525442
research-article2014
JBCXXX10.1177/2329488414525442Journal of Business CommunicationChristensen
Article
Communication as a Strategic Tool in Change Processes
International Journal of Business Communication 2014, Vol. 51(4) 359–385 © The Authors 2014 Reprints and permissions: sagepub.com/journalsPermissions.nav DOI: 10.1177/2329488414525442 jbc.sagepub.com
Marit Christensen1
Abstract The aim of the study was to develop, test, and partly validate a set of organizational communication factors for use in an organizational-change setting. Based on literature reviews and pilot interviews, a survey study was conducted using three samples. First, the testing involved construct validation through exploratory and confirmatory factor analyses. Second, the sample was divided into three groups: employees who experienced change, those who had recently had finished a change process, and those who had not experienced any change processes. The communication factors were then examined as predictors of employee assessment of communication in these three groups, respectively. The results of the study indicated eight robust dimensions included in five categories: social contact, central leadership, information, influence, and barriers to improvement. The results of the factor analyses indicated satisfactory reliability and construct validity of the communication factors, and the confirmatory factor analysis revealed a satisfactory model fit. Keywords organizational communication, organizational change, measurement, validation
Organizations today are confronted daily with the need for change. A major challenge for organizations is to develop both a culture or climate and leadership strategies that allow them to cope with challenges such as downsizing, re-engineering, flattening structures, global competition, and the introduction of new technology (Waddell, Cummings, & Worley, 2011). One approach to a successful change process could be through the establishment of good communication and information strategies within 1Norwegian
University of Science and Technology (NTNU), Trondheim, Norway
Corresponding Author: Marit Christensen, Department of Psychology, Norwegian University of Science and Technology (NTNU), 7491 Trondheim, Norway. Email:
[email protected]
360
International Journal of Business Communication 51(4)
the organization, while implementing change. Communication is well recognized as instrumental to organizational survival and growth (Bordia, Hobman, Jones, Gallois, & Callan, 2004; Schweiger & DeNisi, 1991; Wanberg & Banas, 2000). This fact has served as a driving force in the development of efficient instruments for the measurement and analysis of communication in organizations (Downs, DeWine, & Greenbaum, 1994). The basic idea is that a communication audit creates a foundation for diagnosing the quality of the communication practice. Auditing communication involves the assessment of current communication practices to determine what steps are required for improvement. Based on this line of thought, several instruments for measuring organizational communication have been developed (see Downs et al., 1994, for a review). The existing measures are diverse and differ in their aims regarding communication focus. Additionally, few of them are related specifically to change. Most of the instruments have only been used once. The aim of this study was therefore to develop, test, and partly validate a measure of organizational communication related to change. The first step was a study of the literature on the role of organizational communication during change processes. The following section provides a review of the literature for the pilot interview study. Resistance to change is a phenomenon that slows the change process by hindering its implementation and increasing costs. Some of the drivers of responses to change could be economic fear, uncertainty, inconvenience, threats to interpersonal relationships, the impact on internal processes and systems, and the impact on social functions. Resistance is mostly considered negative; however, that does not tell the whole story. Change is not necessarily beneficial to the organization, and resistance can reveal to managers important aspects of the situation that have not been properly considered (Pardo del Val & Martinez Fuentes, 2003). Change can generate deep resistance in employees and organizations, which makes it difficult and sometimes impossible to implement change processes (Garvin, 2000). Employees mostly resist change when they are insecure about its consequences. The lack of information and communication with managers could give life to rumors and gossip and lead to anxiety associated with the change. Effective communication about the changes and consequences could reduce this anxiety and increase the feeling of mastery (Argyris & Schön, 1996). However, communication is also a difficult and frustrating aspect of managing change in an organization. There is, therefore, a strong need to have wellconsidered communication and information strategies during the planned change process. Another effective strategy to avoid pitfalls and hindrances is to involve employees in planning and implementing the change process with regard to information and ideas. It could increase the likelihood that the employees’ interests and worries are accounted for and thereby increase the motivation to implement the change process (Waddell et al., 2011). The concept of organizational communication is, according to Goldhaber (1999), marked by diversity. The many definitions of organizational communication reflect a wide range of approaches and perspectives. In analyzing the content of these definitions, Goldhaber identified three common features: Organizational communication (a)
Christensen
361
