THE DIMENSIONALITY OF WORKPLACE INTERPERSONAL TRUST AND ITS RELATIONSHIP TO WORKPLACE AFFECTIVE COMMITMENT 43

Trust is fundamental to the existence of human relationships, including the workplace relationship between employees and their co-workers and supervisors. This paper presents the results of a study that investigated the nature and dimensionality of interpersonal trust at work, specifically trust in co-workers and trust in supervisor. Survey questionnaire responses from 278 employees in four South African organisations were analysed. The research findings evidence that interpersonal trust is a multi-foci construct that is differentially related to corresponding foci of affective commitment.

Trust is a psychological state crucial to the formation and sustenance of human relationships (Clarke & Payne, 1997;Kramer, 1999;Mayer, Davis & Schoorman, 1995) and its importance in the workplace has been increasingly recognised over the past 50 years (Butler, 1991;McAllister, 1995, Rotter, 1967).Recent research has demonstrated its practical significance in the determination of citizenship behaviour (Deluga, 1994;Gambetta, 1988), commitment (Bussing 2002;Cook & Wall, 1980), and job satisfaction (Bhattacharya & Devinney, 1998).Trust has also been implicated in the successful management of change (Drucker, 1999;Harvey & Brown, 2000), teamwork (Bews & Martins, 2002) and diversity (Bussing, 2002) initiatives.Colquitt, Scott, & LePine (2007) in discussing the results of their meta-analysis of 132 trust studies concluded that "trust is a vital component of effective working relationships" (p.918).This paper is an attempt to extend the interpersonal trust construct by examining multiple foci of interpersonal trust and examining their relationship to multiple foci of affective commitment.

Trust in the workplace
Globalisation has intensified competition, requiring organisations to seek greater operational efficiencies.Traditional control cultures expressed in closed bureaucratic systems with rigid management hierarchies and Fordist work routines have long been regarded as inhibiting competitiveness and employee creativity, prompting organisations to abandon their obsession with control and attempt to foster high-commitment workplaces (Walton, 1985).Strategies for fostering commitment typically include human resource initiatives in which employees are encouraged to adopt a unitarist perspective and align their personal goals with those of the organisation (Guest, 1995).These initiatives typically incorporate high-trust techniques (e.g.semi-autonomous work teams) that require employees to exercise discretion and assume responsibility for their efforts (Storey, 1995).The importance of interpersonal trust for team performance has been noted (Langfred, 2007).Understanding the nature of trust between co-workers and trust in immediate supervisors therefore becomes important for organisations seeking to create high-commitment, competitive workplaces (McAllister, 1995).
A single widely accepted definition of trust has yet to emerge in the literature (Bews & Martins, 2002;Dirks & Ferrin, 2002;Kramer, 1999;Rousseau, Sitkin, Burt & Camerer, 1998).In economics, trust has been construed as a rational, calculated choice (Fukuyama, 1995;Gambetta, 1988) and in sociology trust is regarded as an institutional phenomenon (Gambetta, 1988;Sztompka, 1999).In psychology trust is treated as an individual predisposition (Farris, Senner, & Butterfield, 1973;Mayer et al., 1995;Rotter, 1967), a predisposition shaped by both experiences and cultural socialisation processes (Clarke & Payne, 1997), an expectancy that another can be relied on (Rotter, 1967), or a "complex set of interpersonal and environmental relationships" (Farris et al., 1974, p. 144).Integrating these approaches, trust can be defined as a psychological state that involves a decisionmaking process, affected by individual attitudes and cognitions, about an individual's willingness to accept vulnerability to another based on positive expectations of his or her actions in the future (Butler, 1991;Clarke & Payne, 1997;Mayer et al., 1995;McAllister, 1995;Scott, 1980).

