lUllllllllflllllllllllllllllllllllllll 10684 0873 l is???” Q This is to certify that the thesis entitled The Influence of Social Information on Employee Anxiety About Organizational Change presented by Katherine 1. Miller .1 has been accepted towards fulfillment of the requirements for M.A. degree in Comunication' Macaw / 783 0-7639 MS U is an Affirmative Action/Equal Opportunity Institution MSU RETURNING MATERIALS: Place in book drop to remove this checkout from LIBRARIES your record. FINES will e be charged if book is returned after the date stamped below. 148 A140 We; {L be”? 'r-r'w '2?“ . 5'2 K27h FEB 061952, E9), K023 THE INFLUENCE OF SOCIAL INFORMATION ON EMPLOYEE ANXIETY ABOUT ORGANIZATIONAL CHANGE By Katherine I. Miller A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Communication 1983 l I ' . . ,I ' ,- 1" ‘.‘ \.‘ ABSTRACT THE INFLUENCE OF SOCIAL INFORMATION ON EMPLOYEE ANXIETY ABOUT ORGANIZATIONAL CHANGE By Katherine 1. Miller Studies of job attitudes have traditionally been conducted in terms of the correspondence between individual needs and objective job characteris- tics. A recently developed theory, however, suggests that job attitudes may be a function of social information received (Salancik & Pfeffer, 1978). This investigation used social information processing theory as . the basis for a study investigating antecedents to employee anxiety about a move to an open office environment. The structural equation model developed from social information pro- cessing theory proved to be a good fit to the data, and a revised version of the model provided an even better accounting for the variance in the data. Anxiety about organizational change was determined by social information, individual needs, and job characteristics, with need for privacy having the largest impact on anxiety. The model is discussed in terms of its support for information pro- cessing theory, its individual significant linkages, and the implications for need satisfaction models of job attitudes and othervresearch on out- comes in organizations. ACKNOWLEDGMENTS It has been said that good friends and good help are hard to find. That was far from true for me while working on this thesis; both good friends and good help were in plentiful supply. I'd first like to thank my advisor, Peter Monge, for having total confidence in me and working me and encouraging me in such a way that I developed confidence in my- self. Vince Farace, Judee Burgoon, and Gerald Miller, in serving on my committee, also provided valuable insight and encouragement. They were good help, and they were always available. My good friends were also a constant source of ideas and an occassional relief from having to deal with those ideas. Three friends in particular deserve special thanks: Milt Shatzer for laughing and talking, Julia Crystler for caring and confiding, and Jim Stiff for arguing and loving._ Finally, I want to thank my parents for bringing me up in a home where loving and learning flourished and for keeping me aware that that home is always available. ii TABLE OF CONTENTS LIST OF TABLES ............................ LIST OF FIGURES ........................... Chapter 1. Chapter II. RELATED LITERATURE AND STATEMENT OF THE PROBLEM ......................... Theories of Job Attitudes ................ Need satisfaction theories .............. Social information processing theory ......... Empirical tests of social information processing theory ........................ Determinants of Occupational Stress ........... Environmental stressors ................ Individual characteristics .............. Stress as person-role fit ............... Social Information Processing Approach to Job Anxiety METHODOLOGY AND PROCEDURES ............... Subjects ......................... . Research Design and Procedures .............. Instrumentation ..................... Chapter III. Information ...................... Individual needs ................... Job characteristics .................. Anxiety ........................ Information helpfulness ....... - ......... Analysis ......................... RESULTS ........................ Preliminary Analyses ................... Measurement Models .................... Theoretical Structural Equation Model .......... Model Modifications ................... TABLE OF CONTENTS (Continued) Page Chapter IV. DISCUSSION ....................... 6l Social Information Processing Theory and the Obtained Model ..................... 61 Determinants of Anxiety: Relative Impact ...... 63 Determinants of Needs . . .. ............. 64 Determinants of Information Helpfulness ....... 65 Impact of Interdependence Needs .......... _. . 66 Impact of Manipulation ................. 67 Limitations ...................... 68 Managerial Implications ........... 4 ..... 69 Directions for Future Research ............. 71 FOOTNOTES ............................... 74 REFERENCE NOTES ........................... 75 REFERENCES ................... ' .......... 76 APPENDICES ........................... .. . 81 Appendix A: Questionnaire Items ................ 81 Appendix B: Informational Manipulations ............ 83 iv Table 10 ll l2 l3 I4 715 16 LIST OF TABLES Items and factor loadings for previous information factor . . . 37 Deviations of observed from expected correlations for internal consistency and parallelism. Previous information . . 38 Items and factor loadings for information helpfulness factor. . 39 Deviations of observed from expected correlations for internal consistency and parallelism. Information helpfulness ..... 40 Item and factor loadings for privacy needs factor ....... 41 Deviations of observed from expected correlations for internal cosistency and parallelism. Privacy needs ........... 42 Item and factor loadings for interdependence needs factor . . . 