takes place within complex open systems (it is influenced by the environment and influences it as well); (b) entails all features of a message (i.e., flow, purpose, direction, and media); and (c) involves people, including their attitudes, feelings, relationships, and skills. Drawing on these features, Goldhaber proposes the following definition: Organizational communication is “the flow of messages within a network of interdependent relationships” (Goldhaber, 1999, p. 36). Research suggests that the concept of organizational communication is multidimensional. For example, on one hand, Tukiainen (2001) found that employees perceive four dimensions of meaning in organizational communication: (1) personnel’s use of the communication system; (2) the management’s conduct and methods of communication, including the perception that separate communication channels exist for workers and leadership, how the leadership’s decisions and communicative behavior are perceived, and the extent to which communication is considered regular, organized, and equal; (3) face-to-face communication; and (4) the efficiency of the communication process. On the other hand, Johnson (1992) reviewed four major approaches to organizational communication structure and identified five dimensions: relationships, entities (e.g., employees, work units), contexts, configuration (e.g., patterns of organizational behavior), and temporal stability. These examples are given merely to illustrate that the structure of organizational communication has been described in different dimensions. The variety is not surprising, considering the wide scope of the field. Organizational communication is considered a specific discipline, although it cuts across a wide range of research fields. The broad scope of the discipline of organizational communication is also reflected in the variety of different organizational communication measurements that have been developed. The most widely used measures are in two categories: process instruments (e.g., measuring issues of conflict management, team building, and communication competence) and comprehensive instruments (involving an overall approach to communication). Validated instruments that have been applied frequently include the Communication Satisfaction Questionnaire (CSQ; Downs & Hazen, 1977), which emphasizes the relationship between communication and job satisfaction; the Communication Audit Survey (also known as the International Communication Association [ICA] audit survey), identified as one of the most comprehensive attempts to measure all aspects of organizational communication (Goldhaber & Krivonos, 1977); the Organizational Communication Development Audit Questionnaire (OCD2; Wiio, 1975), which determines the efficiency of the communication system for achieving an organization’s goals; and the Organization Communication Scale (OSC; Roberts & O’Reilly, 1974), which compares communication practices across organizations. This is by no means a complete list but is meant to give a sense of the scope of the field. The complexity of organizational communication demands a broad perspective. According to Downs et al. (1994), the extensive span of the research area and the use of imprecise parameters created a challenge in establishing the field’s current status. A similar point had already been made in 1974 by Roberts and O’Reilly, who called attention to the lack of systematic development of instrumentation to measure
362
International Journal of Business Communication 51(4)
organizational communication. According to Downs et al., it is unfortunate that so few of the organizational communication measures have been used sufficiently to generate the mandatory levels of reliability and validity. Based on the argument of Downs et al. (1994), two tasks would seem to be entailed in the development of an organizational-communication measure. First, although several instruments have been developed and validated, the multifaceted nature of the field renders it difficult to define the direction to be taken in instrument development. In other words, how does one identify the common denominator to be used to compare a new instrument with previous measurements? It is argued here that a key factor in measurement development is the identification of a background or starting point that can serve as a common reference point, allowing comparison with other approaches. Specifying a background that can be used in any approach to organizational communication increases the possibility of comparing results and identifying the status of the field. The second task concerns the problem of designing an instrument that will be used only once, with no further attempt to certify its validity. Thus, a key issue becomes the development of an instrument that is easy to use, requiring frequent validation. The theoretical background for the study is elaborated next.
Content of the Communication Instrument A study of the literature and a pilot interview study were important to defining the five categories of the communication instrument that were extracted and tested in this paper. As a first step in the development and validation of an instrument of organizational communication, item generation is essential. In this paper, both a deductive approach and an inductive approach were used. The basis for the deductive approach, which involves providing a theoretical foundation for organizational communication and the chosen categories, is provided below. In addition, the inclusion of an inductive approach, in which respondents were invited to share their experiences of organizational communication in change processes, was thought to be of interest. The categories extracted from the pilot studies and literature review included social contact, central leadership, information, influence, and barriers to improvement. Within some of the categories, several themes demonstrated that each category had several facets. An elaboration of the categories’ theoretical and empirical foundation is provided below. Social contact was the first category developed from the pilot interviews. Within this category, respondents identified several elements as crucial to well-functioning communication systems in change processes. In this category, respondents highlighted certain themes: contact between coworkers and leadership, contact among coworkers, and finally the fact that disruptive relations between colleagues and leaders could devastate an organization. The respondents also pointed out the importance of distinguishing between levels, that is, distinguishing between contact with central leadership and contact with the closest leader in times of reorganization and change. The communication and information processes could be experienced quite differently based on who initiates the rationale and strategy behind a planned change process. Central leadership was therefore defined as a specific category, the second category. According to
Christensen
363
