The causal effects of referential vs ideological justification of change

Purpose This paper details an experimental study (« = 197) that explores how different types of managerial change justifications affect employees’ reactions. The purpose of this paper is to explore the impact of managerial justification of a controversial decision in referential terms, ideological terms or a combination of the two. D esign/m ethodology/approach A randomized controlled experiment was used applying case-based video clips to ensure vividness and realism in the experimental manipulation. F ind ings The results show that referential justification caused a drop in the perceived trustworthiness of management, such that it reduced employees’ perceptions of the manager’s integrity. The effect was most pronounced in participants having elevated levels of dispositional resistance to change. The drop in perceived integrity was indirectly associated with reduced intention to support the change together with adverse affective and cognitive reactions to change. O riginality/value A robust test of different change justifications in a randomized, controlled setting, which also highlights the psychological mechanisms through which referential change justifications reduce follower trust. This result should help managers more readily understand the components of successful communication in organizational change. K eyw ords Integrity, Trust, Organizational change, Social accounts, Change reactions P aper type Research paper It is better to fail in originality than to succeed in imitation. (Herman Melville)


JO C M
Research on social accounts and change justifications provides many insights into effective communication during change (Rousseau and Tijoriwala, 1999). Some of these are related to managers' limited ability to control employees' interpretation and construal of change images to which they react (Tucker et al., 2016). Social accounts refer to the justifications and excuses managers give to explain organizational actions, such as citing new regulations as a reason for organizational change (Sitkin and Bies, 1993). Research shows that providing social accounts during change is generally associated with more positive reactions during change but that the efficiency of the social account provision depends on factors such as message credibility and the perceived sincerity of the social account giver. Social accounts research also provides insights into the antecedents, structure and functioning of the provision of justifications during change (Shaw et al, 2003). Apart from research on perceived message adequacy, scant research has focused on the content side of communication or on the possibility that different message types might have differential effects on important outcomes during change. In the present paper, we try to fill some of this void by examining how two common account types are related to trust in leaders.

