The production of trust during organizational change

Abstract This paper investigates the relationships between organizational change and trust in management. It is argued that organizational change represents a critical episode for the production and destruction of trust in management. Although trust in management is seen as a semi stable psychological state, changes in organizations make trust issues salient and organizational members attend to and process trust relevant information resulting in a reassessment of their trust in management. The direction and magnitude of change in trust is dependent on a set of change dimensions that reflect trust relevant experiences and information. We distinguish between dimensions related to trust relevant consequences of the change and trust relevant aspects of how the change process is performed. Empirical results indicate that increases in post change emotional stress and the use of referential accounts for justifying change are both negatively related to post change trust in management. The use of ideological accounts and participation were found to be positively related to post change trust in management, so was perceived decision quality. Findings also indicate that the effects of change on trust are negatively moderated by tenure.


Introduction
The purpose of this paper is to shed light on how organizational change affects trust in management. Several writers have underlined the importance of trust during organizational change, but their main focus has been on trust as a factor that facilitates change or that positively moderates the relationships between facets of the change process (e.g. participation and communication) and successful change (e.g. Kanter, 1977;Burt, 1992;Lane and Bachman, 1998). In this paper, we see organizational change as a critical trust building or trust destroying episode in a long term and ongoing relationship between the organization, represented by its management, and non managerial employees (Korsgaard et al., 2002). It is argued that, contingent on how change tasks are performed and on the consequences for those affected, organizational change can contribute positively or negatively to the evolution of trust in management. Depending on the strength and direction of the relationships between change facets and trust, the benefits of an organizational change may be further strengthened or off set by its associated shift in trust.
In light of trends toward globalization, restructuring and higher paces of change, trust in leaders is playing an increasing role in contemporary employment relationships. Trust has captured the interest of researchers and practising managers partly because trust has been seen as a key contributor to a number of outcomes associated with organizational success. In line with this, trust has been linked to better task performance (Oldham, 1975;Robinson, 1996), openness in communication and information sharing (e.g. Boss, 1978;Dirks, 1999), organizational citizenship behaviour (Konovsky and Pugh, 1994), less conflict (Porter and Lilly, 1996) even between partners in interorganizational relationships (Zaheer et al., 1997), and acceptance of decisions and goals (Oldham, 1975;Tyler and Degoey, 1996).
Theories of trust conceptualize trust as a state that depicts how focal individuals, groups or organizations (the trustor) view other individuals and social units (the trustee). Trust is seen as semi stable and based on the processing of numerous specific experiences with the trustee, often over long periods of time. This raises the question of to what degree can one single episode, such as the experience of an organizational change, affect the level of trust that employees have in their leaders. In this paper, we argue that the level of uncertainty and the vulnerability experienced by organizational members during change lead to an active processing of trust relevant information. Based on this information processing, organizational members reassess their trust in management. Consistent with Morgan and Zeffane (2003) we see organizational change as a critical event that may create or destroy trust in management.
In the next section, we review the literature on trust and identify a set of general factors that have been related to trust in previous work. In the following section, these general trust producing factors are tied to content and process aspects of organizational change, and a set of theory driven hypotheses are developed. In this section the focus is on a theoretical integration of concepts used for analysing change with concepts from the trust literature. Subsequent sections describe our research design, data and results. The paper closes with a discussion of key results of the study.

