out(Penting Rujukan Masterku)