Goldhaber (1999), organizational communication pertains to the attitudes, feelings, relationships, and skills of people. In line with Goldhaber’s suggestion, others have also emphasized the social component of organizational communication. For example, Tukiainen (2001) reported that the management’s conduct and face-to-face communication shaped employee perception of organizational communication. Johnson (1992) also underlined the social component of organizational communication, suggesting that social relationships represent one of the key dimensions of organizational communication structure. Finally, Towers-Perrin (2006) concluded that one of the key hindrances to employee engagement is the difficulty of building appropriate connections between the leadership and the workforce. The third extracted category involved information, especially information content, quality, timing, and structure. Previous research (Johnson, 1992; Tukiainen, 2001) emphasized the importance of employee use of the communication system and the system’s efficiency. Rafferty and Jimmieson (2010) found that the information climate regarding change had a positive relationship with the quality of work life and a negative association with role ambiguity, work overload, and distress. They also found that the participation climate during change had a positive relationship with quality of work life. This is in line with the results of Van Vuuren, de Jong, and Seydel (2007), who reported that feedback from managers to employees is a critical factor in shaping communication climate and employee commitment. Communication and information in the workplace during change and reorganization are essential to employee job satisfaction (De Nobile & McCormick, 2008). The fourth category dealt with influence. Respondents emphasized the importance of both their perceived influence and participation in decision making as well as the important role of the labor unions, especially in the reorganization and change processes. Wilson and Peel (1991) defined participation as allowing individuals to be involved to a high degree in the planning and control of their own work and as having enough knowledge and power to influence processes and results in order to reach desirable goals. A meta-analysis including 43 studies revealed that profit-sharing, worker ownership, and worker participation in decision making were all positively correlated with increased productivity (Doucouliagos, 1995). Participation during change seems to empower the employees and might function as a predictor of acceptance of change (Gagné, Koestner, & Zuckerman, 2000; Wanberg & Banas, 2000). The fifth category was concerned with barriers to the improvement of communication during organizational change and hindrances to good communication practices. Earlier studies demonstrated that relationships, rules, policies, and resources, or lack of them, affect communication team functioning. Koontz (2001) summarized the barriers to communication with the notion that communication problems are often symptoms of problems that are more deeply rooted: for example, poor planning and lack of common goals for the organization, communication overload, badly expressed messages, lack of understanding, and hostility between the participants and bureaucracy. As mentioned, resistance to change might be a major barrier to successful reorganization because it triggers a perceived lack of control, insecurity, and anxiety, among other things (Bruckman, 2008; Erwin & Garman, 2009; Ford & Ford, 2009). Lack of
364
International Journal of Business Communication 51(4)
information and good communication strategies might increase these experiences (Waddell, Cummings, & Worley, 2007).
Theoretical Presupposition The contemporary focus of the theoretical field of organizational communication is closely related to the issues of organizational change (see Salem, 1999, for a review). Communication is regarded as key to the successful implementation of change programs, because it is a tool for announcing, explaining, or preparing people for change, as well as for preparing for positive and negative effects of the impending change (Spike & Lesser, 1995). According to Barrett (2002), effective communication is the glue that holds the organization together. During organizational change, the effectiveness of this glue becomes critical. Efficient communication is essential because it links all organizational processes. As pointed out by Galpin (1996), any organizational change strategy begins by establishing the need for change. A critical decision is then related to how to communicate this need to members of the organization. The rationale for change must be communicated clearly, as must the consequences of not changing, not to mention how the organization will appear after the changes (Galpin, 1996). The rate of change in organizations has been increasing during the past 20 years (see, e.g., Burnes, 2004; Jones, Dunphy, Fishman, Larne, & Canter, 2006), and there has been considerable debate about whether change actually brings the expected benefits to the organizations or those who work in them (Jones et al., 2006). There are two main approaches to change: the planned and the emergent. The main criticism of planned change is that it attempts to make order and a linear sequence out of something that is really unpredictable, messy, and untidy (Buchanan & Storey, 1997). Weick (2000) argues that emergent change consists of ongoing accommodation, adaption, and alteration that produce fundamental change without a priori intentions to do so. Both of these approaches indicate that change in any fashion is unpredictable and unstable and in many cases fails to achieve its objectives. One of the major aims of communication during change is to reduce the experience of job insecurity among employees (Allen, Jimmieson, Bordia, & Irmer, 2007). Good and effective communication seems to reduce psychological insecurity about change and, additionally, increases acceptance, openness, and commitment to change (Bordia et al., 2004; Schweiger & DeNisi, 1991; Wanberg & Banas, 2000). The failure to provide sufficient information or provide inferior information can have negative consequences, such as perceived cynicism about change (Wanous, Reichers, & Austin, 2000). Rafferty and Jimmieson (2010) found that the climate of information regarding change has a positive relationship with the quality of work life and a negative association with role ambiguity, work overload, and distress. They also found that the climate of participation related to change has a positive relationship with the quality of work life. Processes of organizational change are thus intimately related to communication processes. Although it is well recognized in the literature, Galpin (1996) noted that organizations have devoted most attention to technical, financial, and operational aspects of change. This is echoed by Barrett (2002), who argued that some companies
Christensen
365
still do not see that organizational changes, and the management of change, are hindered without effective communication. Strategic communication is a key aspect in Galpin’s (1996) description of the change process, which begins by preparing strategies for communication. Furthermore, communication by means of feedback and employee participation drives the change process forward. Based on the key role of communication in organizational change, we suggest that organizational change may serve as an arena for the development and assessment of organizational communication instruments.