Theoretical Background-Evolution of Trust
Trust is an increasingly important issue in today's organizations. In an ever more chaotic world, distrust becomes the overarching attitude. The novel, unknown, or unfamiliar must be met by suspicion until it has been proven and found reliable. In a chaotic world, xenophobia is not to be considered pathological but a natural strategy for survival. In the small world, where family and friends exist, trust is by most people regarded as a relatively robust phenomenon. In a larger context, in the organization, it is more vulnerable. It takes a long time to build up and can be ruined in a minute.
Trust may be defined as a positive expectation that another will not-through words, actions, or decisions-act opportunistically (Boon and Holmes, 1991). It is also a history-dependent process based on relevant but limited samples of experience (Rotter, 1980). It is built up incrementally and accumulating over time. Most people experience difficulties in trusting total strangers that they meet for the first time. Trust is also a primary attribute associated with leadership.
It has been revealed that part of the leader's task has been, and continues to be, working with people to find and solve problems (Zand, 1997). Whether leaders gain access to the knowledge and creative thinking they need to solve problems depends largely on how much people trust them. Hence, trust and trustworthiness modulate the leader's access to knowledge and cooperation.
Managerial and leadership effectiveness is today, more than ever, depending on the ability to gain the trust of followers (Brockner et al., 1997). Vital threats towards trust may include reengineering, downsizing, and the increased use of temporary employees, all of which undermine a lot of employees' trust in management. Moreover, the opportunities for top management to develop relationships with employees, beyond those with whom they interact frequently, have become more and more scarce (Culbert & MacDonough, 1985).
As a result, employees monitor the organizational environment as a means of assessing whether to place trust in management or not. Therefore, organizational processes are sometimes believed to communicate top management's view of its employees and their roles. If the physical traits of an organization, such as structures, roles, and climate, communicate distrust in employees by top management, it is believed that employees will respond with distrust.
Since trust is such an important feature of leadership, it is not surprising that it has become a key concept in several leadership theories, as noted by Dirks and Ferrin (2002). Transformational and charismatic leaders build trust in their followers (Kirkpatrick and Locke, 1996;Podsakoff et al., 1990). Leaders are also in pursuit of perceivable attributes that promote trust and that may be important to leadership effectiveness (Bass, 1990;Hogan et al., 1994). Furthermore, trust is considered to be an important element of leader-member exchange theory . It has also been noted in leadership theories that trust may be based on deterrence, knowledge or identification (Lewicki and Bunker, 1996;Child, 2002). Hence, trust may be founded on: (i) fear of reprisal if the trust is violated, (ii) behavioural predictability that comes from a history of interaction, or (iii) mutual understanding of each other's intentions and appreciation of the other's wants and desires. Dirks and Ferrin (2002) noted that trust may have different foundations in that they notice that two qualitatively different theoretical perspectives on how trust in leadership may be distinguished. The first perspective focuses on the leaderfollower relationship, especially with regard to how the follower perceives the relationship. In these theories, concepts like trust, goodwill, and the perception of mutual obligation become central from the perspective of social exchange (Blau, 1964;Konovsky and Pugh, 1994;Whitener et al., 1998;Pillai et al., 1999;Schriesheim et al., 1999). Since the emphasis is on relational issues, this perspective has been referred to as the relation-based perspective.
The second perspective centres on to what extent a follower experiences vulnerability in a hierarchical relationship based on the perception of the leader's character (Mayer et al., 1995;Cunningham and MacGregor, 2000). According to this perspective, the followers are actively seeking cues about the leader's characteristics and it is believed that the drawn inferences have consequences for work behaviour and attitudes. The leader's character is important primarily due to the authority of the leader and its possible impact on the follower's ability to achieve his or her goals. This perspective has been referred to as the character-based perspective. In both perspectives, trust is measured as a perception or belief of the follower.
A central theme in recent social psychological research is to clarify the relationship between attitudes and behaviour (Madden et al., 1992). This theme is interesting since current theories on trust in leadership have a tendency either to focus on attitude-related and intentional outcomes or on behavioural and performance outcomes (Dirks and Ferrin, 2002).
In the first group of theories, it is often emphasized that trust is often linked to a variety of attitudinal outcomes, particularly organizational commitment and job satisfaction. Managers should be interested in their employees' attitudes because attitudes give warnings of potential problems and because they influence behaviour. Satisfied and committed employees, for instance, have lower rates of turnover and absenteeism. Given that managers want to keep resignations and absences down they will want to do the things that will generate positive job attitudes. If managers disregard to do those things distrust will occur. It has been noted by Rich (1997) that managers are responsible for many duties that have a major impact on employees' job satisfaction, such as performance evaluations, guidance and assistance with job responsibilities, and training. It is believed that when the employees do not trust their leaders they are more likely to consider quitting, as they are concerned about the decisions that are made and that these may put them at risk.
In the second group of theories, it is frequently believed that when followers feel they can trust their leaders, they will engage in behaviours that put them at risk (Mayer et al., 1995). Such a behaviour may include sharing information with others. It may also exclude a behaviour of 'covering one's back' which has a negative impact on work performance (Mayer and Gavin, 1999). In other theories, it has also been emphasized that employees are often most willing to engage in the desired behaviours if they feel that the leader demonstrates care and consideration, as a form of social exchange (Konovsky and Pugh, 1994).

Towards a Conceptual Framework of Leadership Trust
It has been found that honesty and integrity are among the core traits to be consistently associated with trust in leadership (Cunningham and McGregor, 2000). For instance, people are unlikely to look up to or follow someone whom they perceive as dishonest or who is likely to take advantage of them. When following a leader, people are willing to be vulnerable to the leader's actions if they are confident that their rights and interests will not be abused (Hosmer, 1995;Kouzes and Posner, 1993). Trust may therefore be seen as a psychological state comprising the intention to accept vulnerability based upon positive expectations of the intentions or behaviour of another (Rousseau et al., 1998). It hereby follows that risk, independence, and the willingness to accept vulnerability are necessary conditions in all formulations of trust. It is believed that both the traits honesty and integrity are central in the process of identification.
Other key qualities of managers related to trust include competence, fairness, and openness (Clark and Payne, 1997;Schindler and Thomas, 1993). Competence is vital since people are unlikely to listen to or depend upon someone whose abilities they don't respect. Generally, employees need to believe that the leader has the skills and abilities to carry out what he or she says they will do. A closely related construct is ability, that reflects the group of skills, competencies, and characteristics that make it possible for people to influence a particular environment (Mayer et al., 1995). Fairness is important since employees' evaluations of experiences and relationships are influenced by the form of social interaction. It is important that procedures, judgements, and social processes are perceived to be fair in order to feel trust towards the leadership. There is also a connection to social value orientations, in the sense that the trustee is believed to be driven by different values such as cooperativeness or altruism as a core motive for interacting with the trustor. In the latter case, we speak of benevolence, which denotes the extent to which a trustee is believed to want to do good to the trustor, aside from an egocentric profit motive. Openness matters as it is crucial to most employees whether or not one can rely on the leadership to give the full truth. Thus, Zand (1972:238) included openness in the conceptualization of trust as 'behaviour that conveys appropriate information, permits mutuality of influence, encourages self-control, and avoids abuse of the vulnerability of others'.
It is believed that the factors of identification, competence/ability, fairness/ benevolence, and openness all have great potential to explain the variance observed in leadership trust. It is also assumed that tenure may be a moderating factor in this connection, since this concept, expressed as work experience, appears to be a good predictor of both employee productivity (Gordon and Fitzgibbons, 1982) and turnover (Gatewood and Field, 1987). The concept has also been found to be positively related to satisfaction (Bedeian et al., 1992) as well as negatively related to absenteeism (Popp and Belohlav, 1982). A proposed conceptual framework is presented in Figure 1.