Multiple foci of interpersonal trust at work
Within organisations, there are at least two possible foci for the interpersonal trust of individual employees: their supervisor and their co-workers.The dyadic trust relationship between supervisor and subordinate (Butler & Cantrell, 1984;Deluga, 1994) has enjoyed more research attention than trust in coworkers (Den Hartog, Shippers, & Koopman, 2002).
Trust in supervisor is an interpersonal form of trust (Mayer et al., 1995) that emerges from an employee's perceptions regarding the supervisor's benevolence, integrity, ability, openness to share information, and consistency of behaviour (Butler & Cantrell, 1984;Clarke & Payne, 1997;Mayer et al., 1995, Schindler & Thomas, 1993;Tan & Tan, 2000).Benevolence is the extent to which the supervisor wants to do good for the subordinate, integrity is the extent to which the supervisor's actions are acceptable to the subordinate (as fair, honest and just) and ability refers to the supervisor's technical job related competence.Openness refers to the willingness of the supervisor to share ideas and information and consistency is the reliability and predictability of the supervisor's behaviour.There are contradictory positions regarding whether these five variables are antecedents (Clarke & Payne, 1997;Schindler & Thomas, 1993) or components of organisational trust (Nyhan & Marlowe, 1997).The widely used Organisational Trust Inventory assumes that they are components of trust (Nyhan & Marlowe, 1997).Several variables such as leadership style (Costigan, Ilter & Berman, 1998;Pillai, Schiesheim & Williams, 1999), personal characteristics (Kramer, 1999), relationship history (Kramer, 1999), and individual predisposition to trust (Clarke & Payne, 1997;Mayer at al., 1995) have been identified as antecedents of interpersonal trust in supervisor.& Travaglione, 2004).Nevertheless, there is research suggesting an overall relationship between trust and affective commitment based on social exchange theory, which suggests that processes of reciprocation at work will foster interpersonal relationships at work (see Ferres et al, 2004).It is therefore proposed that if a coworker interacts with benevolence, integrity, consistency, ability, and openness towards a colleague then the relationship between them is likely to be strengthened and maintained.
Research results concerning the relationship between trust in supervisor (vertical interpersonal trust) and trust in coworkers (lateral interpersonal trust) are confusing.Some research evidences construct redundancy and proposes that no distinction been made between these two foci of trust at work (e.g.Schindler and Thomas, 1993).Other research has evidenced that trust in supervisor and trust in co-workers have different antecedents, which indicates that they may be distinct constructs (Costigan et al., 1998;Kramer, 1999, Mayer et al., 1995;Pillai et al., 1999).The major source of confusion in the literature is the contradictory and poor definition of terms in various studies.For example, the term "management" has been used in certain survey items as a proxy for the employee's immediate supervisor (e.g., Flaherty & Pappas, 2002) and in other studies as a proxy for the organisation as an entity (e.g., Pillai et al., 1999).
Trust in the organisation is not an interpersonal form of trust but rather a systems form of trust (McCauley & Kuhnert, 1992;Tan & Tan, 2000) that derives from structures and processes within the organisation such as fairness and perceived organisational support (Barling & Philips, 1993;Folger & Konovsky, 1989;Liden, Wayne & Stilwel, 1993).This study focuses on interpersonal forms of trust and therefore does not consider trust in the organisation.
To assess the benefits of promoting interpersonal trust within organisations, it is instructive to examine the existing literature on the potential effects of trust on organisationally salient work attitudes.Tan and Tan (2000) suggested that trust influences affective commitment.Their suggestion echoed the earlier calls of Nyhan and Marlowe (1997) and Meyer andAllen (1994, 1997) for further research concerning the relationship between organisational trust and affective commitment.Affective commitment refers to the emotional attachment, identification, and sense of involvement that an individual feels toward a specific entity or focus of commitment such as an organisation, career, supervisor, co-worker, or job (Meyer & Allen, 1997).It represents the individuals desire to remain associated with the focus of their commitment and to expend energy for its benefit.Both commitment and trust have been implicated in increasing organisational effectiveness (Bussing, 2002).The relationship between the two constructs was rarely examined (Cook & Wall, 1980;Nyhan & Marlowe, 1997) but commitment is now often considered as a correlate or outcome of trust (Colquitt, Scott, & LePine, 2007).To date, no research has considered the relationship between multiple foci of affective commitment and multiple foci of interpersonal trust corresponding to each of these foci of affective commitment.

Research goals
The primary goal of this research is to advance understanding regarding the measurement of interpersonal trust at work by assessing its dimensionality amongst South African employees.A secondary goal of this study is to examine the relationship of different foci of interpersonal trust with their corresponding foci of affective commitment.

Research approach
A cross-sectional, self-report questionnaire survey design was used in this study.The data were analysed using quantitative statistical techniques.