Method Aim and Design The main objective of the work was to develop and test an instrument of organizational communication and especially examine its performance during phases of organizational change. The design, therefore, included the development and testing of the instrument itself, followed by a test of the instrument’s performance in groups defined in accordance with three phases of organizational change. It was of specific interest to investigate the performance of the instrument in the phase during which employees experienced organizational change. According to Hinkin (1998), the factor-development process includes (1) item generation, (2) questionnaire administration, (3) initial item reduction, (4) confirmatory factor analysis, (5) convergent/discriminant validity, and (6) replication. The results as reported in the present study followed Steps 1 through 4 in this suggested process. As convergent/discriminant validation (Step 5) was not available as an option, a predictive validity test was conducted. The predictive validity of the communication factors was examined by testing how the factors related to communication during organizational change. This was done by examining how the factors predicted the employees’ assessments of communication in three different phases of organizational change. The employees were divided into three groups: employees experiencing no change or reorganization, employees who were in a change process, and employees in organizations that had recently completed their reorganization or a change process. Replication (Step 6) was not performed in the current study. As described above, the questionnaire was developed and based on a review of the literature and a pilot interview study. The questionnaire was administered in three separate samples, involving different organizations, occupations, and work positions, in order to examine the robustness of the dimension structure, which consisted of 76 items. In Step 1, the 76 items were organized into five categories based on their meaning content: (1) social contact (36 items), (2) central leadership (5 items), (3) information (14 items), (4) influence (10 items), (5) barriers to improvement (11 items).
The Aims of this Paper The overall objective of the present work was to develop, test, and partly validate a specific set of organizational communication factors as well as to examine how the
366
International Journal of Business Communication 51(4)
factors relate to organizational change. The rationale lies in the presumed value and centrality of organizational communication, that is, communication as a strategic part of organizational-change management. The content of the communication factors was designed based on both a review of the literature and a pilot interview study. The review of the literature helped to clarify the construct of organizational communication and guided the selection of items. A sample of six respondents from a variety of occupations and positions were asked to describe their experiences of organizational communication, especially related to change processes. Their responses were classified into five categories based on key words or themes: social contact, central leadership, information, influence, and barriers to improvement. After the items had been generated, the next step was to administer a questionnaire to three different samples, with the objective of examining how well those items confirmed expectations about the psychometric properties of the measure of organizational communication. The third step of the process was to ascertain the reliability and validity of the instrument, and it involved initial item reduction through principal component analyses in all three samples. Reliability represents the instrument’s precision and is a necessary condition for validity. The internal consistency of each new dimension was measured by Cronbach’s alpha. In the fourth step, confirmatory factor analyses were conducted to test the adherence of the results of the prior principal component analyses to the data. A weakness of principal component analysis is its inability to test the goodness of fit of the resulting factor structure. A confirmatory factor analysis allows the assessment of the factor structure by statistically testing the significance of the overall model and of item loadings on factors. Two confirmatory factor analyses were conducted in this study. They were both performed on the largest sample. The first analysis was performed to test the model fit of the eight-dimensional measurement model previously found in the PCA analyses. The second analysis was performed to test whether the eight-factor communication instrument was a homogeneous measure of organizational communication with different facets. As a result, a confirmatory factor analysis with a second-order common latent dimension of organizational communication also was conducted. Finally, predictive validity was tested in the last step to examine how the eight communication dimensions predicted the “employees’ assessments of communication,” an item measuring employee evaluation of organizational communication. These analyses were conducted in three different groups in the second largest sample—no reorganization, current reorganization, and finished reorganization—to investigate whether the instrument, with the included eight dimensions, was able to differentiate the performance of the instrument.
Data Collections and Their Design The pilot study included six interviews with three managers and three employees, who were chosen for participation because of their recent experience with change processes in their organizations. Of these respondents, one leader and one employee were chosen
Christensen
367
from the private, municipal, and public sectors. In the pilot interview sample, half of the interviewees were women. Managers and employees were located in Trøndelag, in the region of mid-Norway. The respondents were contacted by telephone, and agreements were reached about their participation and interviews. Each interview lasted 30 minutes, and all were audio-recorded with the permission of the interviewee and thereafter transcribed and analyzed. The pilot interview work, together with the literature review, formed a basis for the development of the 76 items used in the questionnaire. Sample 1 was based on a cross-sectional survey design and consisted of employees working in the social, labor, and health sectors in Norway (n = 470). Five different types of services were contacted for participation. A total of 24 municipalities were carefully chosen on the basis of size and on whether they had a cornerstone industry1 that was downsizing or that had closed down. The design included a control sample. Among the studied municipalities, 12 had a cornerstone industry that had recently closed down or was downsizing. For the purpose of comparison, another 12 municipalities were chosen based not on having a downsizing cornerstone industry but on their resemblance to the former 12 municipalities in size and location. These are referred to here as “control municipalities.”2 Each control municipality was compared with the 12 municipalities with a downsizing cornerstone industry, using the criteria for comparability of the Norwegian Central Bureau of Statistics3: geographical location, settlement pattern, economy of the municipality, average level of education, income after tax per inhabitant, population, age-structure, and employment statistics (Langørgen, Galloway, & Aaberge, 2006). The design also divided the municipalities according to size; half of each subsample was selected from small municipalities (0-4,999 inhabitants), and the remaining half was selected from medium-sized municipalities (5,000-19,999 inhabitants). The division was also based on the grouping by the Central Bureau of Statistics of municipalities according to size. The mean age of the employees was 46 years (SD = 10.50); 73% were female and 27% were male. The leaders of each service received a letter of invitation and were contacted again by telephone for participation. Sample 2 included a selection (convenience sample) of employees working in the production industry and in the transport sector (n = 192). Two local industrial production companies, as well as a local office of a transportation labor union, were contacted and engaged. Participation was voluntary, and the study had the consent of the respective leadership and employee representatives or unions. A contact person in each organization or office distributed the questionnaire to the employees. Participants came from leadership, production, and transportation. The mean age of the respondents ranged between 40 and 50 years of age; 89% were male and 11% were female. Sample 3 was a national quota sample collected in 2004 of Norwegian employees working in different organizations, occupations, and positions (n = 1,002). A quota sample is collected by a stratified sampling technique in which the number of cases to be sampled is predetermined (Reber, 1995); in this case, the sampling represented the overall national distribution of occupations. The distribution of the respondents’ areas of work will be presented in four categories: production and industry (17%; e.g., agriculture, fishing, and industry), service sector (18%; e.g., retail and vehicle repair services, transport, bank/insurance/finance, and
368
International Journal of Business Communication 51(4)
Table 1. Description of the Three Included Samples.