Decision Quality
A factor labelled ability has consistently been identified as a predictor of trust in organizational settings (Gabarro, 1978;Lieberman, 1981). According to Mayer et al. (1995) ability is 'that group of skills, competencies and characteristics that enable a party to have influence within a specific domain' (p. 717). A party's ability has been theoretically linked to trust mediated by several of the mechanisms outlined above. First, ability is related to trust by a calculative, cognitive mechanism because competence is perceived to correlate positively with the party's capacity to conceive of and implement solutions that will lead to high levels of goal achievement. Less able parties may choose solutions of poor quality or choose procedures for implementation that work poorly. Second, the ability-trust relationship can be mediated by an affective mechanism, especially in high-uncertainty situations such as those prevailing during organizational change. Organizational change is characterized by high levels of perceived uncertainty concerning relevant outcomes including job security, the maintenance of social relationships and post change job characteristics. Under such circumstances, leaders that can develop a credible vision of a future state, including a description of a path from the current to this future state are likely to elicit strong positive emotions in followers. A leader's ability and competence, however, are difficult for followers to observe directly. Although some indicators are observable (e.g. formal education, work experience, past performance) their relationships with ability are elusive and some of them are also subject to impression management and thus, provide imprecise information for persons trying to estimate ability. Further complicating the assessment of ability is the context specificity of the ability construct. As argued by Zand (1972), it is difficult to extrapolate experiences and performance from one domain to another. Instead of relying on past performance, experiences or indicators of general abilities, observers are left with the possibility to evaluate performance in the focal domain. One relevant domain for evaluating ability during change is the overall quality of the decision to be implemented. Using this logic, a leader that tries to implement a decision that is perceived to be of high quality is inferred to Figure 1. A conceptual framework of leadership trust possess a high level of domain specific ability while a leader that proposes a low quality decision is perceived to possess little domain specific ability. Based on this, we propose that: H1: There is a positive relationship between perceived decision quality in organizational change and trust in management

Communication -Social Accounts
Organizational change creates high levels of perceived uncertainty for those that will be affected. This uncertainty can relate to whether or not the organization is capable of implementing the change successfully and to what degree the change is required (Armenakis et al., 1993), but often seems to emanate from organizational members' struggle to estimate the change's consequences for their personal goals (Isabella, 1990). The high level of uncertainty leads to a need for information that is partly filled by communication from those in charge of the change. It seems that this kind of communication is particularly demanded after autocratic decision processes where change recipients have had little involvement in the early phases of the change process (Lines, 2004). Managers may be more or less aware of this latent need for change relevant information in the organization and thus respond with small or large amounts of information. Also, the framing of messages may vary according to managers' perception of the information need and the role played by communication during change. Literature on the formation of trust in leader-member relations strongly suggests that leaders exhibiting a general openness towards their followers tend to be perceived as more trustworthy than leaders keeping their thoughts, emotions and emotions close to the chest. In this vein, Whitener et al. (1998) argued that 'open communication, in which managers exchange thoughts and ideas freely with employees, enhances perceptions of trust'. (p. 517). More specifically, it seems that employees see managers as trustworthy when they take the time to explain their decisions thoroughly (Folger and Konovsky, 1989;Sapienza and Korsgaard, 1996). As shown by Cobb and Wooten (1998) managers use different forms of social accounts in order to explain their decisions. Cobb and Wooten (op.cit.) identified four different classes of social accounts (i.e. causal, ideological, referential and penitentiary) but stressed that, most often, some combination of these forms are used in order to construe an adequate message concerning the rationale behind a decision. The construction and delivery of social accounts during change may affect trust in leadership in two ways. First, social accounts can be directly linked to perceptions of trust because it signals a stance of openness on behalf of the leaders. Leaders that do not use social accounts are perceived as concealing their true motives and by consequence are seen as less trustworthy than leaders that are providing explanations for their decisions. This does not necessarily imply that the social accounts truly reflect the motives behind a decision because of the distinction between being trustworthy and appearing trustworthy (Lind and Tyler, 1988). Second, the social account-trust link is mediated by perceptions of procedural justice. Several authors have noted that it is difficult for people to observe trustworthiness directly (van den Bos et al., 1998;Bies and Shapiro, 1988). In the absence of directly observable trustworthiness, people use more observable phenomena as informational cues for trustworthiness. According to van den Bos et al. (1998) it is easier to determine to what degree a procedure is fair than to estimate trustworthiness. Because leaders that use fair procedures are thought to be trustworthy, procedural justice is used as an informational cue for estimating the trustworthiness of leaders. The use of social accounts, in turn, has been linked robustly to perceptions of procedural justice (Cobb and Wooten, 1998). Thus, we propose that: H2: The use of social accounts during organizational change is positively related to trust in management.

