AN ANALYSIS OF A SUPERVISOR-SUBORDINATE TRUST RELATIONSHIP

In view of the importance of interpersonal tru~t a~ recognized by organiutional schola rs and the problems ~S50dated with the study of trun in organiutions. the developmCl1t of a coneeplu~l model of organiution~l trust is essentiaL The aim of this sllldy was to establish empirically the validity of a theoretically sound model of trun in the South Afrian work comeXI. The ovenll results confirmed a positive relationship between intCfpersonal trust, trustwonhiness and successfullruSl relationships.. The propensity to trust, as well as the length of the supcrvisorsubordinue rel~tionship, however. did not prove to hne ~ moderating effect on trusl .... ,onhincss.


ABSTRAC T
In view of the importance of interpersonal tru~t a~ recognized by organiutional schola rs and the problems ~S50dated with the study of trun in organiutions.th e developmCl1t of a coneeplu~l model of orga niution~l trust is essentiaL The aim of this sllldy was to establish empirically the validity of a theoretically sound model of trun in the South Afrian work comeXI.The ovenll results confirmed a positive relationship between intCfpersonal trust, trustwonhiness and successfullruSl relationships..The propensity to trust, as well as the length of the supcrvisorsubordinue rel~tionship, however.did not prove to hne ~ moderating effect on trusl .... ,onhincss.
Recent developments in organizatio nal psychology reflect the need fo r. and impo rtance of, interpersonal trust relationships in promoting individual welfare and org:mizational effectiveness (Clark & Payne, 1997;Kreitner & Kinicki, 1998).In the curren!South African cO n!ext.the issue of mutual trust in working relatio nships is becoming increasingly vi tal to business success (Martins, Watkins, Von der Ohe & De Beer, 1997).
The socio-politi cal situation created a social environment characterized by mistrust among the diverse groups in South Africa, Increased work force d iversity neccssitates that people from ve ry different backgrounds make contact and work together closely.A d iverse work force relies w ith greater d ifficulty o n interpersonal similarity, experience and common backg round to contri bute to m utual attraction and to e nhance the willingness to work together.In this regard, the devdopment of mutual trust provides oue mechan ism for enabling employees to work together more effectivdy (Mayer et al. 1995).Continuing changes in the workplace towards m o re participative manage ment styles and the implementation of self-directed work teams, increase the importance of trust.as cont rol mechanisms are reduced and interaction increases.
In spite of the growing importance of trust in organizations, a diminishing level of interpersonal trust is observed in many companies especially between managers and subordi nates (Martins et al., 1997;Sitkin & Ro th, 1993).Mutual trust is essential for effective communication between supervisors and their subordinates (Blackburn, 1992).Lack of such mutual trust ineviubly results in anxiety.suspicion, unceminty, low monle, low commitment and low job satisfaction (Mishra & Mo rissey. 1990).
Although the importance of interpersonal truSt has been acknowledged, li ttle attentio n has been give n in the literature to the s}'1tematic and theoretical analysis of how trust develops and functions in organizations.The stud y of interpersonal trust in organizations has rem ained problematic, mainly concerning the defi nitio n of trust itself; lack of clarity in the relatiomhip between risk and trust: con fusio n between trust and its antecedents and outcomes; and a failure to consider both the trusting party and the party to be trusted.In o rder to address these iss ues, and to clarify the relationship bet ween a specific trustor and trustee, M aye r et al. (1995) proposed a d yadic model of trust in organizations.Form the literature it appears that this model is supported by previous researeh in the field of in-~1S Ji:n f'tpri"lS slomJd buddrwttJ ,<>: A .s. Engtlbort/u, iAp<>rtnw:nl of l ndustrW l'Jrr""'''fJl /.htivmlry of SIt:II~k.

24
terpersonal trust (Clark & Payne, 1997).However, the validity of thIs model has not been !CSled empirically, which indiCites an important need for furth er research.