Participants
The sample consisted of 278 white-collar administrative workers from four organisations in the Western Cape (N = 278).The sample was predominantly female (57%) and white (61%), most participants (95%) had completed high school and 40% had further tertiary education.The average age of participants in the sample was 36 years (range 19-63).Average tenure was 11 years (range .5-36).

Measuring instruments Interpersonal Trust
Eight items from the trust in supervisor scale that forms part of Nyhan and Marlowe's (1997) 12-item Organisational Trust Inventory (OTI) was used to measure trust in supervisor.Scale items were designed to reflect benevolence, integrity, ability, openness, and consistency (Nyhan & Marlowe, 1997).The OTI items on the trust in supervisor scale were modified to focus on trust in co-workers.This resulted in the development of a new 8-item scale consistent with the literature on the nature of interpersonal trust (Schindler & Thomas, 1993).A seven point Likert scale was used across all items.The response set consisted of confidence in the focus of trust from 1= "nearly 0" to 7 = "nearly 100%".

Affective Commitment
Affective commitment was measured using six items selected and adapted from Meyer and Allen (1997).Three items measured affective commitment to the employee's immediate supervisor (AC-SUP) and three items measured the employee's affective commitment to their co-workers (AC-CW).A five-point Likert scale was used across all items, with possible responses ranging from "strongly disagree" to "strongly agree".

Control variables
The gender, age, race, organisational tenure, educational level, and organisational affiliation of participants were measured to use as control variables in the hierarchical regression analyses.

Factor analysis
Factor analysis was used to assess the dimensionality of interpersonal trust and corresponding affective commitment scales in two ways.First, using principal components factor analysis, as per much of the trust literature, and second, using higher-order factor analysis to yield a more complete assessment of the dimensionality of the constructs (Thompson, 2004).Principal components factor analysis with varimax normalised rotation indicated that trust at work consists of two distinct factors (see Table 1).Principal Components analysis tends to increase the number of factors extracted (Thompson, 2004), but in this study principal factor analysis using the principal axis method with varimax normalised rotation produced the same number of factors.This is consistent with Velicer and Jackson (1990) who concluded that despite textbook recommendations to use principal factors "the choice of method is not a decision that will greatly affect empirical results or substantive conclusions" (p.21) especially for well designed data sets.Due to the propensity of past studies to report principal components analysis these will be reported here (several alternative analyses were conducted and are available from the authors).As expected, all the items relating to trust in supervisor loaded on one factor, named Trust in Supervisor (T-SUP) and all the items relating to trust in co-worker loaded on a second factor, named Trust in Co-worker (T-CW).The two factors explained over 75% of the variance amongst the items.Kaiser's criterion of including factors with eigenvalues exceeding 1 was adopted.The scree plot was also examined and it confirmed that two factors should be extracted.Following the same process, factor analysis on the affective commitment items confirmed that these items loaded on two factors, as expected.The two factors explained almost 68% of the variance amongst the items.It should be noted that one item, AC-SUP 1 loaded less strongly on the AC-SUP factor than the other AC-SUP items and cross-loaded on Factor 1.This result suggests that refinements to this item should be effected in future studies.For the purposes of this study, the item was retained as the factor loading was almost .6 and the magnitude of the difference between the two cross loadings exceeded .3(see Table 2).It should be noted that AC-SUP1 may have been deleted from the scale if a principal factors extraction method had been applied but that this would not have affected the dimensionality of commitment or any of the subsequent results reported in this paper (a full analysis is available from the authors).Note: N = 278 with casewise deletion of missing data.
Principal components factor analysis with varimax normalised rotation.AC-SUP 1 was retained despite moderate cross loading.
Factor loadings greater than 0,7 have been highlighted.
Rather than compute loadings for oblique factors that are often difficult to interpret and have a tendency to replicate less well and following recent recommendations regarding exploratory factor analysis (Thompson, 2004), higher-order factor analysis was applied to the extracted factors (Wherry, 1984).In this approach to factor analysis, clusters of items are identified, axes rotated through these clusters, and correlations between the (oblique) factors computed.The resulting correlation matrix of oblique factors is then further factor-analysed to yield a set of orthogonal factors that divide the variability in the items into that due to shared or common variance (higher-order secondary factors), and unique variance due to the clusters of similar variables (items) in the analysis (primary factors).Careful examination of the factor loadings in Table 3 suggests that (a) There is a general (secondary) interpersonal commitment factor that likely affects all types of commitment measured by the 6 items; and (b) There appear to be two primary, unique areas of interpersonal commitment that can best be described as interpersonal commitment to co-workers and interpersonal commitment to supervisors.Similarly, careful examination of the factor loadings in Table 4 suggests that (a) There is a general (secondary) interpersonal trust factor that likely affects all types of trust measured by the 6 items; and (b) There appear to be two primary unique areas of interpersonal trust that can best be described as interpersonal trust in co-workers and interpersonal trust in supervisors.
The hierarchical factor analyses yield a fuller understanding of the nature of affective commitment and interpersonal trust, showing that there are broader areas of generalisability than indicated by examining the correlations between primary factors.Even critics of principal components analysis with varimax rotation assert that higher-order factor analysis produces a more complete understanding of how the items relate to each other (Gorsuch, 1983).