Participation
It is often argued that the involvement of change recipient in all stages of organizational change processes will lead to a number of positive outcomes including better decisions, less resistance and more generally a positive attitude towards change (Glew et al., 1995;Lines, 2004). Less research has focused on the more long term effects of participation, such as those captured by the trust construct. Although the involvement of those affected by change is strongly advocated in the normative literature on organizational change, managers sometimes choose to use less participation that would seem warranted from this perspective (Cotton et al., 1988). Participation has been linked to trust in prior research not explicitly focusing on trust building and destruction during organizational change. Several authors have argued that a very limited form of participation, the opportunity to voice one's opinions, should have a positive influence on trust in management (Alexander and Ruderman, 1987;Korsgaard and Robinson, 1995). Deeper forms of participation, such as the delegation of decision making power or the use of joint decision making ought to impact trust in management even stronger. Empirically it has been found that employees' trust in management is higher when they are satisfied with their level of participation in decisions (Driscoll, 1978) and when employees can determine their work roles (Deci et al., 1989). Several theoretical mechanisms can be advanced in order to account for the proposed positive participation-trust relationship. First, the use of participation can be seen as reflecting managers' trust in employees. By granting decision influence to employees, managers signal that they (the employees) can be trusted not exclusively to pursue narrow self interests when participating in a decision process, but also keep the organization's and its owners' interests in mind. As argued by Fox (1974: 67) all trust relations are fundamentally reciprocal in nature. Due to strong norms of reciprocity (Gouldner, 1960) employees are likely to reciprocate by perceiving managers who delegate decision influence as more trustworthy than those that do not. Second, the use of participation during change can be thought of as the establishment of temporary groups composed of managers and employees. Even in weak forms of participation (e.g. voice) it is likely that employees will experience the emergence of a common identity between themselves and managers. The stronger the form of participation (e.g. joint decision making, delegation), the stronger the experience of a group composed of managers and non-managerial employees is likely to become. In contrast, purely autocratic decision processes during change are likely to underscore the differences (in identity, values and interest) between managers and employees and make the distinction salient for social categorization (Turner, 1987) and category driven information processing (Hilton and von Hippel, 1996) as the change unfolds. The generation of an ingroup (a group to which one belongs), composed of managers and non managers, is likely to have profound effects on trust in management. Empirical literature on social identity suggests that the positive beliefs associated with similar group membership influence trust (Brewer and Kramer, 1985;Kramer, 2001). Also, categorizing individuals (i.e. managers) into a category to which one belongs lead employees to see managers as more trustworthy than they would have been if they where categorized into an outgroup (Brewer, 1979;Williams, 2001). Based on this, we propose that: H3: The use of participation during organizational change is positively related to trust in management Benevolence in Change -Post Change Job Characteristics As shown in the literature review above, the degree to which trustees are perceived as benevolent strongly impacts the level of trust they are granted. Benevolence is demonstrating concern for others (McAllister, 1995), and managers who act in a way that implies such a concern during change is more likely to be perceived as trustworthy than managers who show little concern for the interests of those affected by the change. Managerial demonstration of concern during change is related to how the change process is structured (e.g. decision quality, communication and participation), but also to how the change content affects the progress towards employee goals and values. Employees have many goals that they try to achieve in their work lives as reflected in the literature on work values (e.g. Kalleberg, 1977). Literature on work values distinguish between an intrinsic dimension, which refers to characteristics associated with the task itself, and a convenience dimension reflecting no excessive amounts of work. Based on this perspective, change recipients are likely to assess the degree to which an organizational change implies more or less intrinsically rewarding tasks and more or less amounts of work. By consequence, the perception of managerial benevolence during change is likely to be based on an assessment of the change's implications for those two sets of consequences. For example, a change that leads to less autonomy, less varied tasks, less feedback and less task identity (Hackman and Oldham, 1980) will create an impression of non benevolent managers. By the same logic, a change that leads to excessive amounts of work in the post change organization, will contribute negatively to the perception of managerial benevolence. Thus, H4a: There is a positive relationship between change caused task variety, autonomy, feedback and identity, and trust in management H4b: There is a negative relationship between change caused amount of work and trust in management.

Production of Trust During Organizational Change 229
The Moderating Effect of Tenure As argued in the introduction, organizational members build their trust in management on the processing of trust-relevant work experiences. The stability of the initial trust, or the degree to which trust in management is affected by the processing of change related experiences is likely to depend on the stock of relevant experiences on which the initial trust is built. The size of this stock, in turn, is dependent on the length of an individual's tenure with the organization. An individual with a long history with the organization has experienced more trust relevant episodes than an individual with a relatively short tenure. For individuals with a long tenure, trust in management is a more elaborate cognitive structure than for individuals with short tenure. This means that the trust construct is more tightly linked in an associative cognitive network to representations of more trust relevant experiences. This tight coupling, to more representations, implies that for trust to be changed, a larger network of relevant representations must be rearranged in order to re-establish cognitive consistency between trust and its underlying representations (Festinger, 1957). For example, a negative shift in trust requires cognitive work in order to account for prior experiences that indicate that management is trustworthy. In the same vein, a positive shift in trust is only feasible if the individual reinterprets prior experiences indicating that management is not trustworthy. For individuals with short tenure, less effort must be made in order to account for experiences during change that are inconsistent with the initial (i.e. pre change) level of trust in management. Based on this, we propose that: H5: The relationship between change experiences and trust in management is negatively moderated by tenure