Hypothesis development
In addition to examining general message content dimensions, such as adequacy, social accounts researchers have proposed schemes for classifying the types of justifications that are given. According to Cobb and Wooten (1998) and Cobb et al. (2001), social accounts can be classified as causal, ideological and referential in their content. The focus in this paper is on ideological and referential accounts. Ideological accounts are accounts that tie the change to higher-order objectives that are often part of shared ideologies (e.g. improved customer satisfaction, more employee involvement, saving money) to which objections are difficult to make. Referential accounts are justifications based on what successful organizations are doing or how they are structured or on future scenarios showing the negative consequences of not changing.
Prior research suggests that ideological justifications of organizational changes produce higher levels of follower trust (Tucker et al, 2013), while reliance on pure referential justification is associated with reduced follower trust (Lines et al., 2005). According to the taxonomy presented in Mayer and Davis (1999), trust in management consists of three subsets of evaluations; management's ability, benevolence and integrity. Research has well established that integrity is among the core traits consistently associated with trust in leaders (Cunningham and MacGregor, 2000). Integrity constitutes a form of moral uprightness that entails consistency in principles and action, even in the face of changing extrinsic incentives. In management research, the tendency to display consistency between expressed values and behavior is a key antecedent to integrity perceptions (Simons, 2002). Maintaining a high level of integrity can function as a form of reputational capital during change processes, conferring managers a wider spectrum of trust and tolerance among followers (Worden, 2003). Building on Simons's (2002) conceptual framework, we do not assume a direct link from social accounts to managerial trust, but rather an indirect one through managerial integrity. This is because managerial integrity is the key antecedent to managerial trust according to the framework. Managerial integrity reflects the degree to which employees perceive their managers as representing themselves correctly in their communications. The perceived sincerity of managers' motivated values is of central importance here. In many respects, managerial integrity indicates the extent to which stated principles are perceived as being in alignment with actions. We can therefore assume that ideological accounts have a positive impact on managerial integrity while referential accounts have a negative impact. Attributing a change initiative to the actions of another organization indicates that the change decision is formed by external forces. Imitative acts thus run counter to the notion of integrity. We therefore propose the following hypothesis: H l. Compared with ideological justification, referential justification of change decisions leads to reduced perceptions of managerial integrity.
To deepen our inquiry into the causal effects of different justifications, we also wanted to explore whether the presence of referential cues or the absence of ideological cues causes the negative effects of reliance on referential justification. Another way to formulate this problem would be to determine whether referential justification leads to a decrease in perceived trustworthiness even in cases when managers provide change justifications that combine referential and ideological reasoning. We predicted that the use of referential justifications would lead to worse perceptions of integrity, even when the referential justification was combined with an ideological justification. Specifically, we expected that the reference to other organizations would lead to reduced perceptions of integrity, even if the decision was explicitly communicated to be both referentially and ideologically motivated. The reason is that ideological accounts should be more effective than referential accounts in communicating an organizational change (Tucker et aL, 2013). Ideological accounts are supposed to facilitate employees' understanding of the change decision from the organization's point of view, leading to greater trust in management. In contrast, referential accounts reduce employees' understanding of the change but do not have any impact on trust in management. Thus: H2. Compared with pure ideological justification, a combination of ideological and referential justification leads to lower perceptions of managerial integrity.
We further expect individual differences in personality to influence the relationship between change justifications and perceptions of integrity. Prior research indicates that the propensity to resist change is a relatively stable trait, across situations (Oreg, 2006(Oreg, , 2003. Dispositional resistance is potentially important because it predisposes employees to view organizational change in a certain way, either as a positive or as a negative phenomenon. However, the level of resistance to change is bound to be influenced by other factors, such as how managers communicate the necessity of the change. For this reason, the importance of dispositional resistance lies in its capacity to influence an organization's approach to the change it needs to adopt. Open communication helps decrease the uncertainty and associated anxiety related to organizational change (Wanberg and Banas, 2000) and increase perceptions of managerial integrity (Schweiger and Denisi, 1991). However, because the type of change justification can have an impact on perceptions of managerial integrity as well, we can reason that employees' disposition to resist change will influence this relationship. For example, employees with a high disposition to resist change tend to be more cue-seeking while taking part in managerial communication than those with a low disposition to resist change. As such, they are likely to be more sensitive to the type of change justification used. In contrast, employees with a low disposition to resist change are more cue-ignorant with regard to the information conveyed and are likely to be more insensitive to the type of change justification applied. Thus: H3. The propensity to resist change moderates the relationship between the type of change justification and perceived integrity.
Perceptions of managerial integrity are one of the most consequential elements of the successful management of change (Simons, 2002(Simons, , 1999) and a crucial factor for employee commitment and managerial support (Colquitt et aL, 2007). Research has demonstrated that increases or decreases in behavioral integrity increase or decrease the trust in management, respectively. For example, Mayer et aL (1995) reported that the perceived characteristics of JO C M the trustee influence the intention to trust. In a similar vein, Gill et at (2005) found that perceived integrity of a trustee predicted an individual's intention to trust. Therefore, we assume that perceived managerial integrity mediates the relationship between different forms of change justification and trust. However, measures of trust might also take different forms; that is, trust can be cognition-based, affect-based or based on intentions to support (McAllister, 1995). Assuming an indirect link from social accounts to managerial trust, we predict that perceived integrity will be an important mediator through which different change justification lead to different forms of employee trust (intention to support, cognition-based trust or affect-based trust). We thus develop a hypothesis specifying the intermediate state that we think mediate observed social accounts-trust relationships. Direct measures of the hypothesized mediational state allow us to shed more light than in previous studies on the mechanisms underlying the purported effects of social accounts. Thus, we conjecture: H4. The relationship between different change justifications and (a) intention to support the decision, (b) affective response toward the change and (c) cognitive response to the change are all mediated by perceived managerial integrity.

Experimental design
To test our hypotheses in a manner that allows for causal inferences, we performed a randomized controlled experiment. Because we wanted to ensure vividness and realism in the experimental manipulation, we used video footage of an actress portraying a manager in an interview setting. The actress was instructed to maintain credible but identical postures, facial expressions and tone of voice in all recordings. The video started with a rolling text describing who the manager was and the nature of her recent change decision. To generate variance in perceived trustworthiness and change reactions, we opted for a decision that would affect the employees negatively and might be considered unfair. The manger's decision was to make her employees pay 25 percent of their own airfare tickets. The text further explained that the manager suspected that her employees were traveling by plane more often than was beneficial for the company and that the new fee presumably would encourage employees to travel by different means or use videoconferencing as an alternative to air travel (for the exact text, see the Appendix). This part of the video was identical in all experimental conditions. After the rolling text, the video continued to a fictitious interview, in which the manager presented one of four different justifications. These justifications represent the experimental manipulation. The control condition showed a still frame picture of the manager in the interview setting. The referential justification manipulation showed the manager explaining that the change decision was a direct imitation of rival companies: "We have been working on collecting practices from other actors in the field and learning from what the others are doing. This is something that our competitors have implemented and succeeded with. As such, we have decided to do the same." The ideological justification video showed the manager justifying her decision solely in ideological terms: "We are now taking measures in order to become more socially responsible. The climate threat is one of the biggest challenges that humanity has ever faced, and our chain has to be a part of the solution." The final version consisted of a combination of referential and ideological justifications of the decision. As in all other versions, this video started with the explanatory text and continued to the manager who provided the ideological justification. After the justification, the screen showed the text: "When asked how she got the idea for these measures, she replied ...". This text was followed by the manager providing the same referential justification as described previously.