The specific purpose of this stud y WaS to test the va lidity of M ayer et ats (1995) model of organizational trust in the South Afri can CO ntext.The present research focussed o n anal yzing the specific nature of the relationships among interpersonal trust, propcnsity to t rust, the factors of trustwo rthiness and a successful trust relationship.The general aim was to get a better understanding of the d ynamics and performance implications of a superviso r-su bordinate truSt relationship.
Conce ptual model ofinterpef' So nal trust Although a g rowing number of researchers in social psychology analyzc trust in fri endship and family relationships, the natu re and basis of trust in such relationships may differ from that in organizatio ns.Mayer et al. (1995) developed a model (see Figure 1) of d yadic trust which focuses on trust in an orga• nizatiOnal enviro nment involving two specific panics, namely the trusting part y (trustor) and the person to be trusted (n us.tc:c).The model includes factors relating to the trustor (propensity to trust) and the trustee (perceived trustwo rt hiness), that were neglected by previous models.They assert that a lack of clear differentiation among factors that contribute to trust, trust itself, and outcomes of t ruSt hindered previous research on trust.According to them all three these facets must be measured in o rd er to test the validit y of their model.

•
Figure 1; Integrated m odel of interpenonal t rust (M ayer et aI., 1995, p.715) Some individuals have a higher general willingness or inherent propensity to trust others.Propensity to trust is proposed to be a suble personality trait that will affect the likelihood that the person will trust others (Clark & Payne, 1997;Mayer et ai., 1995;Rotter, 1%7).
Even though propensity would contribute to the explanation of some of the variance in trust.a given trustor exhibits different levels oftrust with regard to different trustees.To address this variance.the characteristics of the trustee need to be examined.Although a number of factors that lead to trust have been proposed.three chlTacteristics of a trustee are frequently mentioned in the literature.namely abi lity.benevolence.l nd integrity (Butler, 1991;Mayer et al.. 1995).Ability refers to a group of skills.competencies and characteristics thlt enable .11person to exert influence within a specific domain (M ayer et ll.. 1995; Sitkin & Roth.1993).Butler (1991), Mishra (19%) lnd Clark and Payne (1997) refer to a simil:.trconstruct.namely competence.
Benevolence refers to the extent to which a trustee is believed to act in good faith towards the trustor.without an egocentric profit motive (Mayer et aI., 1995).If the trustor perceives that the trustee falls short of any of Butler's (1991)  According to Butler (199 1) and Maye r et ll.(1995).the implication of integrity for interperson al trust involves the trustor's perception that the trustee adheres to principles thlt the trustor finds acceptable.Integrity is characterised by honesty, truthfulness.sincerit y and promise fulfilment (Cbrk and Payne.1997).Butler's (1991) factors of consistency.discreteness and fairness may also be regarded as faceu of integrity.
Each of these thItt determinants (ability.benevolence lnd integrity) captures some unique elements of trustworthiness (Mayer et al.. 1995).If the trustee is perceived as high on all three factors.the trustee will be perceived as quite trustworthy (Mishra. 1996).Thus. it can be postulated that: H ypothesis I: There exist a significantly positive relationship betwccn interpersonal trust lnd the following factors oftrustworthincss: a) ability b) benevolence c) integrit y Cabarro (1978) argues that the research interviews he conducted indicate that, in sequence of importance.integrity, competence and consistency were the moS! salient conditions of a supervisors downward trust in a subordinate.Whilst integrity, motives (loyaltyfrespect) and openness were the most important conditions ofa subordinate's upward trust in a supervisor.Butler andCantrell (1984) tested Gabarro's (1978) rankings of the determinants of trust and listed.from the strongest to the weakest.competence.integrity, consistency.loyalty and openness.for both upward and downward trust.Likewise., Schindler and Thomas (\993) concluded that the relative importance of these five conditions of trust were the same.regardless of the hierarchical relationships between two individuals.
