dynamic nature of trust in virtual teams

27

Click here to load reader

Upload: prasert-kanawattanachai

Post on 05-Jul-2016

221 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Dynamic nature of trust in virtual teams

Dynamic nature of trust in virtual teams

Prasert Kanawattanachaia,*, Youngjin Yoob,1

aFaculty of Commerce and Accountancy, Chulalongkorn University, Phayathai Road, Phatumwan,

Bangkok 10330, ThailandbWeatherhead School of Management, Case Western Reserve University, 10900 Euclid Avenue,

Cleveland, OH 44109, USA

Abstract

We empirically examine the dynamic nature of trust and the differences between high- and low-

performing virtual teams in the changing patterns in cognition- and affect-based trust over time

(early, middle, and late stages of project). Using data from 36, four-person MBA student teams from

six universities competing in a web-based business simulation game over an 8-week period, we

found that both high- and low-performing teams started with similar levels of trust in both cognitive

and affective dimensions. However, high-performing teams were better at developing and

maintaining the trust level throughout the project life. Moreover, virtual teams relied more on a

cognitive than an affective element of trust. These findings provide a preliminary step toward

understanding the dynamic nature and relative importance of cognition- and affect-based trust over

time.

q 2002 Elsevier Science B.V. All rights reserved.

Keywords: Virtual teams; Business simulation; Mutual trust

1. Introduction

Today’s organizations are experimenting with various forms in order to organize and

leverage their human assets. In response to this trend, the emergence of virtual teams

denotes an important new means of organizing workforces (DeSanctis and Monge,

1999). Members in virtual teams are generally geographically dispersed; they normally

interact electronically and work across time and space (DeSanctis and Monge, 1999;

Jarvenpaa and Leidner, 1999; Lipnack and Stamps, 2000). These teams provide many

advantages over traditional teams, including the ability to bridge time and space, and

offer better utilization of distributed human resources without physical relocation of

0963-8687/02/$ - see front matter q 2002 Elsevier Science B.V. All rights reserved.

PII: S0 96 3 -8 68 7 (0 2) 00 0 19 -7

Journal of Strategic Information Systems 11 (2002) 187–213

www.elsevier.com/locate/jsis

1 Tel.: þ1-216-368-0790.

* Corresponding author.

E-mail addresses: [email protected] (P. Kanawattanachai), [email protected] (Y. Yoo).

Page 2: Dynamic nature of trust in virtual teams

employees (Lipnack and Stamps, 2000). However, there are inherent obstacles of

virtual teams that come with flexibility. Given the separation in time and space, the

possible absence of a shared work history, and the limited options of communication

channels, working in virtual team settings can possibly be disastrous. This concern is

echoed in the prediction from the Gartner group that by 2004, more than 60% of the

professional workforce in the Global 2000 Company will work in virtual teams. The

Gartner group also predicted “by 2003, 50% of virtual teams will fail to meet either

strategic or operational objectives due to the inability to manage distributed workforce”

(Biggs, 2000). If this prediction about the change in nature of the work place and the

failure rate of virtual teams were true, what potential factors would lead to a better

organization of virtual teams?

We are just beginning to understand what fundamental factors drive the success and

failure of virtual teams. One of the fundamental factors that are believed to be important in

determining the success and failure of virtual teams is trust. The literature on trust in face-

to-face teams suggests that the establishment of trust is of importance in the working

relationship (Bhattacharya et al., 1998; Golembiewski and McConkie, 1975; Mayer et al.,

1995). Trust also leads to more open communication (Holden, 1990; Smith and Barclay,

1997), cooperation (Parks et al., 1996; Schlenker et al., 1973), higher quality decision-

making (Zand, 1972), risk-taking (McKnight and Chervany, 2000) and satisfaction in the

decision-making process (Driscoll, 1978).

Trust is also important in virtual teams. Lipnack and Stamps (1997), based on their

observations of several virtual teams in companies such as IBM, Sun Microsystems,

and Motorola suggest that the success and failure of virtual teams is primarily

contingent upon trust. This is because trust functions like the glue that holds and links

virtual teams together. Iacono and Weisband (1997) studied virtual teams and found

that trust among team members played an important role in team performance. The

traditional trust literature has recognized that trust is a multidimensional construct

with both cognitive (e.g. competence, reliability, professionalism) and affective

elements (e.g. caring, emotional connection to each other) (Lewis and Weigert, 1985).

The relative importance of these two elements varies depending on the context and

the type of relationship among people. According to Meyerson et al. (1996), the

formation and maintenance of trust in temporary work settings such as virtual teams

depend more on the cognitive element than the affective element. To date, however,

the differences between the cognitive and affective aspects of trust in virtual teams are

not very well understood in the literature.

In addition, the lack of a shared work history, coupled with the absence of face-to-face

communication, makes it harder for virtual team members to gather information and

evaluate one another’s behaviors. Further, the absence of face-to-face interaction creates a

sense of both physical and psychological distances between team members (O’Har-

a-Devereaux and Johansen, 1994). Thus, according to Jarvenpaa and Leidner (1999),

virtual teams need to build trust swiftly at the very outset. However, past studies on virtual

teams also found that the trust in virtual teams appeared to be fragile (Crisp and Jarvenpaa,

2000; Jarvenpaa and Leidner, 1999). It therefore, seems that it is important to maintain

trust as well as to build it. Building on these studies, in this study we ask the following

three research questions:

P. Kanawattanachai, Y. Yoo / Journal of Strategic Information Systems 11 (2002) 187–213188

Page 3: Dynamic nature of trust in virtual teams

Research question 1. What is the relative importance of cognition- and affect-based

trust in virtual team?

Research question 2. Is there a relationship between the presence of swift trust at the

beginning of the project and overall team performance?

Research question 3. Is there a difference between high- and low-performing teams in

terms of the changing patterns of trust level over time?

In addition to contributing to the virtual team literature by looking at both cognition-

and affect-based trust, our research also contributes to the trust literature by looking at the

temporal dynamic nature of trust in virtual teams. Though the role of trust has been widely

studied, few researchers have looked at its dynamic nature. Undoubtedly, however, the

temporal factor plays a critical role in the study of the group and the organization

(McGrath, 1984). Although many studies, as mentioned above, have examined trust in

virtual teams, few have explored the dynamic nature of trust at different stages over the

course of a project’s life. We believe that it is necessary to examine and empirically test

the dynamic nature of trust and its pattern of changes in both the cognitive and affective

elements of high- and low-performing teams over time.

The paper begins by reviewing the literature on trust. Our theory and hypotheses are

then presented and are followed by the research methods. We then present the results of

our study. Finally, the discussions of the results and the contributions of the study are

presented.

2. Theoretical foundation and hypotheses

2.1. Definition of virtual team

Generally, a team is defined as ‘a group of two or more individuals who must interact

cooperatively and adaptively in pursuit of shared valued objectives. Further, team

members have clearly defined differentiated roles and responsibilities, hold task-relevant

knowledge and are interdependent’ (Cannon-Bowers et al., 1993; p. 221). What

differentiates a virtual team from a traditional team is that the virtual team ‘works across

space, time and organizational boundaries with links strengthened by webs of

communication technologies’ (Lipnack and Stamps, 1997; p. 7). Many theorists have

defined such teams differently (DeSanctis and Monge, 1999; Jarvenpaa and Leidner, 1999;

Kristof et al., 1995; Lipnack and Stamps, 2000; Maznevski and Chudoba, 2000). In this

study, following DeSanctis and Monge (1999) and Jarvenpaa and Leidner (1999), a virtual

team is defined as a team whose members (1) are geographically distributed, (2) interact

electronically through the use of computer-mediated communication, (3) are functionally

diverse, and (4) work in a temporary system.

2.2. Trust

Although the concept of trust has been viewed at different levels (group, organization,

society) (Zimmer, 1972), we focus on interpersonal trust among team members, which is

P. Kanawattanachai, Y. Yoo / Journal of Strategic Information Systems 11 (2002) 187–213 189

Page 4: Dynamic nature of trust in virtual teams

defined as ‘the extent to which a person is confident in, and willing to act on the basis of,

the words, actions, and decisions of another’ (McAllister, 1995; p. 25). Interpersonal trust

is a multidimensional construct with both cognitive and affective foundations (Lewis and

Weigert, 1985).

Cognition-based trust refers to the calculative and rational characteristics demonstrated

by trustees. These include reliability (McAllister, 1995; Rempel et al., 1985), integrity,

competence (Mayer et al., 1995), and responsibility (Cook and Wall, 1980). People assess

trust based on various attributes such as certain types of professions (doctor, lawyer) and

levels of familiarity (friend vs. stranger) (Goto, 1996). One basic function of cognition-

based trust is to reduce the complexity among social actors (Luhmann, 1979). Consider an

example of landing a jet fighter on an aircraft carrier deck. A pilot has to rely on other

crews to perform different functions professionally and reliably. Trust is strengthened

when trustees do what they promise to do in a timely and professionally fashion. Without

trust, the cost of controlling and monitoring efforts is likely to escalate rapidly. The

highpoint of cognition-based trust is reached when ‘social actors no longer need or want

any further evidence or rational reasons for their confidence in the objects of trust’ (Lewis

and Weigert, 1985; p. 970). Thus, cognition-based trust relies more on information and

develops through communications among members.