Justification of change
Before the experiments, we pre-tested all video stimuli on a student sample. In sum, 11 participants saw both the ideological justification and referential justification versions of the video, in random order, producing a total of 22 observations. The pre-test examined how the actress came across in each video. The measurement items were how enthusiastic and interpersonally warm the manager was perceived to be. Across the observations, we found no significant differences between the groups. After the test, the participants were told how the stimuli would be used and asked if any of the videos stood out as different from the others, aside from the different wording the actress used. None of the participants indicated that any of the videos differed from the others in such a way.

Participants and procedure
We recruited participants (n = 197) from a Norwegian business school (87 women, mean age: 23 years). Participation was compensated with a gift card to the student cafeteria. Before the experiment, participants were told that the experiment would be about communication, psychology, and leadership. The participants were guaranteed anonymity and allowed to discontinue the study at any time. All participants indicated informed consent electronically, in accordance with the declaration of Helsinki. The study was given ethical approval by the Vice-Rector of Research at the business school where the experiment was conducted.
The participants were instructed to bring their own personal computers or tablets, as well as headphones. All participants assembled in an auditorium, and the entire experiment was conducted via a web-based interface, containing all questions and video manipulation. The participants first responded to a series of dispositional measures. After responding to all these measures, each participant was randomly assigned to one of the four video treatments, as described in the previous section. The video was displayed on participants' individual screens. Participants in the control condition (n = 50) saw the video in which no justification was given. The ideological justification group (n = 47) saw the video in which the manager justified her decision solely by referring to ideological factors. The referential justification group (n = 50) saw the video in which the manager justified her decision solely by referring to it as an imitation of other companies. The final group (n = 50) saw the manager who provided both ideological and referential justifications for her decision. All participants had to view the entire designated video before they could move on in the experiment. After watching their respective videos, all participants continued to give their responses on the outcome measures.

Measures
The measures used in this experiment fall into two main categories: dispositional measures of stable traits, and outcome measures of different attitudes, behavioral intentions and mental states directly related to the experiment manipulation. We randomized the order of the different scales in both dispositional measures and outcome measures, to prevent order effects from affecting our results. The only departure from this randomization was the manipulation check items, which appeared at the very end of the survey, to minimize demand effects and potential consistency bias. The main dispositional measure used in the study was a dispositional change resistance measure derived from Oreg (2003) (Cronbach's a = 0.779), which describes an individual's propensity to resist change, across situations.
For the outcome measures, the participants were instructed to respond as they would if they were employed by the organization at hand. Perceived managerial trustworthiness was measured by assessing the manager's perceived ability, benevolence and integrity on scales derived from Mayer and Davis (1999). We excluded the items measuring top management's past performance from the study because the experimental manipulation gave no information about the manager's past performance. Intention to support change was measured by the ability and is one of the three sub-dimensions of perceived trustworthiness. The third factor, labeled perceived benevolence (Cronbach's a = 0.853), captures participants' views of the manager's benevolence (i.e. the degree to which the manager comes across as being concerned about the welfare and well-being of her employees). This factor also represents one of three sub-dimensions of the concept of perceived trustworthiness. The fourth factor, labeled perceived integrity (Cronbach's a = 0.760), is the final sub-dimension of the concept of perceived trustworthiness. It consisted of four items capturing the degree to which the observed manager comes across as being principled, fair and just. The fifth factor, labeled affective response to change (Cronbach's a = 0.721), consists of two items detailing negative emotional reactions to the communicated change. The sixth and final factor, labeled cognitive response to change (Cronbach's a = 0.709), consists of three items measuring negative cognitions toward the communicated change.