Data, Sampling Unit and Respondents
In order to test the hypotheses above we chose to conduct a multi organizational survey of organizational changes that were completed in the respective organizations. We chose to sample multiple organizations rather than a single organization because testing of the hypotheses required variation in the processes used during change (i.e. participation and account giving) and it was reasoned that the use of data from one organization would limit the amount of variation to be observed in these constructs. The data were collected by Master students at a Norwegian business school as partly fulfilment of the requirements for a course in organizational change and learning. Students recruited organizations by obtaining a contact within the organization (typically a middle manager). During an initial interview with this contact, two changes that had been implemented within the organization were identified. The operational definition of organizational change was 'A change in organizational structures, systems, routines, technology or product market domain that was intended to further the achievement of important organizational objectives'. During this same interview, respondents that could provide information on the changes were also identified. A questionnaire was then sent to each prospective respondent by mail. The students collected the completed questionnaires approximately one week after the solicitation of participation by the respondents. This procedure yielded a total of 118 completed questionnaires (response rate ¼ 92%). None of the questionnaires were discarded. Below we provide a short description of the reported cases as described by their respondents. 62 of the respondents referred to change projects in the public sector (e.g. hospital and defence) whereas 40 respondents referred to change projects in private companies. 67, or most of the respondents referred to the change project as involving some form of reorganization. 45 of the respondents used the terms downsizing or reduction in workforce. 40 respondents referred to the change projects as involving rationalizing and the implementation of new workforms. Of these, most did not involve reorganization. 30 respondents referred to the change project as involving a merger of functions or units within an existing organization. Only 3 mentioned a merger of separate organizations. 18 respondents referred to change projects involving a closure of the unit in which they were presently employed. Of these, however, as many as 15 respondents reported that they were likely to get a new job within their existing organization.

Measures
All variables except the demographic variables were measured using multi item Likert-type scales. Where available, previously published and empirically validated, scales were used. Standard measures for the post change core job dimensions were obtained from Hackman and Oldham (1980) based on the Job Diagnostic Survey (JDS). The JDS measures the amounts of task identity, variety, autonomy and feedback from the job. Three items tap each characteristic. Principal components analysis of this item set revealed three rather than four components, as expected. Items from the variation and autonomy subscales loaded together on one single component. This component was named 'enrichment' and consists of six items (a ¼ 0.92). Reliabilities of the feedback and task identity subscales were a ¼ 0.92 and a ¼ 0.91, respectively. Items were averaged to obtain score for each characteristic. Amount of work was measured by three 7point items (1 ¼ totally disagree, 2 ¼ totally agree) tapping the respondents' thoughts regarding how the change would affect this variable. One item stated 'I have to work harder in order to get the work done' (scale internal consistency at a ¼ 0.83). A second set of three 7-point items tapped perceived changes in the amount of emotional work experienced due to the organizational change. This dimension of work amount was included because recent updates of the job characteristic model has argued that emotional work has become an increasingly important part of overall work strain in contemporary jobs (Parker et al., 1999). One item from this scale stated 'after this change, I more often have to suppress feelings that occur at work'. Participation was measured using three items taping involvement in three main stages of the change process. The items are: (1) I was allowed to participate in the analyses that were performed prior to the change; (2) I was allowed to participate in the development of the change; and (3) I was allowed to participate in the planning of the implementation of the change. Our measures of social accounts were based on the conceptual discussion of Cobb and Wooten (1998) which identified four basic types of social accounts. We used two items to tap each of these dimensions. Principal Components Analysis (PCA) (oblimin rotation) of the items revealed that in this sample, causal and penitentiary accounts were not empirically distinguishable. Oblimin rather than an algorithm for orthogonal rotation was used because the account types were expected to be moderately correlated. According to Cobb and Wooten (1998), change agents are likely to combine different types of accounts into an overall strategy for explaining change. Also, research on power and the use of influence strategies have shown that social agents use a combination rather than one strategy in order to influence their targets (e.g. Brass and Burkhart, 1993). Summary results of the PCA are reported in Table 1 below.
This reflects that in changes where causal accounts were much (little) used, ideological accounts were also much (little) used. This finding is consistent with the discussion in Cobb and Wooten (1998) which underscores that managers usually combine different types of accounts in order to justify their decisions. The two sets of items tapping these two variables were combined into one single measure labelled 'kauspen'. Internal consistency of the social account variables were a ¼ 0.79, a ¼ 0.66 and a ¼ 0.66 for the 'kauspen' accounts, ideological accounts and referential accounts, respectively. Trust in management was measured using the four-item instrument developed by Roberts and O'Reilly (1974) and previously used for studying trust in a change setting by Korsgaard et al. (2002). Internal consistency of this scale was a ¼ 0.82.