On the other hand, affect-based trust involves the emotional elements and social skills

of trustees. Care and concern for the welfare of partners form the basis for affect-based

trust (McAllister, 1995; Rempel et al., 1985). Unlike cognition-based trust, which was

studied mainly in the context of working groups, affect-based trust has typically been

studied in the context of close social relationships such as couples, family members and

friends (Boon and Holmes, 1991). One can share and talk freely about problems with

another person if he or she knows that the other will listen and respond caringly. In

addition, if one is sensitive to her partner’s need, she is willing to fulfill it in both good and

bad times. Although affect-based trust is typically found to be important in the context of

close social relationships, McAllister (1995) found that even in working group

environments, it influences the performance and well-being of the teams in such a way

that one takes another’s problem as his own and is willing to give a helping hand to the

needed party even if they did not request assistance.

Few studies have empirically tested the relative importance of cognition- and affect-

based trust in working relationships. Past research in face-to-face environments shows that

the relative importance of the cognition- and affect-based trust depended on the type of

social relationship, situation, and system under consideration (Lewis and Weigert, 1985;

pp. 972–973). For example, while in close social relationships of couples and family

members, affect-based trust is higher than cognition-based trust. Conversely, cognition-

based trust would be of greater importance in a group less familiar with one another such

as a work group. Gabarro (1978) found that a cognitive aspect, especially competence, is a

key to establishing and sustaining trust in working relationships.

In the context of virtual teams, Meyerson et al. (1996), posit that people working in a

temporary system deal with each other primarily in terms of the professional roles each

individual performs, not in terms of developing social relationships. Therefore, they

argued that in temporary working systems such as virtual teams, the formation and

maintenance of trust relies more on a cognitive dimension than on an affective one.

P. Kanawattanachai, Y. Yoo / Journal of Strategic Information Systems 11 (2002) 187–213190

Page 5: Dynamic nature of trust in virtual teams

Furthermore, we argue that the communications media of virtual teams would also

influence the formation of trust in virtual teams. That is, typically, virtual teams rely on

computer-mediated communication tools as their primary means of communication

(Majchrzak et al., 2000). Although past research has shown that individuals can develop

social relationships in computer-mediated communication environments when given

enough time (Walther, 1995), there is an overwhelming body of literature that shows that

other things being equal, it is more difficult to develop social relationships through

computer-mediated communications due to the depersonalization effect (Kiesler et al.,

1984; Sproull and Kiesler, 1986). Therefore, one can expect that communications in

virtual teams are more task-oriented (Hiltz et al., 1986). Taken together, we expect that

virtual teams will show a higher degree of cognition-based trust than affect-based trust.

Thus, we hypothesize that:

Hypothesis 1. In virtual team environments, the level of cognition-based trust will be

higher than that of affect-based trust throughout the course of the project, regardless of the

team performance.

2.3. Dynamic nature of trust in virtual teams

Past studies have shown that trust in a traditional working relationship develops and

changes over time (Rempel et al., 1985; Rotter, 1980; Zand, 1972) based on on-going

interaction and the experience of working together. This is because these interactions and

experiences allow team members to learn and assess one another’s behaviors (Gabarro,

1978). Relatedly, Butler (1999) also found that an initial trust expectation is positively

related to the quantity of information shared.

In a virtual team where members normally work on a short-lived project, they might not

have enough time to gather sufficient information about their co-workers in order to

determine if their colleague is trustworthy. Moreover, the physical separation of team

members may imply that the levels of trust among virtual team members must be higher

than in traditional work relationships in order to successfully complete the project

(Hartman, 1999). The lack of engagement in a typical social greeting, such as a handshake

as well as face-to-face interaction, makes it harder for team members to establish trust in a

new working relationship. In such environments, team members need to carry out their

tasks by trusting other members from the beginning of the project, not on the basis of past

experiences, but rather on the basis of their background, professional credentials and

affiliations (Meyerson et al., 1996). Such trust is referred to as swift trust in the literature.

According to Meyerson et al. (1996), in a temporary team, “people have to wade in on trust

rather than wait while experience gradually shows who can be trusted and with what: Trust

must be conferred presumptively or ex ante” (p. 170).

Jarvenpaa and Leidner (1999) found that trust could be swiftly developed in the virtual

teams that they studied. They found that 15 of the 29 teams they studied showed a high

degree of trust from the beginning of the project. Among those 15 teams, 10 maintained a

high level of trust until the end of the project. According to the in-depth case analyses by

Jarvenpaa and Leidner, teams that started and ended with high levels of trust showed an

P. Kanawattanachai, Y. Yoo / Journal of Strategic Information Systems 11 (2002) 187–213 191

Page 6: Dynamic nature of trust in virtual teams

outstanding performance. They found that the teams with an initial high trust exchanged

background and personal information, and were more socialized with other members at the

very beginning of the project. On the other hand, they found that only four out of 14 teams

that started with low initial trust were able to overcome the initial lack of trust. Their case

analyses show that teams with low initial trust experienced several unproductive behaviors

right at the beginning (e.g. non-responsive members, unequal participation, general

unwillingness to take initiative to move the project forward, and a lack of socialization)

and most of them never overcame the initial distrust among members. Building on this, we

hypothesize that:

Hypothesis 2. In virtual teams, high-performing teams will be more likely to swiftly

develop cognition-based trust than low-performing teams.

Hypothesis 3. In virtual teams, high-performing teams will be more likely to swiftly

develop affect-based trust than low-performing teams.

Past research on trust demonstrated that the existence of trust is related to high team

performance in a way that it reduces a need for monitoring, increases organizational

citizenship behavior (McAllister, 1995), and escalates group effort and motivation (Dirks,

1999). In an early study of trust in virtual teams, Iacono and Weisband (1997) noted that

teams with a high level of trust tended to engage in continuous and frequent

communication, to focus on work content, and to adequately socialize during the early

stage of the project. They found that both the affective and cognitive elements of trust were

prominent in high-performing teams.

Yet, little is known on how trust changes over time and whether high-performing teams

follow different patterns of changes vis-a-vis low-performing teams. In a virtual team

setting, even if teams are able to initially begin with high trust, it is possible that such swift

trust can be easily destroyed later in the project. As noted earlier, in-depth case analyses by

Jarvenpaa and Leidner (1999) showed that one third of the teams that started the project

with high trust ended up with low trust and low performance. Jarvenpaa and Leidner

(1999) also found that teams that were able to maintain high levels of trust throughout the

project performed well, along with teams that overcame initially low levels of trust. In

another study, Crisp and Jarvenpaa (2000) found that trust in some teams increases over

time; however, an overall trust level in virtual teams decreases over time. However, they

did not examine whether high- and low-performing teams experienced the same

decreasing pattern of trust over time.

With one faulty action, mutual trust can be easily destroyed (Deutsch, 1958). When the

other party feels that trust is violated, cognitively, he or she assesses the degree of

violation. Affectively, he or she may get angry, experience stress, and become

disappointed (Lewicki and Bunker, 1995; p. 162). Poor performance can adversely affect

the cognition-based trust, which in turn negatively influences affect-based trust

(McAllister, 1995). Lowered trust levels will then interfere with the team’s ability to

effectively carry out its task, creating a vicious cycle of low trust and low performance. In

particular, swift trust in virtual teams can be especially vulnerable to such a vicious cycle

since it is based on expectations rather than concrete real experiences. Meyerson et al.

P. Kanawattanachai, Y. Yoo / Journal of Strategic Information Systems 11 (2002) 187–213192

Page 7: Dynamic nature of trust in virtual teams

(1996), for example, argued that once swift trust is developed, it is there to be tested and

proven through actions. Therefore, we argue that if team members cannot live up to the

expectations of others, the initial swift trust can be easily destroyed. This suggests that

swift trust exists in virtual teams only during the very early stages of the project until team

members accumulate enough experiences together, which in turn provides the basis for a

more traditional form of trust (or lack thereof).

In the case of teams that start with no swift trust, thus showing a low level of initial

trust, it will be difficult for them to overcome the lack of trust among team members, albeit

not impossible. In Jarvenpaa and Leidener’s study (1999), about two-thirds of the teams

that started with low trust could not overcome the initial barrier and performed poorly.