Type of employees Gender distribution n
Sample 1
Sample 2
Sample 3
Public 73% female 470
Private 11% female 192
Diverse 56% female 1,002
the service industry), public administration (46%; e.g., research and education, health and social services), and others (19%; see the Statistical Yearbook of Norway 2005 [Central Bureau of Statistics, 2005]). The mean age of the employees was 44 years (SD = 11.49); 44% were male and 56% female. See Table 1 for a description of the three samples.
Questionnaire and Measures Organizational communication was measured using 76 items in the questionnaire. The respondents were presented with statements to be checked on five-point scales ranging from 1 = not correct to 5 = absolutely correct, and 1 = never to 5 = always, as well as 1 = not useful to 5 = invaluable. The first category, social contact, consisted of three dimensions: (a) contact with leadership (nine items, e.g., “the dialogue between my closest leader and me is good”), (b) disruptive relations (five items, e.g., “rumors are flourishing between the employees”), and (c) coworker contact (four items, e.g., “the communication between my coworkers and me is good”). This category of social contact measured the social relationship and dialogue between the leader and the employees and among coworkers. The first dimension, contact with leadership, measured the relationship between the employee and the closest leader and included aspects such as tolerance, openness, respect, and trust. The second dimension, disruptive relations, measured detrimental relationships among the employees and included such topics as rumors, conflicts, and lack of communication. The third dimension measured the relationship and contact among coworkers. The items included trust and solidarity. The second category consisted of only one dimension, central leadership (three items). It measured the relationship between employees and their central leadership, related to topics such as openness, involvement, and regular contact (e.g., “The communication with the leadership is marked by openness”). The third category, information, also had one dimension: information (seven items, e.g., “the information provided is relevant for me”). The items dealt with relevance, clearness, and comprehensibility as well as the amount of information that is given and the timing of the information. The forth category, influence, had two dimensions: (a) perceived influence (three items, e.g., “my opinions are heard”) and (b) perceived union assistance (two items, e.g., “The unions bring my view/opinion further on”). The first dimension measured the influence experienced by the employees in their workplace, involving items of
Christensen
369
being heard, time and place for discussion, and inclusion. The second dimension measured the employees’ experience of union assistance regarding information about rights and bringing the employees opinions forwards. The fifth category, barriers to improvement, had only one dimension (six items). It measured different barriers to well-functioning communication in an organization (e.g., “certain individuals may hamper communication”). Employee assessment of communication was measured by a single item from the General Nordic Questionnaire (QPS Nordic) for psychological and social factors at work (Dallner et al., 2000). The item asks, “Is there sufficient communication in your department?” The item was measured on a five-point scale where 1 = very seldom or never, 2 = rather seldom, 3 = sometimes, 4 = rather often, and 5 = very often or always.
Data Analyses The pilot study used qualitative interview techniques to gather data on organizational communication during change. The interview questions concerned the respondents’ perceptions of communication in their organization and their appraisal of ideal organizational communication. The interviews were read and an overall framework was established. The analysis process started with data reduction by focusing on relevant parts in the text to be included in a survey on organizational communication during change. The next step was to sharpen, focus, and organize the data by developing units within organizational communication for use in a survey instrument. Together with a review of the literature, this material was used as a foundation for developing the organizational communication instrument. Three principal component analyses (PCAs) and reliability analyses were performed, in three separate samples. The samples were all part of the project “Work and Health in a Changing World” (financed by the Research Council of Norway). Preliminary Factor Analyses. Principal component analyses were used to examine the dimensionality of the hypothesized communication components in all three samples. The questionnaire was based on items describing the aforementioned five specific areas within organizational communication. The assumption was that the variables were indicators of different dimensions as presented above. The Kaiser criterion has been referred to as the most appropriate for evaluating the results of a principal component analysis (see Kim & Mueller, 1978). The dimensionality was examined using both varimax (orthogonal) rotation and promax (oblique) rotation, both with the Kaiser’s extraction criterion (i.e., eigenvalue ≥1). Confirmatory Factor Analysis. Two confirmatory factor analyses (CFAs) were performed to investigate the model fit of the data. The first model tested the model fit of the eight-dimensional model previously found in the principal component analyses, and the second model further tested whether the eight-factor communication instrument was a homogeneous measure of organizational communication with different facets. The latter confirmatory factor analysis with a second-order common latent
370
International Journal of Business Communication 51(4)
dimension of organizational communication was therefore conducted. The analyses were performed using the AMOS program 18.0. In comparing the fit of the factor structure, the following fit indices were reported: chi-square, degrees of freedom, root mean square of approximation (RMSEA), goodness-of-fit index (GFI), and comparative-fit index (CFI). Reliability of the resulting dimensions is reported by Cronbach’s alpha (α) coefficient. The analyses were run separately in all three samples to test whether a robust dimensional solution had been achieved: that is, whether it could be reproduced in all three samples, based on respondents from different organizations and different occupations. Correlational analyses were examined by means of Pearson’s r correlation coefficient. Regression analyses (multiple linear regressions) were performed to examine how the communication components predicted employee assessment of communication. The Statistical Package for the Social Sciences (SPSS) V18 was used to analyze the data.