According to Mayer et al. (1995), trustworthiness should be thought of as a continuum along which each of the factors can vary.rather than perceiving the trustee as absolutely trustworthy or not.Ability, benevolence and integrity are important for trust and each may vlry independently from the other, al though still related.Hence. it can be postulated that: Hypothesis 2: The factors of trustworthiness influence one another's effect on interpersonal trust reciprocally.
Ap~rt from propensity to trust ~ffecting interpersonal tru.<t whe n there is no information available o n the characteristics of the trustee, propensity can enhance the effect of trust-worthiness, thereby producing a moderating effect on trust (Mayer ct 011..1995).Therefore, it can be postulated that: Hypothesis 3: Propensity to trust has a moderating effect on the relationship between the factors of trustworthiness and interpersonal trust.
The Mayer et aL model (1995) can explain interpersonal trust (based on propensity to trust) bcfore any relationshi p bet wecn two persons has developed.When a relationship begins to develop, thc trustor may be able to obtain information on the trustee's charactcristics through third-party sources :and observation.Integrity, for instance.will be important to the formation of truSt early in the relationship.because little in formation is avaibble about the trustee's hcnevolence toward the trustor.
As the rebtionship develops.interaction with the rrustee enables the trustor to gain insight into the trustee's ben evolence, and the relative implct of benevolence on trust will grow.The development of the interpersonal relationship is therefore likely to modify the relative importance of the deterrninlnts of trustworthiness.Consequently.it can be postulated that: Hypothesis 4: The effect of integrity on illlerper:ional truSt will be most sllient early in the supervisor-subordinate relationship.prior to acquiring meaningful information on benevolence.
Hypothesis 5: The effect of perceived benevolence on interpersonal trust will increase as the relationship between supervisors and subordinates develops over time.
According to various researchers (Blackburn, 1992;Butler. 1991;Gabarro, 1978;Kreitner& Kinicki.1998)The Conditio ns of T rust Inventory established by Buder (1991) was used to measure trustworthiness and interpersonal Irust.The item s for the Conditions of Trust Inventory were sdected through a nnge of confirmatory fact or analyses and the factor pattern supported the content <l nd construct validity of the melsure (Butler. 1991).
A questionnai re consisting of eight items was developed for measuring a successful trust relationship between a specifi c subordinate and his/her supervisor.The behaviours identified in the literature as the strongest indication of thc outcomes of a successful trust relationship were included as item s (Cloete, 1998).A six-point Likert scale was used as a response fo rmal.
Both the Interpersonal Trust Scale and the Conditions of Trust Inventory were adapted for the purposes of this study.On the one hand.some items were rewritten to ensure that the meaning was more accurate within the South African cultural context.O n the other hand.a six-point Likert sole response format replaced the original five-point scale to counteract the central respo nse tendency (Kerlinger, 1986).
The study found that the adapted Conditions of Trust Inventory (0-= 0.80 to 0.93) and the Successful Trust Relationship Scale (0 = 0.95) showed high levels of internal consistency, and the adapted Interpersonal Trust Scale (a = 0.76) showed a satisfactory level of reliability.The subscale of the adapted Conditions of Trust Inventory used for m easuring interpe rsonal trust produced a high coefficient alpha of 0.93.

T he relatio nship between interpersonal t rWit and the factors of trwtworthiness
The Pearson product-moment correlation coefficients (Kerlinger. 1986) between intefJxrsonal trust and the factors of trustworthiness were determined.High and significant (p • 0.01), positive relationships exist between interpersonal trust and each of the fact o rs of trustwo rthiness (ability: r = 0.77; benevolence: r = 0.88; integrity: r = 0.92).Thus, H ypothesis 1 was confirmed.
Standard multiple regression (Tabachnick & Fidell, 1989) was performed between interpersonal trun as the dependent variable and ability, benevolence and integrit y as independent variables, to determine the relative importance of each of the factors of trustworthiness.