Taken together, past research suggests an on-going reciprocal relationship between

trust and team performance. Given that past research has demonstrated the association

between high levels of trust and team performance, it is only logical to conjecture that

high-performing teams are not only able to quickly develop but also maintain a high level

of trust, while that of low-performing teams deteriorates over time (Crisp and Jarvenpaa,

2000; Jarvenpaa and Leidner, 1999). Given the lack of past empirical studies with

longitudinal data, however, it is an empirical question precisely how quickly that trust

development will take place and the shift/change over time. Thus, we expect that:

Hypothesis 4. In virtual teams, high-performing teams will be more likely to maintain the

high level of cognition-based trust than low-performing teams.

Hypothesis 5. In virtual teams, high-performing teams will be more likely to maintain the

high level of affect-based trust than low-performing teams.

3. Method

3.1. Participants

Participants were recruited through an e-mail announcement that was broadcasted by

an Internet list-serve popular among faculty members in the information systems area. Six

different MBA courses that were taught by five professors in four different countries were

recruited for the study.2

A total of 146 MBA students (100 males; 46 females) of 10 nationalities participated in

the study. The average age and length of work experience of the participants were 28 and 5

years, respectively. Students took part in the project as part of their course and were

randomly assigned to 40, four member teams. Team members were students from four

different universities; two teams had two members from the same university. For all

participants, the participation to the game was a required part of the course that they took

and the performance of the game was accounted for 10–30% of the final course grade.

During the course of the project, four teams were removed due to having more than one

2 The second author taught two of them.

P. Kanawattanachai, Y. Yoo / Journal of Strategic Information Systems 11 (2002) 187–213 193

Page 8: Dynamic nature of trust in virtual teams

member from the same university or having a non-active member. This left 36 teams for

the data analysis.

3.2. Task

Inc 2000w, a web-based strategic business simulation game built on generic business

concepts, was used for the study. Unlike other games, Inc 2000 equally emphasizes the

four major functional areas of business-marketing, finance, production and operations, and

human resources. The engine of the game was developed by the first author and has been

used regularly in both academic institutions and corporations in more than 100 sessions

over the last 4 years. It assumes that every team has been in business for 2 years. All teams

start with the same position in terms of market shares, financial resources, human

resources, inventory, etc. Each team manages a $356 million company, producing and

selling high-end server computers in competition against other teams. In Inc 2000, the

interdependency among team members is, by design, built into the task. Each management

team makes quarterly decisions covering four major business functions. Their aim is to

make a decision collectively in order to optimize the performance of the firm as a whole.

The game was conducted over an 8-week period. Each member was randomly assigned

to one of the four business roles: VP of marketing; VP of productions and operations; VP

of finance; and VP of human resources. Apart from the first week of the project, during

which participants spent time getting to know other team members, reading the game

manual and collectively setting the vision and objectives for their fictitious companies,

teams were required to decide on 25 variables in the four functional areas each week.

Team members discussed how they should run their company for each week (from weeks

2–8—each week represented a quarterly decision in this game), primarily through text-

based, computer-mediated communication. At the end of each week (after all weekly

decisions were submitted), the game administrator processed the decisions. Each team’s

weekly performance results were then distributed. The outcomes from prior weeks were

taken into account in the subsequent week.

A web-based interface was designed to support and facilitate communication and

knowledge coordination among team members in different places (Figs. 1 and 2). The

interface design allowed the participants to (1) enter/edit/view their decisions and to

see their team’s performance, and (2) to communicate and exchange ideas/information

from anywhere at anytime through a web-based discussion database that was tightly

integrated into the game. In addition to this discussion database, members were

provided an electronic mailing list for e-mail communications. All e-mail messages

sent via the mailing lists were archived.

The web interface is purposely designed to allow only the member who is assigned to a

particular functional area to input decision variables in that area, while other members can

only view these variables once they are entered. The purpose of this split interface design,

along with the interdependence among four functional areas in the game’s logic, was not

only to make the business game more realistic, but also to make each individual’s decision

part of a larger system. Individual members’ ability to effectively interrelate actions

through communication therefore became critical to the team’s performance.

P. Kanawattanachai, Y. Yoo / Journal of Strategic Information Systems 11 (2002) 187–213194

Page 9: Dynamic nature of trust in virtual teams

Fig. 1. Decision form of a member in charge of marketing area.

Fig. 2. Team discussion database screen.

P. Kanawattanachai, Y. Yoo / Journal of Strategic Information Systems 11 (2002) 187–213 195

Page 10: Dynamic nature of trust in virtual teams

3.3. Measures

We hoped to capture the level of trust at the beginning of the project,3 at the midpoint,

and at the end of the project. Therefore, the survey was administered three times at the end

of weeks 2 (T1), 5 (T2), and 8 (T3).

A questionnaire was administered via a web page once all the decisions were

submitted. A reminder e-mail was automatically sent out 1 day before the deadline to

participants who had not yet completed the questionnaire for that respective quarter.

Except for team performance, all measures were assessed using 5-point Likert scales.

All items were listed in Table 1.

Team performance (PERF). Team performance was measured using the cumulative

financial performance (net profit)4 of each team at the end of the simulation game. Teams

were told that the goal of the game was to maximize the performance of the firm in the

market.

Disposition to trust (DT). Since individuals’ trust can be influenced by their

disposition to trust (McKnight et al., 1998), we measured individual disposition to

trust using a four-item scale developed by Pearce (1992). Respondents assessed items

by rating them on a scale ranging from 1 (Strongly disagree) to 5 (Strongly agree).

This scale was measured at T1 and T2.

Trust. We drew upon items developed by Cook and Wall (1980); McAllister (1995)

to measure cognition- and affect-based trust, respectively. Because some original

items were deployed at the dyad level (supervisor/subordinate), some wordings were

modified to suit the group-level measurement. For example, items from the original

scale such as “I can talk freely to this individual about difficulties I am having at

work and know that (s)he will want to listen.” was re-worded to “I can talk freely to

my team about difficulties I am having at work and know that my team will want to

listen.” Because the measure was developed for the different context, we re-examined

the reliability and validity of these items through a pilot study which included 55

students (13 teams) playing the Inc 2000 game. From the pool of 11 items (six items

for cognition- and five items affect-based trust, respectively) two items from

cognition-based trust and one item from affect-based trust were dropped. For the final

scale, each dimension is measured by four items. Respondents assessed items by

rating them on a scale ranging from 1 (Strongly disagree) to 5 (Strongly agree). Both

cognition- and affect-based trust were measured at T1, T2 and T3.

3 Since the teams did not start the real decision-making task until week 2, the trust level measured at the end of

week 2 is considered to be the initial trust level.4 Although there are six criteria for performance evaluation (net profit, stock price, ROA, ROE, unit cost,

and unit sold) in the instructions we gave to students, when we looked at each team’s vision, mission, and

objective, we found a consensus that in most cases each stated the goal to be the maximization of profit. A

supplemental correlation analysis also revealed that net profit is highly correlated with other five criteria

(r . 0.90). Thus, we used net profit as a single measure of team performance in the study.

P. Kanawattanachai, Y. Yoo / Journal of Strategic Information Systems 11 (2002) 187–213196

Page 11: Dynamic nature of trust in virtual teams

4. Analyses

4.1. Levels of analysis

The data were analyzed at the team level. In order to check the appropriateness of

aggregating individuals’ scores into a group-level score, we conducted two statistical tests:

Table 1

Results of confirmatory factor analysis and construct reliability of constructs for each time period (using EQS

5.7b)

Time

T1 T2 T3

Cognition-based trust (CBT); construct reliability ¼ 0.89 0.92 0.93

Most of my teammates approach his/her job with professionalism

and dedication

0.87 0.86 0.89

I see no reason to doubt my teammates’ competence and

preparation for the job

0.81 0.88 0.87

I can rely on other teammates not to make my job more difficult

by careless work

0.78 0.82 0.90

Most of my teammates can be relied upon to do as they say they

will do

0.82 0.90 0.87

Affect-based trust (ABT); construct reliability ¼ 0.86 0.90 0.93

I can talk freely to my team about difficulties I am having at work

and know that my team will want to listen

0.71 0.78 0.88

I would feel a sense of loss if one of us was transferred and we

could no longer work together

0.77 0.83 0.89

If I shared my problems with my team. I know (s)he would respond

constructively and caringly

0.79 0.84 0.88

I would have to say that we (my team) have made considerable

emotional investments in our working relationship

0.85 0.87 0.85

Disposition to trust (DT); construct reliability ¼ 0.72 0.86

Most people tell the truth about the limits of their knowledge 0.55 0.76

Most people can be counted on to do what they say they will do 0.49 0.75

Most people are honest in describing their experience and abilities 0.72 0.90

Most people answer personal questions honestly 0.74 0.73

Goodness of fit index

Chi-square 81.900 100.684 28.825

df 54 53 19

p 0.008 0.001 0.068

NFI 0.932 0.941 0.973

CFI 0.975 0.971 0.991

RMSEA 0.064 0.085 0.068

RMSEA (90% confidence interval) (0.03, 0.09) (0.06, 0.11) (0.00, 0.11)

Note. All loadings were significant; t-values ranged from 4.60 to 11.21. Construct reliability were based on

Fornell and Larcker’s formula (1981).