Results Results From the Pilot Study The data from the pilot study were organized in five categories, based on the analyses. The first category, social contact, included three dimensions, namely contact with leadership, disruptive relations, and coworker contact. The interviewees emphasized the important relationship and interaction with both colleagues and leaders during a change process, including issues such as regular contact, trust, and justice, and items within this area of communication were therefore developed. The second category included the topic of contact with central leadership. This category was included because of the importance of the impact of different levels of leadership to the better understanding of communication during a change process. Items related to central leadership (e.g., the top leadership, corporate leadership, or political leadership) were therefore developed. The third category concentrated on information. The processes of change were described as being dependent on high-quality information: the right amount of information given at the right time. The fourth category focused on influence and included both perceived influence and perceived union assistance. Influence and participation were mentioned as some of the main factors for successful implementation of change, especially related to overcoming resistance in the change process. Items related to these issues were therefore included in the instrument. The fifth category emphasized barriers to improvement, a topic that was mentioned in each of the prior categories. Therefore, items related to barriers in each of these themes were included in this category.
Establishing the Factor Structure The first step in validation involved determining dimensionality and reliability of the items and dimensions. The objective was to establish a robust dimension structure that could be reproduced in different samples and settings. Hence, the main criterion for dimensionality and reliability for the communication components was a structure that was reproduced in each of the three separate samples.
Christensen
371
Communication Components Dimensionality was examined by means of explorative principal component analyses with varimax and promax rotation and Kaiser’s extraction. Social Contact. The 36 items measuring social contact were analyzed with explorative principal component analysis. A robust three-dimension structure involving 15 of the items was found in all samples. However, some minor exceptions were related to Samples 2 and 3. The principle component analysis was performed with Kaiser’s extraction criteria revealing seven dimensions in the first principal component analysis done in Sample 3. Only three dimensions made theoretical sense, and in the initial analyses, 21 items were excluded from further analysis because they did not adhere to a clean structure and presented weak factor loadings. The resulting structure is presented in Table 2. In Samples 1 and 2, there were some remaining cross-loadings, but those were retained because the major sample demonstrated a clean structure. A further analysis revealed that removing the items cross-loading the structure did not improve anything but only moved problems to the other dimensions. Based on the content of the variables, the dimensions were named “contact with leadership,” “disruptive relations,” and “coworker contact.” The results of the PCAs are shown in Table 2. Only factor loadings at .35 or above are included; this is close to the conventional .40 criterion that is frequently applied as a cutoff (Ford, MacCallum, & Tait, 1986). Central Leadership. The five items measuring degree of central leadership were extracted as one dimension. Two items were omitted owing to weak factor loadings. Table 3 presents the results. An additional principal component analysis was performed including the items from the social contact category and the items of central leadership, because of similarities in the content of the two categories. The results revealed, however, that central leadership emerged as an individually distinct dimension. The category was therefore kept in the initial form. Information. The respondents were asked to assess 14 items regarding how messages are communicated (e.g., for quality, timing, and comprehensibility). The principle component analysis was performed with Kaiser’s extraction criteria revealing one dimension. Seven items were excluded from further analysis because they did not adhere to a clean structure (see Table 4). Influence. The principal component analysis revealed that 5 of the 14 items fell into a robust two-dimensional solution with acceptable reliability. The two dimensions were named perceived influence and perceived union assistance (see Table 5). Barriers to Improved Communication. The results of the items measuring barriers to communication improvement demonstrated that six items constitute a robust dimension (see Table 6).