Assessment of common method variance
The measurement approach used in this research, i.e. single informants and perceptual measures, makes the results liable to contamination by common method variance (CMV, e.g. Podsakoff and Organ, 1986). CMV is defined as the overlap in variance between two variables attributable to the type of measurement procedure rather than due to a relationship between the underlying constructs (Avolio et al., 1991). CMV can arise because on consistency motives, respondents' awareness of existing theories, social desirability and mood connotations with certain questionnaires (Podsakoff and Organ, 1989;Boone and de Brabander, 1997) and its magnitude has been shown to vary across research areas (Crampton and Wagner, 1997). Our choice of measures reflects the usual trade off between this threat to internal validity and the low availability of more objective indicators with a satisfactory level of content validity (Cook and Campbell, 1979). Our judgement was to accept some possible CMV for a satisfactory level of content validity. Interpretation of the findings, however, should be made with the possibility that the estimated coefficients could be somewhat inflated due to CMV. Although this procedure has some limitations and ambiguities, we performed the Harman's one-factor test to examine the CMV problem in our data. As explained by Podsakoff and Organ (1986), the likelihood of finding one factor that accounts for the majority of variance in the original data decreases with the number of variables. Thus, we performed a series of four tests by varying the number of variables used in each. Summary results of these tests are reported in Table 2 above. As shown in Table 2, the number of variables used in the Harman's one-factor tests, ranged from 16 (Model 1, post change job characteristics and trust items) to 8 (Model 4, participation and trust items). In all four tests, the number of components that emerged was greater than one (ranging from 2 to 4 factors) and the observed pattern of item loadings were as expected. Also, trust emerged as a distinguishable factor in all tests, a result that indicates a satisfactory level of discriminant validity for our dependent variable. Based on the results above, we conclude that although some level of CMV might be present in our data, the scales discriminate sufficiently between constructs to make them suitable for hypothesis testing.

Results
Zero order correlations, means and standard deviations of the study variables are reported in Table 3 below. As shown in the table, all pairwise correlations are significantly different from 1, indicating a substantial degree of discriminant validity (Anderson and Gerbing, 1988). Also, interscale correlations are significantly lower than the estimated a-values for the respective variables, a further indication of discriminant validity (Nunnally, 1967).
In order to test the hypotheses above, we used a hierachical regression approach, where variables reflecting the content of the change were entered in the first step and variables reflecting the chosen change process were entered in the second step. Results of these analyses are reported in Table 4. The first block of variables was  found to be significantly related to post change trust in management (R 2 ¼ 0.13; p , 0.05). As shown in this table, decision quality is the content variable most strongly related to trust (b ¼ 0.21, p , 0.05). Thus, H1 is supported. Also, a significant relationship between emotional strain and trust in management is observed. No such relationship with trust was found for amount of work. Based on this, mixed support is found for H4b. No significant relationships were found between the job characteristics variables and trust. Thus, H4a is not supported. As shown in the table, the second block of variables (the process variables) was also found to be significantly related to trust. The use of ideological accounts was found to be significantly related to trust in management, so was the use of referential accounts (negatively). Also, the use of participation during change was found to be significantly related to trust. In sum, a substantial amount of support was found for hypotheses H2 and mixed support was found for H3.

The Moderating Effect of Tenure on the Relationship between Change Experiences and Trust in Management
In order to test the interaction hypothesis (H5) concerning tenure, we first attempted to follow a traditional hierarchical regression approach, using the algebraic product between tenure and the relevant independent variables as representing the moderator effect. This approach indicated a significant interaction effect (DR 2 ¼ 0.07, F DR2 ¼ 3.2, p , 0.05), but variation inflation indices (VIFs) of the tenure variable and several of the interaction variables indicated severe multicollinearity problems. The highest VIF under this model was found at 46.7, for the tenure variable. As high levels of multicollinearity can make  Table 5 above. The subgroup analyses were performed by using a mean split (tenure . 12.86 years and tenure , 12.86 years, respectively) on tenure, and estimating the regression model in each of the two tenure subgroups. Overall, the model performed significantly better for explaining trust in the low tenure subgroup. In fact the overall F-value indicated significant relationships between predictors and trust only in the low tenure subgroup regression. Results from this analyses revealed that emotional strain is weakly (p , 0.1) related to trust in management in the high tenure group whilst no such relationship is found in the low tenure group, indicating that the negative effect of emotional strain is moderated (positively) by tenure. However, the low level of significance, together with a non significant overall model leads us to interpret this finding with caution. A second difference across the subgroups was that the use of ideological accounts only is positively related to trust in management in the low tenure group (b ¼ 0.48, p ¼ 0.002). No such relationship was found in the high tenure group (b ¼ 0.07, p ¼ 0.46), indicating that the effect of ideological accounts is moderated (negatively) by tenure. Third, a stronger relationship between the use of referential accounts and trust was found in the low tenure subgroup (b ¼ 20.24 ÃÃ ). For the remaining variables, no significance between group differences in relationships with trust in management was found. In sum, mixed support was found for H5.