P. Kanawattanachai, Y. Yoo / Journal of Strategic Information Systems 11 (2002) 187–213 197

Page 12: Dynamic nature of trust in virtual teams

(1) James’ index (rwg), commonly known as the interrater agreement index (James et al.,

1984), and (2) Intraclass correlation (ICC) (Kenny and la Voie, 1985). James’ index

measures the homogeneity of members’ perceptions. Generally, an aggregation is

considered appropriate if the median rwg of the scale is greater than 0.70 (George, 1990).

The calculation of the intraclass correlation (Kenny and la Voie, 1985) is based on the

computation that compares within-group and between-group variances. To justify

aggregation, the ICC values of scale should be higher than 0.12 (James, 1982).

The results show that all rwg medians of cognition- and affect-based trust were 0.87 and

0.72, respectively, scores that are well above 0.70. Intraclass correlations of cognition- and

affect-based trust were 0.15, and 0.13, respectively, indicating that an aggregation of

individuals’ scores into the group-level score is warranted.

4.2. Test of measurement model

We first demonstrated that items would be loaded on their targeted factors. We

conducted three separate confirmatory factor analyses using the EQS 5.7b package for T1,

T2, and T3. The results showed that all items were loaded on the target factors at all three

phases. All loadings were greater than 0.70 and stable across all periods (Table 1). Also, as

shown in Table 1, all goodness of fit indexes for all periods clearly indicated that the model

fit well with the data, thus providing strong evidence for the convergent and discriminant

validity of the measures used in the study.

Then, to demonstrate the multidimensionality of the trust scale, we examined the

differences in chi-square between one- (x2ð20Þ ¼ 90:29; p , 0.001) and two-factor

(x2ð19Þ ¼ 29:88; p , 0.05) trust models. The results revealed that the two-factor model is

superior to the one-factor model, as suggested by a significant change in chi-square

(Dx2ð1Þ ¼ 60:41; p , 0.0001) as well as a moderate increase in the GFI index value of

0.13. Based on this evidence, a two-factor model appears to be warranted.

We further examined the discriminant validity using the square root of the average

variance extracted. As shown in Table 2, all square roots of the average variance extracted

displayed in a diagonal of a correlation matrix are greater than the off-diagonal construct

correlation in the corresponding rows and columns for each separate time period. This

indicated that each construct shared more variance with its items than it shared with other

constructs (Fornell and Larcker, 1981), thereby confirming the discriminant validity.

Finally, we estimated the reliability of these measures using Fornell and Larcker’s

(1981) construct reliability. All factors achieved high reliability as shown in Table 1.

Table 2 shows the descriptive statistics and the correlation matrix of all measures for all

three phases of measurement.

5. Results

We first split 36 teams into 18 high- and 18 low-performing teams based on the

performance score. Then, to make sure that individuals’ disposition to trust did not

influence the study results, we examined whether their levels of disposition to trust vary

across time and team performance. An analysis of variance (ANOVA) revealed no

P. Kanawattanachai, Y. Yoo / Journal of Strategic Information Systems 11 (2002) 187–213198

Page 13: Dynamic nature of trust in virtual teams

Table 2

Means, standard deviations and correlation of constructs at the group-level (n ¼ 36)

Low-performing

teams

Hi-perform-

ing teams

All teams Correlation of constructs (all teams)a

Mean SD Mean SD Mean SD ABT1 CBT1 PERF1 ABT2 CBT2 PERF2 ABT3 CBT3 PERF3

Time 1 ABT1 2.84 0.63 2.73 0.60 2.78 0.61 0.78

CBT1 3.19 0.52 3.25 0.55 3.22 0.53 0.74** 0.82

PERF1 43.94 20.51 65.28 19.36 54.61 22.44 20.14 20.04 na

Time 2 ABT2 2.88 0.56 3.06 0.47 2.97 0.51 0.43** 0.33 0.19 0.83

CBT2 3.25 0.60 3.60 0.56 3.42 0.60 0.18 0.29 0.29 0.54** 0.87

PERF2 36.33 17.34 69.72 14.91 53.03 23.25 20.03 0.14 0.22 0.32 0.34* na

Time 3 ABT3 2.66 0.50 2.93 0.80 2.80 0.67 0.39* 0.16 0.15 0.74** 0.45** 0.26 0.88

CBT3 3.00 0.60 3.34 0.80 3.17 0.72 0.20 0.29 0.08 0.54** 0.59** 0.47** 0.65** 0.88

PERF3 39.61 18.14 64.50 12.04 52.06 19.74 20.01 0.09 0.26 0.21 0.24 0.58** 0.30 0.39* na

CPERFb 217.06 38.53 44.72 19.76 13.83 43.50 20.03 0.10 0.21 0.33* 0.42* 0.72** 0.34* 0.54** 0.68**

* Correlation is significant at the 0.05 level (2-tailed); ** correlation is significant at the 0.01 level (2-tailed).a Diagonal boldface elements were the square root of the average variance extracted.b Cumulative performance.

P.

Ka

na

wa

ttan

ach

ai,

Y.

Yo

o/

Jou

rna

lo

fS

trateg

icIn

form

atio

nS

ystems

11

(20

02

)1

87

–2

13

19

9

Page 14: Dynamic nature of trust in virtual teams

differences in the levels of disposition to trust (F(1, 238) ¼ 0.69, p ¼ 0.41) between a

high-performing team (mean ¼ 3.33, S.D. ¼ 0.76) and a low-performing team

(mean ¼ 3.25, S.D. ¼ 0.72). The ANOVA results also revealed no changes of disposition

to trust between time 1 and time 2 (F(1,238) ¼ 0.66, p ¼ 0.42). These results indicated

that individuals working in both high- and low-performing teams have no difference in the

level of their disposition to trust at the beginning and the middle stages of the project.

We expected that in virtual environments, teams would develop a higher level of

cognition-based trust than affect-based trust (Hypothesis 1). To test our hypothesis, we

conducted a paired t-test comparing the levels of cognition-based and affect-based trust at

all three periods, regardless of team performance. The results strongly support Hypothesis

1. The level of cognition-based trust is higher than that of affect-based trust at all three

periods (t(35) ¼ 6.23, 5.05, and 3.84, for T1, T2, and T3; p , 0.0001 for all three times).

Hypotheses 2 and 3 aim at identifying the differences between high- and low-

performing virtual teams in terms of cognition- and affect-based trust at the outset of the

project. We hypothesized that high-performing teams would be more likely to show higher

initial cognition-based trust than low-performing teams (Hypothesis 2). In order to

examine these hypotheses, we first dichotomized samples (using median scores) into high-

and low-trust teams in T1 for cognition-based trust. One-way ANOVA results show that

there were significant differences (F(1, 34) ¼ 54.11) between high (mean ¼ 3.63) and low

trust (mean ¼ 2.81) teams in terms of the cognition-based trust level at T1. If the H2 were

true, we would expect to see (a) a large number high-performing teams in the high-trust

group in T1, which reflects the presence of swift trust, and (b) a large number of low-

performing teams to be found in low-trust teams, which reflects the absence of swift trust.

To examine this statistically, we conducted a crosstabs analysis (Table 3) with a chi-square

test with team performance and cognition-based trust at T1. The chi-square test

(x2ð1Þ ¼ 4; p ¼ 0.046) shows that there was a statistically significant relationship between

team performance and the presence of swift trust in the cognitive dimension. More

specifically, the results show that two-thirds of high-performing teams showed initial swift

trust in cognitive dimension, while only one-third of low-performing teams showed swift

trust. Thus Hypothesis 2 was supported.

We also hypothesized that high-performing teams would be more likely to show higher

initial affect-based trust than low-performing teams (Hypothesis 3). Following the same

logic that we used to test Hypothesis 2, we again dichotomized the sample into high- and

low-trust teams in T1 for affect-based trust. One-way ANOVA results show that there

Table 3

Crosstabulation of performance £ cognition-based trust (T1)

Performance CBT

Low High Total

Low 12 6 18

High 6 12 18

Total 18 18 36

P. Kanawattanachai, Y. Yoo / Journal of Strategic Information Systems 11 (2002) 187–213200

Page 15: Dynamic nature of trust in virtual teams

were significant differences (F(1, 34) ¼ 61.76) in terms of the affect-based trust level

between high (mean ¼ 3.27) and low trust (mean ¼ 2.30) teams at T1. We then conducted

a crosstabs analysis with chi-square test (Table 4). The results showed that there was no

statistically significant relationship between team performance and the presence of swift

trust in the affect-based level (x 2(1) ¼ 1.78; p ¼ 0.182). Thus, Hypothesis 3 was rejected.