Reliability analysis
Reliability was assessed by calculating the internal consistency of each scale using Cronbach's coefficient alpha.The Cronbach alpha for each scale well exceeded 0,7 (see Table 5), the criterion suggested by Nunnally (1978).Inter-item correlations between items within each scale were also high and exceeded 0,7 for each scale, exceeding the criterion of 0,3 suggested by Hair, Anderson, Tatham, and Black (1998).

Descriptive statistics
The means, standard deviations, skewness and kurtosis of each scale are presented in Table 5, which shows moderately high levels of trust across both interpersonal trust foci and moderately high levels of affective commitment across both foci.The standard deviations (ranging from 0,79 to 0,90) demonstrate that there was sufficiently high variation between scores in the sample to permit further analysis.The skewness and kurtosis of the distribution for each scale were within the guidelines suggested by Hair, Babin, Money, and Samouel (2003).
Post-hoc Scheffe tests were conducted to assess differences across the four participating organisations on all the trust and commitment variables but no significant differences across organisations were found on any of the variables examined.
Similarly, no differences were found on any of the variables across age or gender.
There were no significant differences between black and white participants on A-CW or T-CW.There was a significant difference between black and white participants on A-SUP, affective commitment to their immediate supervisor (p = 0,017) and T-SUP, trust in supervisor (p < 0,001).

Correlation analysis
Correlation analysis revealed significant relationships between and within the trust and affective commitment scales (See Table 5).A significant positive relationship (r = 0,474; p < 0,0001) between T-SUP and AC-SUP was found.A significant positive relationship (r = 0,252; p < 0,0001) was found between T-CW and AC-CW.These significant relationships suggested the usefulness of proceeding with regression analyses.Though statistically significant the correlation between T-CW and T-SUP (r = 0,426; p < 0,0001) was not sufficiently high as to suggest possible problems of multicollinearity in the regression analyses (Hair et al., 1998).

Regression Analysis
Hierarchical multiple regression analysis was performed to examine further the relationship between interpersonal trust and corresponding dimensions of affective commitment, while controlling for demographic variables.The demographic control variables were entered in the first step of the hierarchical regression analysis and the two interpersonal trust variables were added in the second step of the regression analysis.Two such analyses were conducted, first with AC-SUP as the dependent variable and then with AC-CW as the dependent variable.

At
Step 1 of the hierarchical regression analysis for AC-SUP, the six demographic variables were entered.They explained less than 5% of the variance in AC-SUP.Race was the only significant variable (beta = -0,119, p = 0,021) at the 0,05 level of significance.The overall model was not statistically significant.

At
Step 2 of the hierarchical regression analysis for AC-SUP, the two interpersonal trust variables were added to the model.Now, only T-SUP was a significant predictor of affective commitment to the supervisor (beta = 0,450, p < 0,0001).

DISCUSSION
The results of this study highlight two findings of interest.First, interpersonal trust at work has two clear dimensions: T-SUP and T-CW.Second, there is a significant relationship between interpersonal trust foci and corresponding affective commitment foci.Each of these findings will be discussed in turn.