Discussion
Overall, substantial empirical support has been found for the theoretical framework that was proposed in this paper. Interestingly, our results indicate that aspects of the change process are more strongly related to trust in management than are aspects of the change in terms of its consequences for work load, emotional strain and post change job characteristics. In particular, the weak relationship between change consequences and trust in management merit some further comment. One dominating perspective portrays the development of trust as the outcome of an exchange history experienced by two self-interested parties. Previous research strongly suggests that job characteristics are important factors entering the evaluation of whether an organizational change has positive or negative consequences for those affected. Consequently, these factors should affect perceptions of managerial benevolence and -thus -trust in management. However, our findings indicate that change recipients' consideration of post change job characteristics plays little or no role in the development of trust during organizational change. At least two explanations could account for this finding. First, it is possible that change recipients find it hard to estimate the consequences of organizational change for job characteristics. If a high level of uncertainty is attached to the post change job characteristic estimates, their relationships with outcomes, such as trust in management, is likely to be weak. Second, it is likely that managers are seen as having more influence on process aspects of change than content aspects. The content of change could be seen as imposing itself upon the organization due to poor results, changes in competition or institutional pressures. How the process is designed and executed, however, is possibly seen as the result of managerial choice. If content aspects of the change are attributed to factors outside managerial control, its consequences for trust in management should be weaker than for the process aspects. Also, support was found for the notion that the relationship between organizational change and trust in management is stronger for employees with short tenure than for those with long tenure with the organization.
The findings of this research indicate that there is a positive relationship between perceived decision quality in organizational change and trust in management (H1). Hence the salience of perceived ability was confirmed. The observed relationship between the variables underscores the importance of considering the perception of decision quality in organizations as an antecedent of trust in management. In past research, authors have argued that decision processes sometimes the direction of the organization. Such symbolic decisions often address management's view of its employees and their roles (Culbert and MacDonough, 1985). Quite often, such processes are more perceivable to the employees than, for instance, top management's behaviour itself. Also, the nature of organizational change implies that it becomes difficult for the employees to make inferences from past experience of managerial decision making to the present due to its context dependence. Employees are simply left with the possibility to evaluate performance in the focal domain. The present research indicates that most of this argumentation presented by these authors is valid. This means that a leader that tries to implement a decision that is perceived to be of high quality is inferred to process a high level of domain specific ability. The opposite is true for leaders whose decisions are perceived to be of low quality.
Results from the hierarchical regression analyses indicate that there is no reliable effect between change caused amount of work and trust in management. This implies that the perception of managerial benevolence may not be tied to the perceived amount of work among employees. Therefore, H4b must be rejected in its present form. On the other hand, the present research also indicates that change caused emotional strain is a reliable predictor of trust in management, and that there is a negative relationship among the variables. It is therefore assumed that this factor may be a better indicator of perceived managerial benevolence. However, it could still be the case that our lack of a reliable effect of perceived amount of work is due to the measurement of the concept during a change process, since we did not either get support for task identity as a significant predictor of trust in management. Both these factors have in previous research been found to be reliable predictors of job satisfaction (Hackman and Oldham, 1980). Nevertheless, overall the results indicate that value related characteristics associated with the task itself, such as emotional strain, have a conceptually stronger link with perceived managerial benevolence than characteristics associated with convenience, such as amount of work. These are facts that support the idea that perception of managerial benevolence still is an important predictor of trust in management (Meglino and Ravlin, 1998).
Our results show that the use of participation during change is significantly related to trust in management. The question is: How can we explain and make sense of this finding, and what are the implications for the management as a change agent? The theoretical discussion contributes to several explanations for the relation between participation and trust in management, and in the following we will add some more.
Participation in change processes affect satisfaction with the process, outcomes, and acceptance (March, 1994;Glew et al., 1995). Participation means that the participants in change processes can speak out or use 'voice' in order to influence ends and means (Hirschman, 1970). The freedom to speak one's mind or participative openness might be related to norms that arise through interaction within a group and between the group and the management (Meyerson, 1994). By repeating and negotiating normative beliefs in all aspects of work, the participants can bring the norms to the level of institutionalization, and the outcome of such organizational development can place trust in management. In developing trust in management the participants will look for cues to inform them whether they should place their trust in the management.
Trust in management will depend on the management's 'willingness to be vulnerable' (proposed by Mayer et al., 1995), or depend on how and to what extent the management gives subordinates freedom to decide how to act. Therefore, trust is a way of dealing with risk represented by another person's freedom to decide how to act, and it is this freedom that represents the potential uncertainty. From this view, Kramer (2001) discusses how uncertainty in organizational settings creates a need for trust-based relations. The source of trust in Kramer's work is identitybased trust, and he shows how identity-based trust or trustworthy relationships with the management might work to help participants in change processes to cope with uncertainty. One way of doing this might be to establish relationships based on preservation of what Edmondson (1999) calls 'psychological safety' for the participants-in other words assuaging their fears of failure and personal harm.
Participation through freedom to use 'voice' means that the management gives discretionary power to the participants, and it also means a suspension of suspicious watchfulness, i.e. the trustee is given space to exercise discretionary power to make good judgement and to act appropriately. Trust placed in management is endangered if decision-making is not closely allied to the participants' interests and values (March and Olsen, 1989: 118). Appropriate management therefore means attaining a position, whereby trust is based on the ability to communicate to the participants that their interests and values are understood and taken seriously.
Hypotheses 3 suggested a general positive effect of causal accounts on trust. The results however indicate a more complex relationship between social accounts and trust. More specifically we find that different types of account show different relations to trust. Thus, whereas ideological accounts show a clear positive relation to trust, referential accounts show a negative relation to trust. For other types of accounts we find no significant relation to trust. In the development of hypothesis 3 we suggested two main effects of the use of social accounts; first social accounts were believed to signal a stance of openness on behalf of the leaders, and second; the use of social accounts was believed to influence trust through procedural justice.
Following this line of reasoning it is possible to think of different forms of accounts as varying in terms of legitimacy. Social accounts may be seen as more or less legitimate depending on their conforming to a set of norms that specify what constitutes a proper account or even what constitutes a proper account under a specific set of circumstances (Rosseau and Tijoirwala, 1999). Such norms relate to the general strength of the argument, how elaborate the argument for change is developed and to the legitimacy of the particular type of arguments used. Referring back to the results it is possible to think of employees as perceiving ideological accounts as being more persuasive and well developed and thus more legitimate than referential accounts.
Second, we posited that the use of social accounts would signal a stance of openness or transparency on behalf of the leaders. Different types of accounts may however also be seen as diagnostic of dimensions other than transparency. To the extent that some types of accounts are believed to be more difficult to apply successfully, the use of these different types of accounts may be thought of as signalling different levels of ability or even self-confidence. Thus, ideological accounts in which leaders actively attempt to spell out the justification for change may highlight the causal agency of the leaders and signal ability on the hands of the leaders. Referential accounts on the other hand may signal the opposite, by resorting to the example of other companies or organizations, leaders may be perceived by employees as to abdicate responsibility for the change process, signalling a lack of ability on behalf of the leader (Ashforth and Mael, 1989). This increased identification and in-group solidarity may generate positive affect and lead employees to develop more positive expectations regarding the motives and intensions of leaders, and increase trust. Referential accounts in contrast, by referring to the example of other companies or organizations, emphasize the superiority of other organizations.
Rather than reflecting the effects of the social accounts, the relation between social accounts may reflect the underlying properties of either the situations in which social accounts are used or of leaders that use them. First, different types of social accounts may be associated with different types of circumstances. Following this line of reasoning and the results, it is possible to see ideological accounts to be predominantly used in association with proactive and favourable change scenarios. Similarly it is also possible to imagine referential accounts being used in association with more coercive and restrictive change scenarios. The relation here would be partially spurious reflecting underlying properties of the change scenario more than the type of account provided.
Second, the use of social accounts may reflect the abilities and resources of the leaders using them. Different types of social accounts may be associated with a different set of abilities and resources or with varying levels of such abilities and resources. Highly trusted leaders may feel more comfortable using ideological accounts than would leaders that are less trusted and who command a less secure power base. This can be seen as analogous with the discussion in leadership research as of whether the observed relation between influence tactics and influence outcomes reflect the choice of tactic, properties of the person employing the tactic (powerbase) or some interaction effect (Yukl and Tracey, 1992).
Highly trusted leaders may from previous experience have learned that such accounts work and accordingly, be highly motivated to use them (Gist and Mitchell, 1992). This does not necessarily negate the suggested effects of causal accounts, accounts may still have a direct effect on trust, but we would also expect to see a selection effect in that we would expect to find trusted leaders using causal accounts that further increase trust, whereas leaders with little trust or even distrusted leaders would use causal accounts producing the opposite result. The relation between causal accounts and trust may thus reflect the differences between virtuous and negative cycles of social exchange between leaders and employees. Empirical studies of influence tactics can be seen as analogous here. These show that outcomes (resistance/acceptance) partially reflect the power base of the person using a particular tactic as well as the content of the tactic (Yukl et al., 1996;Yukl and Falbe, 1990).