Finally, we hypothesized that high-performing teams would be more likely to

maintain trust at a high level for both dimensions than low-performing teams

(Hypotheses 4 and 5). To test our hypotheses, we conducted four repeated-measures

ANOVA tests that examine the changing patterns of cognition-based trust and affect-

based trust in high- and low-performing teams. The repeated-measures ANOVA allow

us to see whether the changing patterns of trust over time follow linear or quadratic

patterns (Bray and Maxwell, 1985). If the repeated-measures ANOVA detected either

linear or quadratic patterns, we performed post-hoc pairwise comparisons among T1,

T2, and T3 using the Scheffe test to understand the nature of the changing patterns of

trust more clearly. One can view the three repeated observations of cognition-based

trust and affect-based trust (within-subjects factor) on each team as nested within the

team. In the repeated measures ANOVA analyses, each team is blocked so that the

variability among teams is completely removed from the error term (Stevens, 1996).

Using the method proposed by Green (1990), sample sizes for repeated measures

needed for power ¼ 0.80 in a single group measuring the subjects at three points in

time for the small effect size at alpha ¼ 0.05 are 14 (for average correlation of the

subjects’ responses to the scales at three points).

Fig. 3 graphically presented a pattern of changes in trust levels between high- and low-

performing teams over time. The results of these tests are shown in Table 5. The repeated-

measures ANOVA for the cognition-based trust of high-performing teams revealed a

significant quadratic trend (F(1, 17) ¼ 5.31, p ¼ 0.03). To better understand the nature of

this trend, we conducted post-hoc pairwise comparisons between T1 vs. T2 and T2 vs. T3.

The results of the post-hoc test showed that there was a significant increase of cognition-

based trust in high-performing teams from T1 (mean ¼ 3.25, S.D. ¼ 0.55) to T2

(mean ¼ 3.60, S.D. ¼ 0.56) (F(1, 17) ¼ 4.60, p ¼ 0.047). However, there was no

significant difference between T2 and T3 (mean ¼ 3.34, S.D. ¼ 0.80) (F(1, 17) ¼ 2.85,

p ¼ 0.11). This suggests that high-performing teams in our sample were able to develop

high levels of cognition-based trust during the first half of the project and maintained these

Table 4

Crosstabulation of performance £ affect-based trust (T1)

Performance ABT

Low High Total

Low 11 7 18

High 7 11 18

Total 18 18 36

* Correlation is significant at the 0.05 level (2-tailed).

P. Kanawattanachai, Y. Yoo / Journal of Strategic Information Systems 11 (2002) 187–213 201

Page 16: Dynamic nature of trust in virtual teams

Fig. 3. Pattern of changes in trust levels between high- versus low-performing teams over time.

Table 5

Results of pairwise comparisons

Source SS df MS F Sig.

CBT-high-performing teams

TIME Linear 0.09 1 0.09 0.31 0.59

Quadratic 1.13 1 1.13 5.31* 0.03

Error Linear 4.69 17 0.28

Quadratic 3.62 17 0.21

CBT-low-performing teams

TIME Linear 0.35 1 0.35 1.19 0.29

Quadratic 0.27 1 0.27 1.87 0.19

Error Linear 5.05 17 0.30

Quadratic 2.49 17 0.15

ABT-high-performing teams

Time Linear 0.36 1 0.36 1.31 0.27

Quadratic 0.59 1 0.59 4.22 0.06

Error Linear 4.67 17 0.28

Quadratic 2.37 17 0.14

ABT-low-performing teams

Time Linear 0.28 1 0.28 1.36 0.26

Quadratic 0.22 1 0.22 2.81 0.11

Error Linear 3.45 17 0.20

Quadratic 1.30 17 0.08

P. Kanawattanachai, Y. Yoo / Journal of Strategic Information Systems 11 (2002) 187–213202

Page 17: Dynamic nature of trust in virtual teams

levels during the second half.5 On the contrary, the repeated-measures ANOVA of

cognition-based trust of low-performing teams did not show any significant results. This

implies that the level of cognition-based trust of low-performing teams did not change

throughout the project life cycle.

We found an almost identical pattern for affect-based trust. Again, the repeated-

measures ANOVA for affect-based trust of the high-performing teams showed a moderate

significant quadratic trend (F(1, 17) ¼ 4.22, p ¼ 0.056). The post-hoc test revealed a

significant increase in affect-based trust from T1 (mean ¼ 2.73, S.D. ¼ 0.60) to T2

(mean ¼ 3.06, S.D. ¼ 0.47) (F(1, 17) ¼ 4.56, p ¼ 0.048), while showing no statistically

significant changes from T2 to T3 (mean ¼ 2.93, S.D. ¼ 0.80) (F(1, 17) ¼ 0.93,

p ¼ 0.35). On the other hand, the repeated-measures ANOVA of affect-based trust of

the low-performing teams did not show any significant results. Thus, we concluded that

the level of affect-based trust of low-performing teams did not change over time.

To gain a better understanding of the changing patterns of cognition- and affect-based

trust in high- and low-performing teams, we extended the analyses that we conducted to

test Hypotheses 2 and 3. That is, we further dichotomized the sample for cognition- and

affect-based trust for each phase. Then, we tracked each team for all three phases in terms

of their trust levels (high vs. low). The summary of the results as shown in Table 6

provides interesting insights. Among high performing teams, 28% and 33% of teams

started with high levels of cognition- and affect-based trust, respectively, and maintained

these trust levels throughout the project (pattern 1). To the contrary, only 6% (cognition-

based) and 17% (affect-based) of low-performing teams were able to maintain a high trust

level. It is also interesting to note that 11% (cognition-based) and 17% (affect-based) of

high-performing teams started with low trust and never managed to improve these levels

(pattern 8). At the same time, a much larger percentage of low-performing teams (28% for

both dimensions of trust) had low levels of trust throughout the project. Another

interesting observation is that among the teams that ended with a high level of trust in T3

(pattern 1, 3, 5, and 7), a much larger number of teams started with a high level of

cognition- (45%) and affect-based trust (50%) among high-performing teams than low-

performing teams (12 and 17% for cognition- and affect-based trust, respectively). Thus,

Hypotheses 4 and 5 were both supported.

6. Discussion

Our main focus of this study was to examine the dynamic nature of trust in virtual

teams and the pattern of changes of both cognition- and affect-based trust between

5 An examination of the mean values of cognition-based trust of high- (3.25) and low-performing teams (3.19)

might lead one to believe that there is not enough evidence for the presence of swift trust among high-performing

teams, contradicting our analysis of H2. However, our test of the presence of swift trust among high-performing

teams was based on the number of high-performing teams that were classified as high-trust teams in T1. One must

note that as shown in Table 3, there were six low-performing teams that showed swift trust and six high-

performing teams that showed low initial trust at T1. Therefore, it is possible that although there were a

significantly larger number of high-performing teams that showed swift trust than of low-performing teams at T1,

the difference in mean scores of trust between high- and low-performing teams is statistically insignificant.

P. Kanawattanachai, Y. Yoo / Journal of Strategic Information Systems 11 (2002) 187–213 203

Page 18: Dynamic nature of trust in virtual teams

high- and low- performing teams over time. In research question 1, we inquired about

the relative importance of cognition- and affect-based trust in virtual teams. We found

that virtual teams in our sample developed a higher-degree of cognition-based trust than

that of affect-based trust. In research question 2, we asked whether there is a

relationship between the presence of swift trust at the outset of a project and overall

team performance. We found that the presence of swift trust in the cognitive dimension

is related to team performance. However, we failed to find a similar relationship in

affect-based trust. Finally in research question 3, we asked whether there is a difference

between high- and low-performing teams in terms of the changing patterns of trust

level over time. We found that high-performing teams not only quickly establish trust

at the beginning, but also manage to maintain it at a high level throughout the project.

Our results strongly support the proposition made by Meyerson et al. (1996) that in

temporary work teams like those in our study, the cognitive element is more important

than the affective element of trust. While we do not downplay the importance of affect-

based trust, we emphasize that virtual teams should explicitly attempt to develop

cognition-based trust in virtual teams.

Consistent with past research on virtual teams (Jarvenpaa and Leidner, 1999), we found

the evidence of swift trust at the beginning of the project among high-performing teams.