Results
To explore the hypotheses, we conducted analyses of variance (ANOVAs), with least significant difference post hoc tests. We first determined that the manipulation checks gave results in line with our predictions. The groups differed significantly, and in the expected directions, in their responses to the statements this company has original ideas (F(3,173) = 9.851,/) < 0.001), the company is innovative (F(3,171) = 4.489, p = 0.005), the company imitates others (F(3,171) = 9.236, p < 0.001), the company communicates that the change is imitative (F(3,169) = 50.47, p < 0.001) and the company claims ideological reasons for the change (F(3.171) = 4.132, p = 0.007).
Our main prediction was that referential change justifications would lead to reduced trust in management. H l stated that the groups would differ in their perceptions of managerial integrity. The results from the ANOVA revealed between-group differences in perceived integrity (F(3,173) = 3.705,/) = 0.013). The post hoc test indicated that the group that saw the referential change justification video perceived the manager as having significantly less integrity than the group that saw the ideological change justification video (/) = 0.037). This finding provides support for H l.
H 2 declared that the combination of referential and ideological change justifications would lead to lower perceptions of integrity than ideological justification alone. The post hoc test indicated that the group that saw the combined justification video gave significantly lower integrity ratings than the group that saw only the ideological justification video (/> = 0.021). This finding provides support for H2.
H 3 proposed that the tendency to consider referential justifications indicative of a lack of integrity would be more pronounced for people with a stronger dispositional resistance to change. To test this hypothesis, we employed the PROCESS macro for SPSS (Hayes, 2013). We entered the referential vs ideological change justification in the first step of the regression analysis. In the second step, we entered the interaction term between the different justifications and propensity to resist. The interaction term explained a significant increase in variance in perceived integrity (A/?2 = 0.081, fl = 0.470, t = 2.855,/) = 0.005). Thus, the propensity to resist change significantly moderated the relationship between type of change justification and perceived integrity. This result provides support for H3. Participants with a strong dispositional propensity to resist change perceived the ideological change justification as a significantly stronger indication of managerial integrity than participants with a low propensity to resist change. Figure 1 illustrates the moderating effect of dispositional propensity to resist on the relationship between type of change justification and perceived integrity.
H4 stated that perceived managerial integrity would mediate the important outcome variables: (a) intention to support the decision; (b) affective response to the change; and (c) cognitive response to the change. This prediction implies that differences in perceived change commitment scale derived from Fedor et a,l (2006). This scale consists of four items that capture behavioral intentions as representations of commitment. The main benefit of using intention statements rather than pure attitudinal statements is that the former is more strongly associated with actual behavior. Finally, we measured affective and cognitive intentions to resist the proposed change using scales from Oreg (2006). We reworded the items used as hypothetical statements, rather than measures of past behavior. The five items in this scale that measure behavioral resistance fit poorly with the purpose of our experiment, as they all entail actual past behavior (e.g. "I protested against the change"). As such, we excluded these items from the experiment. All measurement scales used in the study were administered in their original language (English).
We entered all outcome measurement items into a maximum likelihood factor analysis, with direct oblimin rotation. The maximum likelihood estimator is asymptotically efficient and less inclined to recover weak factors (Winter and Dodou, 2012). Five items produced cross-loading and thus were excluded from further analyses. The final measurement model (see Table I) consisted of six factors, congruent with our expectations. The first factor was labeled intention to support (Cronbach's a -0.902). The second factor, perceived ability (Cronbach's a = 0.898), captures participants' views of the observed manager's professional

JO C M
integrity, caused by different change justifications, would be directly associated with the outcomes of those variables. The ANOVAs did not reveal any direct effect of type of justification on the ultimate outcome variables. However, although direct effects between A and Y failed to materialize, it is still possible and useful to test and report the indirect relationships pertaining to H4. Exploring mediated effects provides information about whether the proposed mediator has a positive or negative denotation and can also suggest directions for future research in terms of identifying suppressing variables that are not measured in the presented experiment (see Hayes, 2009). In mediation analyses in which the direct effect between X and Y is absent, the proposed mediators are better referred to as indirect relationships rather than mediated relationships (Mathieu and Taylor, 2006). We employed Hayes's (2013) PROCESS macro for SPSS. The macro uses a bootstrapping method with bias-corrected confidence estimates. All three tested models used the referential vs ideological change justification as the manipulated variable (A) and perceived integrity as the mediating variable (AT). In the first tested model, we used intention to support the change as the outcome variable (E). The analysis shows that the indirect path from type of justification, through perceived integrity, to intention to support the change was statistically significant, with confidence intervals below zero (from -0.384 to -0.029). The coefficients from the mediation model indicate that referential change justification significantly reduced perceived integrity (ß = -0.261, t = -2.201, p = 0.03), and, in turn, perceived integrity was associated with an increase in intention to support the change iß = 0.641, t = 4.288, p < 0.001). This finding provides support for H4a. We then undertook the same analytical procedure, only with affective response to the change as the outcome variable (E). Again, the indirect path from type of justification, through perceived integrity, to affective response to the change was statistically significant, with confidence intervals above zero (from 0.019 to 0.302). The coefficients from the mediation model indicate that referential change justification significantly reduced perceived integrity iß = -0.256, t = 2.157,/> = 0.034), and, in turn, perceived integrity was associated with a decrease in negative affective response to the change iß = -0.487, f = -3.398, p < 0.001). This finding provides support for H4b. Finally, we performed the same statistical operation using cognitive response to change as the outcome variable (E). The indirect path from type of justification, through perceived integrity, to cognitive response to the change was statistically significant, with confidence intervals above zero (from 0.024 to 0.356). The coefficients from the mediation model indicate that referential change justification significantly reduced perceived integrity ß = -0.256, t = -2.157, p = 0.034), and in turn, perceived integrity was associated with a decrease in negative cognitive response to the change ß = -0.594, f = -5.083,/» < 0.001). This finding provides support for H4c. Across all the outcome measures proposed in   Figure 2, we found that referential justification produced worse perceived integrity than ideological justification. This negative effect of referential justification was indirectly associated with less intention to support the change, more negative affective reactions to the change and more negative cognitive reactions to the change. Figure 2 displays the results from the mediation analysis.