372
International Journal of Business Communication 51(4)
Table 2. Dimensionality of the Items Measuring Social Contact (Factor Loadings). Sample 3 1. Contact with leadership I can be open in bringing up subject matters with my closest leader. The dialogue between my closest leader and me is good. There is tolerance/acceptance in my communication with my leader. My closest leader keeps an open door policy for the employees. My closest leader is available if I wish to bring up personal matters. My closest leader trusts the employees. My closest leader respects me. My closest leader and I have a common understanding of responsibility distribution. The manager takes the employees’ different needs into consideration. Explained variance Cronbach’s alpha 2. Disruptive relations Rumors are flourishing between the employees. My coworkers and I use a lot of time in discussing our frustrations concerning the job situation. We have personnel conflicts. Explained variance Cronbach’s alpha 3. Coworker contact The communication between my coworkers and me is good. I can bring up work related topics with my coworkers. I can bring up personal issues with my coworkers. Explained variance Cronbach’s alpha
1
2
3
Sample 2
Sample 1
1
2
3
.83 — —
.78
—
— .82 — —
.82 — —
.84
—
— .81 — —
.81 — —
.83
—
— .80 — —
.80 — —
.79
—
— .70 — —
.76 — —
.79
—
— .78 — —
.71 — — .70 — — .69 — —
.69 .61 .76
.42 — .59 .46 — — .38 .74 — — — — .72 — —
.66 — —
.51
.47 — .72 — —
48% .931
36% .930
1
2
3
50% .933
— .80 — −.36 −.71 — — .80 — — .78 — — −.45 .51 — .73 — — .76 — 10% .758
—
— — .79
—
—
.68 — — .77
— — .77
—
—
.70 — — .65
— — .68
—
.36 .57 — — .72
8% .687
−.78 — — .72 — 10% 9% .550 .747
7% .682
7% .638
Note. — = factor loadings below .35.
So far, the analyses have examined each dimension separately. Having established dimensionality and reliability of the communication components, the next step was to examine the bivariate correlations between the dimensions. See the correlation matrix in Table 7.
373
Christensen Table 3. Dimensionality of the Items Measuring Central Leadership (Factor Loadings). 1. Central Leadership The communication with the leadership is marked by openness. The leadership is personally involved in the employees’ on-the-job well-being. The leadership contacts the employees on a regular basis. Explained variance Cronbach’s alpha
Sample 3
Sample 2
Sample 1
.91
.87
.91
.90
.86
.92
.89
.84
.92
81% .884
74% .821
84% .905
Table 4. Dimensionality of the Items Measuring Information (Factor Loadings).
1. Information Generally there is an insufficient amount of information provided. The information uncovers the consequences of future proceedings. The information uncovers the causes of future proceedings. The information was provided too late. The message contained within the information is comprehensible. When the information is provided, there is not enough time for discussion/questions/feedback. The information provided is relevant for me. Explained variance Cronbach’s alpha
Sample 3
Sample 2
Sample 1
.77
.76
.77
.76
.78
.76
.75
.80
.74 .72
.69 .57
.75 .73
.70
.70
.69
.56 52% .842
.51 46% .805
.58 54% .852
78
Table 7 demonstrates that the overall tendency suggests associations between the communication components. Because some of the dimensions were highly correlated, it was found necessary to retest the dimensionality of the communication components by using principal component analyses with oblique rotation (promax). The results of the oblique rotation reproduced the initial dimensions that were found by use of orthogonal rotation (varimax).
Confirmatory Factor Analyses Having established robust dimensionalities across the three samples, the next step was to test these dimensions simultaneously as an overall model in a confirmatory factor analysis. This model included first-order factor only. This confirmatory factor analysis was performed only on Sample 3. The intentions of the first confirmatory
374
International Journal of Business Communication 51(4)
Table 5. Dimensionality of the Items Measuring Influence (Factor Loadings).
1. Perceived influence I am included in the evaluation process. My opinions are heard. There is not enough time and/or place for discussion and feedback. Explained variance Cronbach’s alpha 2. Perceived union assistance Unions provide complete information about my choices/rights. The unions bring my view/opinion forward. Explained variance Cronbach’s alpha
Sample 3
Sample 2
Sample 1
1
2
1
2
1
2
.90
—
.91
—
.86
—
.89 .80
— —
.87 .74
— —
.87 .80
— —
45% .837
43% .789
46% .836
—
.96
—
.94
—
.90
—
.95
—
.94
—
.91
37% .908
35% .864
35% .881
Note. — = factor loadings below .35.
Table 6. Dimensionality of the Items Measuring Barriers to Improved Communication (Factor Loadings). 1. Barriers to Improvement A lot remains unsaid because of negative consequences for those who introduce difficult topics. Certain individuals may hamper communication. There are easily too many rumors. Information content is poor in its quality Certain groups may hamper the flow of communication. Communication is too unstructured. Explained variance Cronbach’s alpha
Sample 3
Sample 2
Sample 1
.84
.75
.86
.83 .80 .76 .75
.79 .68 .76 .77
.85 .85 .78 .80
.68 61% .871
.64 54% .825
.63 64% .884
factor analysis was to test the eight-factor structure found in the previous principal component analyses. The results revealed a satisfactory model fit to the data, as indicated by the following fit indices’ values: χ²(566) = 1,907.44, p < .000, CFI = 0.92, GFI = 0.87, and RMSEA = 0.056. An RMSEA of 0.05 or less indicates a close model fit; 0.05 to 0.08 indicates a fair fit; and 0.08 to 0.10 indicates a mediocre fit. An RMSEA exceeding 0.10 reflects a poor fit to the observed data (Brown & Cudeck, 1993).
375
Christensen
Table 7. Correlation Matrix of the Communication Components (Pearson’s r Coefficient). 1 1. Contact with leadership 2. Disruptive relations 3. Coworker contact 4. Contact with central leadership 5. Information 6. Perceived influence 7. Perceived union assistance 8. Barriers to improved comm.