Only the regression coefficients ofintcgrity and benevolence differed significantly from zero. with F-values of70,76 (p .0,(01) and 9.ff1 {p I O,OS~ respcaively.Thus, only two of the independent variables contributed significantly to the prediction of the trust a subordinate would have in hisfhcr supervisor, namely benevolena: (sr 2 = 0,01) and integrity (sr2 = 0,08).Integ rity is the most important determinant regarding a subordinate's upward trust in a supervisor, with benevolence in second place.Ability, integrity and benevolence in combination explained 86% of the variance in trust.Although the correhltio n between interpernmal trust and lbility was 0,77 (p • 0,01), ability did not contribute signifiantly to regression.The trust l subordinate hOlS in his/her supervisor an therefOr(" be predicted by only taking into ilCCO\lnt the scor("s gained on perceived integrity md benevolence.Appar("ntl y the r("btionship between interpersonll trust lnd lbility is I n indirect result of the rdltio nships between benevolence, integrity and truSt. The interrelationship of the fac ton of trustworthiness Hierarchial regr ("ssion (Tlblchnick & Fidell, 1989) wu e mployed to determine if ability, benevolence and integrity hlve l mutull effect o n elch other's effect on interpenonll trust.
Furthermore, the aim was to determine iflbility, benevolence lnd integrity have an ldditional effect on interpersonal trust besides the inten.ction-effect of the three flCtors o f trustworthiness o n trust.
In step one, the interaction between the three flcwn of trustworthiness were PUI in the cqul tion.Secondly, lbility.benevolence and integrity were ldded to the equltion in o:ne step.The multiple regression coefficient (R ) wu significantl y (p , 0,001) different from zero after elch step.After step t wo. with all the vlril bles in the equation, R = 0,93.F (4; 126) = 198.p ' 0.001 .

T h e moderating effect of propensity t o trust o n the rela t ion ship between the factors of trustworthiness and interpersonal trust
The results of the stl ndud multiple regression anllysis (Tablchnick & Fidell, 1989) between interpersonal trust as dependent variable.and ability and ability-propensity to truSt intenction as independent vuiables, showed that the lbilitypropensity to trust intenctio n was superfluous in t he relationship between ability and interpersonal trust.Only the regression coefficient of lbility differed significantly from zero, F (2; 128) = 34,34; p .0,001.The lbility-propensity to trust interaction did not contribute significlntly to the prediction of trust.
Simi larly, it was found that the bene vole nce-propensity to trust interlction lnd the integrity-propensity to trust intenction did not contribute significlntly to the prediction of trust.Consequently, H ypothesis 3 WlS not supported.
The moderating effect ofthe length of the supervisor-subordinate working relationship on the relatiomhi p be tween integrity and trust The aim was to test the proposition that the impact of integrity o n trust is moderated by the length of supervisor-subordinl te relationship; that is, the longer the wo rking relationship, the lower the impact of integrity.If the length of working relationship (short, medium, long) is represented by two dumm y variables (D1 l nd 02). the research hypothesis implies the following regression m odel: E(Y I X I; D2)=o+f31 XI +.82 01 +f33 0 2+f3401 +f35 0 2' XI where E = expected vllue Xl = integrity 0\ = length of working relltionship from medium to long O 2 = length of working relationship from short to long .8401• X t = interaction between IJl lnd integrity f3 S 0 2 • X \ = interaction between 02 and integrity By using hierarchical multiple regression lnllysis, it was found that the ldding of O J, O 2 , 0 \ • X\lnd O 2 • Xl in l regression equl tion in which integrity alreldy lppeued, did not lead to a significant incrcOlSe in R 2. ThUs, it seems as if the length of the working relationship did nOt moderate the impact of integrity on trust.H ypothesis 4 Wl S therefore not supponcd.