Table 6

Frequency of the changing patterns of cognition- and affect-based trust in high- and low-performing teams

Pattern High-performing Low-performing All teams

Frequency % Frequency (%) Frequency (%)

Cognition-based trust

1. HHH 5 28 1 6 6 17

2. HHL 1 6 2 11 3 8

3. HLH 3 17 1 6 4 11

4. HLL 3 17 2 11 5 14

5. LHH 2 11 4 22 6 17

6. LHL 2 11 1 6 3 8

7. LLH 0 0 2 11 2 6

8. LLL 2 11 5 28 7 19

Total 18 100 18 100 36 100

Affect-based trust

1. HHH 6 33 3 17 9 25

2. HHL 0 0 2 11 2 6

3. HLH 3 17 0 0 3 8

4. HLL 2 11 2 11 4 11

5. LHH 1 6 3 17 4 11

6. LHL 2 11 1 6 3 8

7. LLH 1 6 2 11 3 8

8. LLL 3 17 5 28 8 22

Total 18 100 18 100 36 100

Note. H ¼ high trust; L ¼ low trust; the three letters represent trust level at T1–T3, respectively. For

example, HHL means high trust at T1 and T2, and low trust at T3.

P. Kanawattanachai, Y. Yoo / Journal of Strategic Information Systems 11 (2002) 187–213204

Page 19: Dynamic nature of trust in virtual teams

In fact, about two-thirds of the high-performing teams showed the presence of swift trust

in the cognitive dimension at T1. Although we do not attempt to draw any conclusive

causal relationship between swift trust and team performance, it at least seems safe to say

that the presence of swift trust in the cognitive dimension can be used as an early predictor

of team performance. Our results also suggest that the teams that establish swift trust at the

beginning of the project are more likely to maintain a level of trust throughout the project.

We also found that as teams moved along, high-performing teams are more likely to

maintain their levels of cognition- and affect-based trust. High-performing teams seemed

able to bridge the physical and psychological distances through both dimensions of trust.

For example, a member of a high-performing team wrote in his e-mail in week 8:

I definitely think our success is a direct result of ourcommitment and professionalism in the pursuit of a commongoal. Best of luck to each of you in your future endeavors.

This e-mail message indicates that these team members developed a high degree of

cognition-based trust among each other as a result of their ability to perform their

individual tasks in a professional way. In addition to cognition-based trust, evidence of

affect-based trust can also be found in most high-performing teams. Members exchanged

photos, sent a birthday card in PowerPoint file, shared the feelings of becoming a parent,

congratulated each other on new jobs, and invited other members to visit their homes.

Below is an e-mail message in week 4 from another high-performing team, expressing

emotional bonds the team had.6

Team,First of all, congratulations Steve!!! My little girl is 10mos.old, and as a relatively new father I can tell you that it is thegreatest thing that ever happened to me.I suggest you order 2500 of each part, and produce 2500 units.Increase maintenance level to 3, R&D to 4,500,000, and pro-ductivity improvement to 7,000.I know you have more important things to right now, so I’ll checkthe decision around 3pm. If you haven’t had a chance to take careof it, I’ll e-mail [the game administrator] and ask him to enterthe decision.Best wishes to you and your family.John

One could argue that high-performing teams were able to continue to perform at a high

level since trust among the team members facilitated the flow of knowledge and

cooperation (Deutsch, 1958; Huemer et al., 1998) while reducing the level of uncertainty

(Sorrentino et al., 1995). At the same time, it is also important to recognize that high

performance helps team members develop both cognition- and affect-based trust.

6 All names are disguised.

P. Kanawattanachai, Y. Yoo / Journal of Strategic Information Systems 11 (2002) 187–213 205

Page 20: Dynamic nature of trust in virtual teams

The excerpts taken from a series of e-mail messages exchanged among the members of a

high-performing team in weeks 4 and 5 show the dynamic interplay between both

dimensions of trust and team performance.

Week 4: 10/3/2000 9:01:27 PM [HR manager]

Hi all,

[Detailed HR decision was proposed.]

Summary: 2 more laborers, 2 more salespersons, 4% sales commissions, and 10%

production incentive.

Comments?

P.S. I started my new position at [name of a company] (Public relations firm) on

Monday (10/2). I am an accountant doing Financial Statement consolidations or at least

I will be in the future when I’m familiar with my new position. Also, I’m still trying to

get my computer set up at home (I had a laptop at my last job which also served as my

home computer) so I will probably not put my decisions in until Friday morning, but I’ll

try for Thursday morning. Sorry if this causes any inconveniences.

Week 4: 10/4/2000 1:00:51 AM [Marketing manager]

Hi,

your HR justification sounds fine with me Jeff…and good luck in your new job. I think

our decision is fairly well finalised for this week (pending any comments from the CEO

of course)!!!

Regards,

Andrew.

Week 4: 10/4/2000 4:19:42 PM [Production manager]

Excellent job, thank you. (I did not had to do anything…) I took back production

increase and left the other variables as agreed. My only concern is that we changed a

little bit too many variables at one time so it will be harder to detect direct relationships.

But I also think that there are few rounds and we need to act if we want to break through.

That’s the only comment as CEO. I think we are OK, but be sure you did type in your part

to the decision form. I will have a look at on Friday and write if anything comes up. Bye.

Week 4: 10/4/2000 9:32:42 PM [Finance manager]

After reviewing everyone’s input, I agree to not increasing production capability this

quarter. The proposed sales force should be sufficient along with the price increase.

Thanks for ordering those specific reports; if no value added after this quarter, we can

drop them. Just wanted to see one more round at a minimum.

I dropped the stock repurchase part! If someone disagrees, just let me know!

I increased the dividend pay out from $200 K to $300 K in an effort to see what our ROE

will do while dropping the stock repurchase (as mentioned above). Hopefully it will

increase our overall

Equity!

Congrats to Jeff with the new job—enjoy and to Jenny as the CEO!

Regards,

Alan

Week 5: 10/8/2000 8:28:45 PM [Marketing manager]

you beauty,

P. Kanawattanachai, Y. Yoo / Journal of Strategic Information Systems 11 (2002) 187–213206

Page 21: Dynamic nature of trust in virtual teams

we are now the leaders!! no other team has sold more units than us, and we have the best

cumulative ROA and ROE. We again sold all of our servers at a selling price of $36,000

which I think was the highest selling price of all teams this quarter—great for our profit

and cash flow. Well done folks!!!

[Detailed marketing decision was proposed.]

Regards,

a very, happy Andrew.

p.s. Jeff is CEO this quarter

Week 5: 10/10/2000 12:16:16 PM [Finance manager]

I concur, the team did very well overall! Congrats to all!

The price drop while keeping other variables constant (R&D, Advertising, etc) should

be a good strategy for the next quarter input!

[Detailed financial decision was proposed]

Take care and will be in touch!

Once again—good job everyone!

Regards,

In the above example, team members showed cognition-based trust toward other

members in terms of their ability to make sound decisions. Yet, such cognition-based trust

was reinforced through the team’s continuing success in the game. Furthermore, team

success also improved affect-based trust as well. Our results thus indicate that the

relationship between trust and team performance is reciprocal.

On the contrary, low-performing teams experienced no change in the level of

cognition- and affect-based trust. In fact, our post-hoc analysis showed that there was a

significant drop in the affect-based trust of low-performing teams from T2 (mean ¼ 2.88,

S.D. ¼ 0.56) and T3 (mean ¼ 2.66, S.D. ¼ 0.50) (F(1, 17) ¼ 6.70, p ¼ 0.02), although it

did not influence the overall pattern as examined by the repeated-measured ANOVA. One

can argue that the low performance level of these teams can be attributed to their inability

to develop an adequate level of trust among team members in virtual team environments;

this in turn hindered the cooperation among and withdrawal of members (Luhmann, 1979),

which further deteriorated their trust.

Our post-hoc observation of the communication contents of low-performing teams

indicate that inconsistencies between words and actions can easily destroy trust. The

following comment is from a member of a low-performing team in week 2.

A week after my initial e-mail, there was still no input fromother team members. Promises were made to get up to speed andprovide thoughts on our options but nothing was delivered.

Cognition-based trust among team members can be also negatively influenced by the

careless work of one member. An entire week’s effort to make a sound decision was

worthless when this team discovered that one member did not enter decisions based on

what the team as a whole had agreed upon. The following is an excerpt from an e-mail

message from a member of a low-performing team in week 4.

P. Kanawattanachai, Y. Yoo / Journal of Strategic Information Systems 11 (2002) 187–213 207

Page 22: Dynamic nature of trust in virtual teams

dear Adam,i don’t know what has happened in the end. are we increasingplant capacity or not. i guess the whole discussion startedbecause we all had the consensus that we were increasingcapacity.let us first understand the rule of the game. we are a part of thesame team. whether we like it or not we are going to sink or swimtogether. if your VP marketing is doing a rotten job then beingthe part of the same team you are also responsible for it. and ifmy VP production is doing a good job then no matter how lousy aperson i am, in this virtual team i am going to share the credit.therefore instead of pulling the legs of each other let us helpeach other.now the question arises how can we help each other. if thequarter ends at 4 o’clock on 30th september and the VPProduction wakes up at 3.30 just half an hour before the quarterends and tell us that the whole quarter we have been doing a lousyjob and that he does not accept the consensus then naturallythis is not going to help.[…omitted]

In this e-mail, the participant showed a great deal of concern about other members’

commitment to the task and the quality of their decisions. Not surprisingly, this team

showed low degree of cognition-based trust. A similar example can be found in the

following e-mail message in week 4 from another low-performing team.