Discussion and conclusion
The reported experiment tested the effects of different types of change justifications on employees' reactions. We find that referential justification causes a drop in trust in management, in that it reduces employees' perceptions of managers' integrity. This effect is most pronounced in participants with elevated levels of dispositional resistance to change.
The drop in perceived integrity, caused by referential rather than ideological justification, is also associated with reduced intention to support the change as well as more negative affective and cognitive reactions to the change, respectively. This study corroborates previous research finding that social accounts are related to trust (Lines et al., 2005), which has not been demonstrated experimentally before. From a causality standpoint, it is important to note that social accounts used by managers also influence follower trust in organizational change settings that are experimentally manipulated and controlled. Previous studies have used a retrospective method of measuring social accounts. However, followers' evaluations of a change project could, after a certain period, be the victim of memory decay. This is especially true when investigating perceptions. Our experimental method remedies this potential flaw, as followers' recall of the social account was not distorted by time lapse and thus should be more accurate.
Our findings suggest that communication explaining the reasoning behind changes is necessary and important. When such information is not provided, serious negative consequences for the change initiative might result. It is reasonable to assume that the individual use of social accounts results in an increased search for information, as most individuals normally try to understand the context of the explanation provided and the possible consequences of the decision or the change initiative (Lines et al., 2005). The use of different social accounts in the communication of the organizational change can therefore lead to differing supporting attitudes and behavior among followers (Tucker et al., 2013).
This study is not without limitations. One of the main limitations of the experiment is its lack of external or ecological validity. Although we designed the manipulation to be as vivid and realistic as possible, generalizing results from experiments to a real-life setting must always be done with caution. However, in this particular case, the reported findings are largely congruent with past findings from survey studies and observational data from real-life change processes in actual organizations. As such, we argue that the results should generalize better to ecological settings than would be the case with most experimental data. An important goal of research in change management is to predict reactions to different organizational changes under different circumstances. As such, the ideal measurement would be observable behavior from managers and employees. Our measures thus carry an inherent limitation to the presented conclusions. We measured responses and reactions to i the change through an experiment rather than directly observable behavior. However, every operationalization is flawed relative to the construct on which it is based (Shadish et aL, 2002). As prior research has demonstrated that measures of attitudes alone have limited predictive value on actual behavior, we used mainly measures of behavioral intentions, which have a stronger association with actual behavior. However, field experiments may prove useful in exploring the reported relationships further. Finally, although we demonstrated followers' acceptance of different social accounts during change, we offer limited insight into the legitimacy of their sense making. Future research could address this issue by applying an appropriate experimental design. Moreover, although we tested the social accounts-trust relationship in a rigorous way, many definitions of trust exist. Future research could explore additional types of trust that may be relevant during organizational change (see McAllister, 1995). The nature of social accounts is also complex because it is tied to different levels of the organization (i.e. aligns strategic-level leadership with operational-level followership). Our focus was on how followers perceive social accounts rather than on their creation and design. Thus, future research could experimentally test a multi-level study design of social accounts (Cobb et aL, 2001;Frey and Cobb, 2010;Skarlicki et aL, 2004).
Despite these limitations, our study has convincingly argued for the necessity of examining the conditions under which employees are likely to develop trust in management. Our findings could thus be used to suggest explanatory mechanisms related to organizational climate and trust in management. We suggest that the type of justification used by management, employee personality and employee reactions to change should be modeled as contextual variables in an attempt to build a general theory of sources of trust in management.