Interpretation
This study confirms that interpersonal trust is a multidimensional construct and implies the need to draw on multiple disciplines to further our understanding of interpersonal trust (see Kramer, 1999;Lewicki & Bunker, 1996;Rousseau et al., 1998).Findings regarding the unidimensionality and independence of T-CW are particularly important, as this focus of trust has not enjoyed much research attention.
The results indicate that T-SUP has a statistically significant positive relationship with T-CW.That high levels of one focus of interpersonal trust were accompanied by high levels of the other focus of interpersonal trust in this study may indicate the importance of contextual factors that facilitate high levels of interpersonal trust.Future studies should therefore control for aspects of the corporate culture across participating organisations (see Kramer, 1999;McAllister, 1995;Whitener, Brodt, Kosgaard, & Werner, 1998), particularly as it is communicated during socialisation processes (Lewicki & Bunker, 1996).The high relationship between different foci of interpersonal trust may simply not be generalisable across organisations (Tyler & Kramer, 1996), as suggested in this study.Nevertheless, an important finding in this study is that interpersonal trust has a "g factor" that can be shown using higher-order factor analysis.
The findings in this study demonstrate an unexpected result.
Trust at work was found to be moderately high and the highest levels of trust were in supervisors.This surprising finding has possible implications for understanding trust in the South African context.Firstly, it demonstrates that the participants in this study tended to perceive their supervisors as demonstrating high levels of benevolence, integrity, ability, openness and consistency at work (Mayer et al., 1995;Schindler & Thomas, 1993, Tan & Tan, 2000).It also indicates that co-workers are demonstrating average to high levels of the same.Finally, it could indicate that employees in South African organisations have a high predisposition to trusting others (but this remains speculative given the limited sample in this study).Further research on the nature of trust would be required in order to determine the effect of personal characteristics on trust in South Africa.High levels of interpersonal trust were found across race groups but dyadic analyses were not conducted and further research is therefore necessary before any claims can be made in this regard.
The majority of previous research that has been conducted on the relationship between trust and affective commitment has demonstrated a relationship on an organisational level.Research concerning trust relationships on the interpersonal level has been largely ignored (Bussing, 2002;Cook & Wall, 1980;Nyhan & Marlowe, 1997).The results of this study showed a statistically significant and practically significant explanatory relationship between trust in supervisor and affective commitment to the supervisor.A statistically significant but moderately low explanatory relationship between trust in co-workers and affective commitment to co-workers was found in this study.
The paucity of research on understanding the nature of T-CW or developing appropriate measures of the construct could have influenced this finding (Schindler & Thomas, 1993).

Recommendations
There is little empirical research evidence for a direct causal relationship between trust and affective commitment (Nyhan & Marlowe, 1997) but the results in this study showed that trust in supervisor was significantly related to affective commitment in supervisor and explained significant variance in affective commitment in supervisor.Although research on the relationship between trust and affective commitment on a supervisory level is limited, this study supports previous research that demonstrated the strength of this relationship on an interpersonal level (Bussing, 2002;Nyhan & Marlowe, 1997;Schindler & Thomas, 1993).This has implications for supervisors given Meyer and Allen's (1991) research that evidenced how higher levels of affective commitment result in organisationally salient outcomes such as citizenship behaviours and willingness to stay in the organisation.For executives, this study suggests that fostering trust between co-workers may be an essential component of strategic change initiatives such as self-directed teams, which require high levels of collaboration to be effective.For human resource managers, it is probably self-evident that if managers attempt to foster interpersonal trust within an organisation this will positively affect the effectiveness of human resource practices by influencing perceptions of the fairness of these practices.Finally, the results suggest that the promotion of trust in supervisor may contribute to employee retention and productivity, through its effect on affective commitment to the supervisor.

Suggestions for future research
The findings presented in this paper advance our understanding of interpersonal trust at work by demonstrating its dimensionality and examining the relationship between foci of interpersonal trust and corresponding foci of affective commitment to the organisation.This study is, to our knowledge, the only study that has investigated these relationships.It is hoped that future studies will develop this line of research, perhaps with the application of longitudinal designs and the use of structural equation modelling so that our understanding of these relationships and the causal order between them can be further refined.The development of workplace relationships based on trust may be challenging, especially in the South African context where existing levels of trust are typically thought to be low, but this study does indicate that fostering interpersonal trust should become a managerial imperative as it results in valued organisational outcomes.

Table 1 FacTor analysis: inTerpersonal TrusT
Note: N = 278 with casewise deletion of missing data.Principal components factor analysis with varimax normalised rotation.T-SUP 1 removed due to cross loading.Factor loadings greater than 0,7 have been highlighted.