Once again, it is very difficult to make a decision about HR when Ihave no idea how much we plan to produce! Whoever is in charge ofOperations really needs to input decisions earlier in the game.The decisions to hire/fire takes affect in two periods so if youplan to increase/decrease production I need to know in advance.Please help by inputting/discussing your decision earlier inthe game.

Reflecting the low degree of affect-based trust, many low-performing teams did not

even exchange farewell messages to each other.

6.1. Limitations

Our study has several limitations. First, we arbitrarily assigned roles to participating

students. We felt that this might have suppressed the influence of individual team

members’ real abilities and expertise, which consequently might have suppressed the level

of cognition-based trust. Future research should attempt to align participants’ expertise

with their assigned role. Alternatively, future research may also examine the dynamic

P. Kanawattanachai, Y. Yoo / Journal of Strategic Information Systems 11 (2002) 187–213208

Page 23: Dynamic nature of trust in virtual teams

nature of trust in real organizational settings. These alternative designs will allow us to see

the real impact of cognition-based trust in virtual teams.

Second, we measured trust through perceptual measures. Although these measures

were taken from the literature and showed excellent measurement properties, we felt that a

better understanding can be gained by analyzing team members’ communicative actions.

In the context of virtual teams where members are geographically separated, this means

one needs to examine the contents of their communication interactions. A micro-level

content analysis of team communication interactions would undoubtedly improve our

understanding in this area.

Third, turning to the issue of external validity, we acknowledge the variations and

different types of virtual teams. The virtual teams in this study represented an extreme case

of virtual teams where members have no prior history of working together. In real world

settings, some of them may have known each other and worked together before. Moreover,

the limited choices of communication technology in this study (e-mail and thread-style

discussion) may not be realistic when other forms of technology such as video

conferencing, instant messaging, and other groupware tools are widely available,

accessible and affordable.

6.2. Implications for virtual team management and future research

Despite these limitations, a meaningful and interpretable pattern of changes in trust

levels over time was attained. As an initial empirical study for a better understanding of the

dynamic nature of trust, our study provides a few important implications for the practice of

virtual team management. First, our study confirms the findings of past virtual team

research which identified the impact of trust among team members on team performance.

As we noted at the beginning of the paper, many organizations today utilize virtual team

methods both internally and externally in order to stay ahead of the competition. There are

abounding examples of virtual teams that carry strategically important tasks (Majchrzak

et al., 2000; Malhotra et al., 2001). Therefore, today’s managers need to recognize the

strategic importance of managing trust in teams that are technology-enabled.

Second, our findings suggest that the presence of swift trust at the early stage of a

project can have diagnostic value for managing virtual teams. As shown in our study, the

presence of swift trust might suggest a high probability of achieving high performance and

lasting trust among team members. On the contrary, the lack of swift trust can be

interpreted as an early warning sign for virtual team managers. Managers can therefore

utilize various instruments in order to assess the levels of trust at the earliest stage and

detect possible warning signs of trouble in their teams.

Third, our finding implies virtual team members rely more on cognition-based trust

than on affect-based trust throughout the project, as demonstrated in the testing of

Hypothesis 1. Thus, we suggest that the managers of virtual teams should focus on

cognitive dimension of trust. While the typical socialization strategies suggested by

previous virtual team studies might help teams develop affect-based trust, they probably

will not adequately develop and maintain cognition-based trust. Given that cognition-

based trust is based on a calculative, rational process, managers need to provide

P. Kanawattanachai, Y. Yoo / Journal of Strategic Information Systems 11 (2002) 187–213 209

Page 24: Dynamic nature of trust in virtual teams

task-relevant background information on virtual team members so that members can

quickly develop cognition-based trust as well as affect-based trust.

Fourth, we suggest that the managers of virtual teams should focus on the maintenance

as well as the development of trust in virtual teams. Our results clearly showed that a large

number of high-performing teams were able to maintain high levels of both dimensions of

trust until the end of the project. Again, typical socialization strategies may help managers

develop trust, but they may not be enough to maintain it once conflicts among team

members emerge. Thus, managers need to be equipped with various conflict resolution

strategies in order to alleviate conflict before it leads to the degradation of trust among

members.

Finally, technology used in virtual teams should not merely facilitate effective and

efficient information sharing and exchanges among teams. One should also look into a

possibility of using technology to create a digital workplace atmosphere that supports a

building of trust. Video conferencing, for example, could reduce the sense of both physical

and psychological distance.

Our study also provides several directions for future research. First, we raised many

new interesting research questions related to the changing patterns of trust over time. How

are certain teams able to quickly establish swift trust at the outset of the project? How does

swift trust become experience-based trust or distrust in virtual teams? How and why does

the level of trust in both dimensions change over time? What particular incidents lead up to

the rise and fall of trust among team members in virtual teams? Second, as indicated

earlier, future research can examine the micro-level communication processes by which

teams develop different levels of cognition- and affect-based trust over time. Such studies

can provide invaluable insights to managers and researchers alike about how to ‘read’ the

health of the team in terms of trust in both dimensions.

Third, recent studies suggest that socio-cognitive constructs such as transactive

memory and collective mind have a direct and important influence on team

performance above and beyond typical socio-psychological constructs such as trust.

Future studies can examine how those socio-cognitive constructs and socio-

psychological constructs are interrelated in virtual teams. Fourth, we studied trust

among peer groups without a designated leader. However, past research in face-to-

face environments shows that trust in leaders is also important for team performance

(Roberts and O’Reilly, 1974). Future studies can examine the influence of leader-

subordinate trust on team performance.

Fifth, findings on the study focus on the pattern of changes in trust level between

high- versus low-performing teams over time. It will be important in future work to

examine the causal as well reciprocal relationships of trust and performance in a

longitudinal fashion. Sixth, although our study did not look at the impact of cultural

differences among team members regarding the development of trust, it is well

understood that national culture influences individuals’ attitudes and behaviors

(Hofstede, 1983). Kayworth and Leidner (2000) found from their study of 12 global

virtual teams, language and time differences can make it difficult for team members to

communicate and interpret meanings accurately in cross-cultural settings. The cultural

differences thus add another layer of complexity in the development of trust. Finally,

future research needs to consider the role of technology artifacts and their features in

P. Kanawattanachai, Y. Yoo / Journal of Strategic Information Systems 11 (2002) 187–213210

Page 25: Dynamic nature of trust in virtual teams

the formation of swift trust and its maintenance in virtual teams. Given the prominent

roles that communication technology play in virtual team communications and the

emergence of agent technology, we need to examine if the meaning of trust among

distributed individual members can be extended to technology artifacts that may act

on behalf of other members.

References

Bhattacharya, R., Devinney, T.M., Pillutla, M.M., 1998. A formal model of trust based on outcomes.

Academy of Management Review 23 (3), 459–472.

Biggs, M., 2000. Assessing risks today will leave corporate leaders well-prepared for the future of

work. InfoWorld September http://www.infoworld.com/articles/op/xml/00/90/25/

000925opbiggs.xml.

Boon, S.D., Holmes, J.G., 1991. The dynamics of interpersonal trust: resolving uncertainty in the

face of risk. In: Hinde, R.A., Groebel, J. (Eds.), Cooperation and Prosocial Behavior, Cambridge

University Press, Cambridge, England, pp. 190–211.

Bray, J., Maxwell, S., 1985. Multivariate Analysis of Variance, Sage, Newbury, CA.

Butler, J.K.J., 1999. Trust expectations, information sharing, climate of trust, and negotiation

effectiveness and efficacy. Group and Organization Management 24 (2), 217–238.

Cannon-Bowers, J.A., Salas, E., Converse, S., 1993. Shared mental models in expert team decision

making. In: Castellan, N.J., (Ed.), Individual and Group Decision Making: Current Issues,

Lawrence Elbaum Associates, Hillsdale, NJ, pp. 221–246.

Cook, J., Wall, T., 1980. New work attitude measures of trust, organizational commitment and

personal need non-fulfilment. Journal of Occupational Psychology 53 (1), 39–52.

Crisp, C.B., Jarvenpaa, S.L., 2000. Trust Over Time in Global Virtual Teams, Academy of

Management Meeting, Toronto,.

DeSanctis, G., Monge, P., 1999. Introduction to the special issue: communication processes for

virtual organizations. Organizational Science 10 (6), 693–703.

Deutsch, M., 1958. Trust and suspicion. Journal of Conflict Resolution 2 (4), 265–279.

Dirks, K.T., 1999. The effects of interpersonal trust on work group performance. Journal of Applied

Psychology 84 (3), 445–455.