The moderating effeet of the length of the supervisor-subordinate working relatioruhip on t he relatiomhip between benevolenee and trwt The lim WlS to test the proposition that the impact ofbenevolence on trust is moderated by the length of supervisor-subordinate rebtionship; thlt is, the longer the working relationship, the stronger the impact of benevolence.The procedurc followed for the testi ng of Hypothesis 5 WlS the same lS for Hypothesis 4. It was found that the adding of 0 1.02> 0 1 • X I and O 2 ' Xl in a regression equation in which benevolence already appeared, did not lead to a significant increase in It 2. Hence, it seems as if the length of the working relationship did not moderate the impact of benevolence on trust.H ypothesis 5 was consequently rejected.
The relationship between interpersonal trust and a successful trust relationship A high Pearso n product-moment correlation coefficient between interpersonal trust and a successful trust relationship was found (r = 0,85; p , 0,01).It appears that a strongly positive relationship exists bel\veen interpersonal trust and a successful (fUSt relationship.Accordingly, Hypothesis 6 \VOl! supported.

DISCUSSION
A significantly positive relationship was found between interpersonal trust and the factors of perceived trustworthiness.The results indicate that the truSt of a subordinate in his/her sup~visor is dependent on the supervisor's trustworthiness as perceived by the subordinate.Given these results.organizations should pay more attention to enhance the ability, benevolence and integrity of employees in their efforts to increase interpersonal trust between employ~s.
These fi ndings regarding the relationship between interpersonal trust and the factors of trustworthiness were supported by the findings o( several researchers (Butler. 1991;Butler & Cantrell. 1984: Clark & Payne. 1997;Mishra, 1996).Moreover, these results are in congruence with Mayer et at's (1995) proposition stemming (rom their model ofinterpcrsonal trust. It was further found that the sequence of importance of the fa ctors o( trustworthiness (or the subordinate's upward trust in his/her manager, was integrity and then benevolence.This result rega rding the sequence of importance o( the factors of trustworthiness is similar to the findin g of Gabarro (1978).h .however.differs from the results obtained by Butler and Cantrell (1984), and Schindler and Thomas (1993).
It appears that integrity and benevolence arc predominant in determini ng the manager's trustworthiness as p~ceived by the subordinate, and that ability does not playa major role.Mayer et "ts (1995) model docs not specify the sequence ofimpon ance of the factors of trustworthiness, although they contend that each of these factors portrays some unique characteristics of trustworthiness.Hence, it seems that the resuhs concerning the role that ability plays in determining the manager's trustworthiness as experienced by the subordinate, is contrary [Q Mayer et ats (1995) contention.
The limitations of significance tests must be kept in mind as a possible explanation of the lack of v"riance in interpersonal trust through ability.According to Tabachnick and Fidell (1989) ability may appear insignificam beause it shares variance with integrity and benevolence, although the three variables in combination are m"inly responsible (or the size of R 2, Because of this, the correlation coefficient between ability and trust must be interpreted additionally to the regression coefficient and the F-value of ability.The significant bivariate correlation.however, can lead to stronger conclusions th"n is justified if the relationship between ability alone and trUSt is measured. Furthermore, it must be kept in mind that th is study examined interpersonal trust from one position only, namely from the subordinate upward to the supervisor.Consequently, it is possible that the importanCe of ability as de terminant of trust may increase when truSt is examined from the position of the supervisor downward to the subordinate.Then it would be in accordance with Gabarro's (1978) finding that the sequence of importance for a manager's downward trust in a subordinate differs from that for the upward trust of a subordinate in a manager.
h was found that the factors of trustworthiness, namely ability, benevolence and integrity, have a reciprocal effect on each factor's effect on interperson al trust.These findings suggest that for instance, benevolence alone is not enough to lead to trust if the manager does not display hi gh ability and integrity.These findin gs arc similar to Mishra's (1996) andMayer et al.'s (1995) proposition that ability, integrit y and benevolence are rebted, and that it is possible for a perceived lack of any of the three factors to undermine trust.