Implications for Further Research
Our findings indicate that further search for general effects of social accounts per se on trust is likely to be misguided. Instead further research should direct its attention to the effects of more specific categories of accounts. Although accounts are widely acknowledged as important for employees reactions to grievances and change, few studies have tested the effects of different types of accounts. Few of these have looked at trust as an outcome. Rather than being passive recipients of accounts employees are likely to develop their own accounts as to why the organization is changing. Such accounts may be influential in forming employees' response to leaders' accounts. Most studies so far have been cross sectional. In order to fully capture the effects of accounts on trust we may have to follow organizational change processes over time. More explorative designs may be necessary to fully understand employees' interpretations of and reactions to accounts.

Implications for Practice
One key finding from this study is that organizational change represents one class of events where trust can be produced or destroyed, depending on how the change process is structured. Trust in management is itself a relatively stable construct and, in times of stability, organizational opportunity for producing trust is best seen as a scarce resource. Under stability, when organizational members follow established routines and apply established cognitive frameworks for processing task relevant information, trust in management seldom rise as an issue for reconsideration. Consequently, few opportunities for building trust are present. In organizations where trust in management is at a low level, organizational change processes can be actively used as a vehicle for building trust. More radically, organizational change could have trust building as its main purpose and other positive effects of change become by products rather than core change objectives. Especially for organizational members with medium to low tenure, a careful design and execution of change processes could positively affect trust in management.
One obvious interpretation of the findings presented above is that providing accounts does not necessarily lead to increased trust. As shown in the results some accounts seem to be associated with less, not more trust. However, from this to conclude that leaders should use ideological accounts would be premature. Neither can we conclude that referential accounts are always wrong.
As mentioned in the discussion, the use of different account types may reflect properties of both the situation and the leader using them. The effect of a specific type of account may depend on how it is used and the credibility of the leader who is using such accounts (Simons, 2002). Rather than sticking to one type of accounts, skilled leaders are likely to combine these, knowing when to use one type rather than the other.
The strong relationship between the use of participatory change processes and trust found in this study is encouraging for managers with an inclination towards democratic work practises. Arguments for increased use of participation are usually based on its assumed influence on decision quality and commitment to a decision. Additionally, in change contexts participation has also been shown to have a positive influence on organizational learning , Lines, 2005. The finding that participation is also positively linked to trust in leadership provides an additional argument for involving those affected by change in the analysis, conception and implementation of organizational change.