Driscoll, J.W., 1978. Trust and participation in organizational decision making as predictors of

satisfaction. Academy of Management Journal 21 (1), 44–56.

Fornell, C., Larcker, D., 1981. Evaluating structural equation models with unobservable variables

and measurement error. Journal of Marketing Research 18, 39–50.

Gabarro, J.J., 1978. The development of trust, influence, and expectations. In: Athos, A.G., Gabarro,

J.J. (Eds.), Interpersonal Behavior: Communication and Understanding in relationships, Prentice-

Hall, Englewood Cliffs, NJ.

George, J.M., 1990. Personality, affect, and behavior in groups. Journal of Applied Psychology 75,

107–116.

Golembiewski, R.T., McConkie, M., 1975. The centrality of interpersonal trust in group process. In:

Cooper, C.L., (Ed.), Theories of Group Processes, Wiley, New York, pp. 131–185.

Goto, S.G., 1996. To trust or not to trust: situational and dispositional determinants. Social Behavior

and Personality 24 (2), 119–131.

Green, S.G., 1990. Power Analysis in Repeated Measures Analysis of Variance with Heterogeneity

Correlated Trials, The Annual Meeting of the American Educational Research Associate, Boston,

MA,.

P. Kanawattanachai, Y. Yoo / Journal of Strategic Information Systems 11 (2002) 187–213 211

Page 26: Dynamic nature of trust in virtual teams

Hartman, F., 1999. Teams and team building. In: Dorf, R.C., (Ed.), The Technology Management

Handbook, CRC Press/IEEE Press, Boca Raton, FL, pp. 8–12.

Hiltz, S.R., Johnson, K., Turoff, M., 1986. Experiments in group decision making, designated

humand leaders and statistical feedback in computerized conferences. Journal of Management

Information Systems 8, 81–108.

Hofstede, G., 1983. The cultural relativity of organizational practices and theories. Journal of

International Business Studies 14 (2), 75–89.

Holden, R.K., 1990. An exploratory study of trust in buyer–seller relationships. Unpublished

doctoral dissertation, Boston University.

Huemer, L., von Krogh, G., Roos, J., 1998. Knowledge and the concept of trust. In: von Krogh, G.,

Roos, J., Kleine, D. (Eds.), Knowing in Firms: Understanding, Managing and Measuring

knowledge, Sage, London, pp. 123–145.

Iacono, C.S., Weisband, S., 1997. Developing Trust in Virtual Teams, Proceedings of the thirtieth

Hawaii International Conference On System Sciences, Hawaii, January.

James, L.R., 1982. Aggregation bias in estimates of perceptual agreement. Journal of Applied

Psychology 67 (2), 219–229.

James, L.R., Demaree, R.G., Wolf, G., 1984. Estimating within-group interrater reliability with and

without response bias. Journal of Applied Psychology 69, 85–98.

Jarvenpaa, S.L., Leidner, D.E., 1999. Communication and trust in global virtual team. Organization

Science 10 (6), 791–865.

Kayworth, T., Leidner, D.E., 2000. The global virtual manager: a prescription for success. European

Management Journal 18 (2), 183–194.

Kenny, D.A., la Voie, L., 1985. Separating individual and group effects. Journal of Personality and

Social Psychology 48 (2), 339–348.

Kiesler, S., Siegel, J., McGuire, T.W., 1984. Social psychological aspects of computer-mediated

communication. American Psychologist 39 (10), 1123–1134.

Kristof, A.L., Brown, K.G., Simps, H.P., Smith, K.A., 1995. In: Beyerlein, M.M., Johnson,

D.A., Beyerlein, S.T. (Eds.), The virtual team: a case study and inductive model, Advances

in Interdisciplinary Studies of Work Teams, vol. 2. Jossey-Bass, San Francisco, pp.

229–253.

Lewicki, R.J., Bunker, B.B., 1995. Trust in relationships: a model of development and decline (H.C.

1995, Trans.). In: Bunker, B.B., Rubin, J.Z. (Eds.), Conflict, Cooperation, and Justice: Essays

Inspired by the Work of Morton Deutsch, Jossey-Bass, San Francisco, pp. 133–173.

Lewis, J.D., Weigert, A., 1985. Trust as a social reality. Social Forces 63 (4), 967–985.

Lipnack, J., Stamps, J., 1997. Virtual Teams: Reaching Across Space, Time and Organizations with

Technology, Wiley, New York.

Lipnack, J., Stamps, J., 2000. Virtual Teams: People Working Across Boundaries with Technology,

2nd ed., Wiley, New York.

Luhmann, N., 1979. Trust and Power, Wiley, New York.

Majchrzak, A., Rice, R.E., Malhotra, A., King, N., Ba, S., 2000. Technology adaptation: the case of

computer-supported inter-organizational virtual team. MIS Quarterly 24 (4), 569–600.

Malhotra, A., Majchrzak, A., Carman, R., Lott, V., 2001. Radical innovation without collocation: a

case study at Boeing-Rocketdyne. MIS Quarterly 25 (2), 229–250.

Mayer, R.C., Davis, J.H., Schoorman, F.D., 1995. An integration model of organizational trust.

Academy of Management Review 20 (3), 709–734.

Maznevski, M.L., Chudoba, K.M., 2000. Bridging space over time: global virtual team dynamics and

effectiveness. Organization Science 11 (5), 473–492.

McAllister, D.J., 1995. Affect- and cognition-based trust as foundations for interpersonal

cooperation in organizations. Academy of Management Journal 38 (1), 24–59.

P. Kanawattanachai, Y. Yoo / Journal of Strategic Information Systems 11 (2002) 187–213212

Page 27: Dynamic nature of trust in virtual teams

McGrath, J.E., 1984. Groups: Interaction and Performance, Prentice-Hall, Englewood Cliffs, NJ.

McKnight, D.H., Chervany, N.L., 2000. What is Trust? A Conceptual Analysis and an

Interdisciplinary Model, AIS 2000, California,.

McKnight, D.H., Cummings, L.L., Chervany, N.L., 1998. Initial trust formation in new

organizational relationships. Academy of Management Review 23 (3), 473–490.

Meyerson, D., Weick, K.E., Kramer, R.M., 1996. Swift trust and temporary groups. In: Kramer,

R.M., Tyler, T.R. (Eds.), Trust in Organizations: Frontiers of theory and Research, Sage,

Thousand Oaks, CA, pp. 166–195.

O’Hara-Devereaux, M., Johansen, R., 1994. Global work: Bridging Distance, Culture, and Time,

Jossey-Bass, San Francisco, CA.

Parks, C., Henager, R., Scamahorn, S., 1996. Trust and reactions to messages of intent in social

dilemmas. Journal of Conflict Resolution 40 (1), 134–151.

Pearce, J.L., Sommer, S.M., Morris, A., Frideger, M., 1992. A configurational approach to

interpersonal relations: profiles of workplace social relations and task interdependence. Working

paper #OB092015, Graduate School of Management, University of California at Irvine.

Rempel, J.K., Holmes, J.G., Zanna, M.P., 1985. Trust in close relationships. Journal of Personality

and Social Psychology 49 (1), 95–112.

Roberts, K.H., O’Reilly, C.A., 1974. Failures in upward communication in organizations: three

possible culprits. Academy of Management Journal 17, 205–215.

Rotter, J.B., 1980. Interpersonal trust, trustworthiness, and gullibility. American Psychologist 35 (1),

1–7.

Schlenker, B., Helm, B., Tedeschi, J., 1973. The effects of personality and situational variables on

behavioral trust. Journal of Personality and Social Psychology 25, 419–427.

Smith, J.B., Barclay, D., 1997. The effects of organizational differences and trust on the effectiveness

of selling partner relationships. Journal of Marketing 61 (1), 3–21.

Sorrentino, R.M., Holmes, J.G., Hanna, S.E., Sharp, A., 1995. Uncertainty orientation and trust in

close relationships: individual differences in cognitive styles. Journal of Personality and Social

Psychology 68 (2), 314–327.

Sproull, L., Kiesler, S., 1986. Reducing social context cues: electronic mail in organizational

communication. Management Science 32 (11), 1492–1512.

Stevens, J., 1996. Applied Multivariate Statistics for the Social Sciences, 3rd ed., Lawrence Erlbaum

Associates, Mahwah, NJ.

Walther, J.B., 1995. Relational aspects of computer-mediated communication: experimental

observations over time. Organization Science 6 (2), 186–203.

Zand, D.E., 1972. Trust and managerial problem solving. Administrative Science Quarterly 17 (2),

229–239.

Zimmer, T., 1972. The impact of Watergate on the public’s trust in people and confidence in the

mass media. Social Science Quarterly 59, 743–751.

P. Kanawattanachai, Y. Yoo / Journal of Strategic Information Systems 11 (2002) 187–213 213