Based on the resuhs, it can be concluded that propensity to trust does not have a moderati ng effect on the relationship between the fa ctors of trustworthiness and interpersonal trust.This finding is in contrast to Maye r et al. 's (1995) assumption in their model o(interperson al trust.Butler (1991).and Clark and Payne (1997).however, found that trust in a specific individual is mo re relevant in the prediction of work Outcomes than the general trust in others (i.e.propensity to trust), which lends some support to the results obtained.
It was found that the length of the supervisor-subordinate relationship, hence the development of the truSt relationship over time, did not moderate the relative impact of integrity or benevolence on trust.This finding consequently indicates that the relative impact of integrity and benevolence on trust remain constant as the interpersonal trust relationship between the manager and subordinate develops through interaction over time.This findi ng is in contrast with the asse rtion by Mayer et al. (1995) that the effect of perceived integrity on trust will decrease over time, and the effect of perceived benevolence will increase as the trust relationsh ip between supervisors and subordinates develops.
The significantly positive relationship that \VOIS found between interpersonal trust and a successful trust relationship was supported by the studies of Cangemi et al. (\989), Mi shra and Morrisey (1990), and Blackburn (1992).This findin g implies that the outcomes of high interpersonal trust between the subordinate and his/her supervisor can be open communication, empathy, autonomy.cooperation.participatory decision making, fairn ess, mutual respect.and the creation of opportunities for the utilization of the subordi n"te's potential.
Since th is study regarded interpersonal trust from one position only, it is recommended that future studies should analyze trust (rom both positions, hence also from the position of the manager downward to the subordinate, to get better insight into the development of mutual trust between the two specific panies. Further development and operationalization of Mayer et al.'s model of organizational uust would benefit the study of organizational behaviour and performance.To accompl ish a high level of construct validity, new instruments should be developed to measure the perceptions of the factors oftrustworthiness, interpersonal trust, and propensity to trust in accordance to the definitions provided by Mayer et al. (1995).Since risk taking is an essenti,,1 compOnent of organizational trust (Mayer et aI., 1995), it should also be accurately measured and included in future validation of the model.In addition to model-specific hypotheses, it is also important [Q further investigate the process by which trust develops.
In conclusion. it is clear from this study that trust is an essential element of interpersonal relatio nships "nd effective management.It is thereforc in the interest of organizations to proactively build the level of interpersonal truSt between supervisors and subordinates.
w~s used to measure propensity to trust.Tile validity of the Interpersonal Trust Scale has been tested and the results ~~~~':':~ ~~~rt:l.b~econstruct and discriminant validity for the scale (Rotter.1967).
Blackburn found that specific forms of behaviou r. namely participltion in decision making.open and clear communication.honesty.and empathy.had a positive effect on trust in a supervisor-subordinate relltionship.The results of Mishra and Morrissey's stud y supported the main advantages of a successful tru~t rel:.ttionworking at various organizations in the Western Clpe were selected as S:.l.mplc.Purposive sampling (Kerlinger.1986)ensured thlt the sample was repres(:ntative regarding gender (47% female), age (varied from 18 to 48 yeus).work experience (varied from one to 31 years), and length of working relationship with supervisor (50% more than one year).
Blackburn (1992)rust relationship can lead to greater co-operation and support of organizational goals.anincrease in productivity, th.:: advancement of the developmental potenti,,) of subordinltes.opencommunicationandhonesty.Mutual trUSt between supervisors and subordinates is consistent with Graen's(Butler.l99I)venicaldyadlinkagemodel.Trust is an essential component in the reciprocal rein forcement in high-quality leader-member exchanges, promoting satisfaction and productivity over time.Tnnsformational leadership is also characterized by mutua!trust.Accordi:ng to Bass & Avolio (1994) ..11 transformational leader supports his/her followers and is trusted and respected by them.A mlllsformational leader becomes effective through empowering hisfhcr fol1owers by giving them autonomy.facilitatingthc!ir self-development, and serving as a role model of integrity and equity.Blackburn (1992)conducted a study in South Africa that built on the study ofMishn and Morrissey (1990).instrumenu The Interpersonal.Trust Scale developed by Rotter (1%7)