.51* −.52* .55* .57* .55* .10* −.54*
2
−.27* −43* −.538* −.42* −.01* .67*
3
.32* .37* .36* .16* −.35*
4
.59* .50* .01 −.47*
5
.54* .09* −.60*
6
7
.15* −.46*
−.05
* p < .01.
Furthermore, it was of interest to see whether the eight-factor communication instrument was a homogeneous measure of organizational communication with different facets. Therefore, a second confirmatory factor analysis with a second-order common latent dimension of organizational communication was conducted. The analysis was also conducted in Sample 3. The results revealed that the model had a satisfactory fit to the data according to the following values: χ²(586) = 2,105.84, p < .000, CFI = 0.91, GFI = 0.86, and RMSEA = 0.058. See Table 8 for the confirmatory factor analysis with the second-order latent dimension. The results revealed that seven of the eight factors loaded strongly on the latent second-order dimension of organizational communication. The union assistance dimension exhibited a low factor loading with a loading of only .10 (see Figure 1). The goodness-of-fit indices of both models are summarized in Table 9. Consequently, the dimensions were used for further analyses.
Predictive Validity Having established the structure and dimensionality of the communication factors, the next step was to examine the predictive validity. This involved examining how the communication factors predicted the “employees’ assessments of communication,” an item measuring the employees’ evaluation of communication (Is there sufficient communication in your department?). This was examined in Sample 1, and analyses were performed in the total sample and three separate groups: Group 1 consisted of employees experiencing no reorganization; Group 2 included employees going through reorganization; Group 3 contained employees who had recently been involved in completed reorganization. These analyses provided the opportunity to differentiate between groups within specific stages of a change process. A regression model was then specified by entering all of the communication factors as predictors for the “employees’ assessments of communication.” This regression model was tested in the total sample and each group (see Table 10).
376
International Journal of Business Communication 51(4)
Table 8. Results of a Confirmatory Factor Analysis With a Second-Order Common Latent Dimension (Sample 3). Estimate Contact with leadership (ξ 1) X1: I can be open in bringing up subject matters with my closest leader. X2: The dialogue between my closest leader and me is good. X3: There is tolerance/acceptance in my communication with my leader. X4: My closest leader keeps an open-door policy for the employees. X5: My closest leader is available if I wish to bring up personal matters. X6: My closest leader trusts the employees. X7: My closest leader respects me. X8: My closest leader and I have a common understanding of responsibility distribution. X9: The manager takes the employees’ different needs into consideration. Disruptive relations (ξ 2) X1: Rumors are flourishing between the employees. X2: My coworkers and I use a lot of time in discussing our frustrations concerning the job situation. X3: We have personnel conflicts. Coworker contact (ξ 3) X1: The communication between my coworkers and me is good. X2: I can bring up work related topics with my coworkers. X3: I can bring up personal issues with my coworkers. Central leadership (ξ 4) X1: The communication with the leadership is marked by openness. X2: The leadership is personally involved in the employees’ on-thejob well-being. X3: The leadership contacts the employees on a regular basis. Information (ξ 5) X1: Generally there is an insufficient amount of information provided. X2: The information uncovers the consequences of future proceedings. X3: The information uncovers the causes of future proceedings. X4: The information was provided too late. X5: The message contained within the information is comprehensible. X6: When the information is provided, there is not enough time for discussion/questions/feedback. X7: The information provided is relevant for me.
SE
1.00
β .84
0.92 1.04
0.03 0.03
.81 .88
0.99
0.04
.83
0.92
0.04
.71
0.72 0.76 0.82
0.03 0.03 0.03
.74 .73 .76
0.79
0.04
.71
1.00 1.31
0.08
.64 .81
1.14
0.08
.71
1.00 1.09 1.23
0.08 0.10
.73 .70 .60
1.00 1.00
0.03
.84 .89
1.03
0.04
.84
1.00
.71
1.17
0.06
.75
1.00 −0.78 −0.93
0.06 0.05 0.05
.67 −.65 −.70
−0.87
0.05
−.68
−0.55
0.04
−.48
(continued)
377
Christensen Table 8. (continued)
Perceived influence (ξ 6) X1: My opinions are heard. X2: I am included in the evaluation process. X3: There is not enough time and/or place for discussion and feedback. Perceived union assistance (ξ 7) X1: Unions provide complete information about my choices/rights. X2: The unions bring my view/opinion further on. Barriers to improvement (ξ 8) X1: A lot remains unsaid because of negative consequences for those who introduce difficult topics. X2: Certain individuals may hamper communication. X3: There are easily too many rumors. X4: Information content is poor in its quality. X5: Certain groups may hamper the flow of communication. X6: Communication is too unstructured.
β
Estimate
SE
1.00 1.07 0.87
0.04 0.04
.87 .69 .83
1.00 0.96
0.24
.92 .89
1.00 0.88 0.97 0.71 0.74 0.62
.82 0.04 0.04 0.03 0.04 0.04
.77 .77 .71 .68 .60
Note. All p values are