43 Deviations of observed from expected correlations for internal consistency and parallelism. Interdependence needs ...... 44 Item and factor loadings for anxiety factor .......... 45 Deviations of observed from expected correlations for internal consistency and paralleilism. Anxiety ............ 46 Item and factor loadings for revised anXiety factor ...... 48 Deviations of observed from expected correlations for parallelism. Revised anxiety factor ............. 49 Correlation matrix among theoretical variables ........ 51 Structural equations for theoretical model .......... 53 Model comparisons ....................... 57 Structural equations for revised model ............ 59 LIST OF FIGURES Elam: BEBE 1 Theoretical model of employee anxiety ............ 19 2 Combined theoretical and measurement model of employee anxiety ........................... 3O 3 Theoretical model with structural coefficients ....... 52 4 Revised model with structural coefficients ......... 58 vi CHAPTER I RELATED LITERATURE AND STATEMENT OF THE PROBLEM The importance of a strong theoretical framework as the basis of research has long been accepted in social science. Researchers in the field of com- munication clearly support the goal of formulating theory to explain human communication behavior. However, in the field of organizational communication, this goal has rarely been reached, as attested to by recent review pieces (Jablin, I980; Redding, 1979; Richetto, 1977). Nowhere is the preponderance of atheoretical work seen more clearly than in research on the impact of communication in influencing job attitudes and behaviors. Communication vari- ables such as network participation, openness, social relationships with coworkers, and participation in decision-making have been related to a large variety of of attitudinal and behavioral outcomes such as job satisfaction, job involvement, organizational commitment, stress, and productivity. However, these empirical studies have rarely been supported by a theoretical framework linking communicative activities with individual and organizational outcomes. Although it has thus far been applied Sparingly by organizational com- munication scholars, one theoretical framework which makes the link between communication and individual and organizational outcomes is the social infor- mation processing theory of job attitudes (Salancik & Pfeffer, l978). This theory, a response to weaknesses of the traditional need-satisfaction models, proposes that job attitudes are a function of the communicative activities of l 2 employees. In the next section, the need satisfaction model will be described and critiqued. Then the social information processing theory will be pre- sented, and research which supports it will be discussed. Social information processing theory will then be used to explore the link between organizational change and work characteristics in the anxiety and stress felt by employees. Theories of Job Attitudes Salancik and Pfeffer (l978) proposed the social information processing theory of job attitudes as a response to prevailing theories of job attitudes in industrial and organizational psychology. Thus, need satisfaction theories of job attitudes will be briefly discussed and criticized to provide a con- text for social information processing theory. Need satisfaction theories. Need theories of job attitudes are based on three propositions. These are that (1) persons have basic, stable, rela- tively unchanging and identifiable attributes, including needs, (2) jobs have stable, identifiable sets of characteristics that are relevant to needs, and (3) job attitudes result from the correspondence between the needs of the individual and the characteristics of the job. Although there are many varia- tions of need theories, one which has gained considerable prominence was for- mulated by Hackman and Lawler (l97l). They suggested that there are four core dimensions of job characteristics -- variety, autonomy, task identity, and feedback, and that these dimensions serve as determinants of job satisfaction and motivation, with the relationship moderated by an individual‘s level of higher order needs such as accomplishment and personal growth. Other varia- tions of the need satisfaction model have been presented by Aldefer (l972), Newman (l975), and Porter (l962, 1963). The need satisfaction models of job attitudes have been criticized on conceptual, empirical, and methodological grounds (Salancik & Pfeffer, l977). First, need satisfaction models implicitly assume causality from job charac- teristics tg_attitudes without considering that a satisfied person can impute characteristics to a job. Need models are also inconsistent in the specification of the functional form of the job characteristic/attitude rela- tionship, and there is the assumption of a direct link between attitude and behavior. Salancik and Pfeffer's largest criticism of need models, however, involves the conceptualization of job characteristics and human needs. Al- though need theorists conceed that need strength often changes, needs are con- ceptualized as stable characteristics of persons. However, even highly-popular theories of human needs, such as Maslow‘s hierarchy of needs (Maslow, l943) or Aldefer's (l972) typology of existence, relatedness and growth needs, have not received much empirical support. Thus, Salancik and Pfeffer argue that, ". . . the fact that the concept of need is ambiguous on the points of the origins of needs makes the possibility of empirical refutation remote and the concept, in its present stage of development, of limited utility" (l977, p. 443). The concept of Job characteristics is also criticized, as the pro- cess of defining job characteristics has been largely incumbent on the researcher and the choice of job dimensions used may influence the observa- tions of the researcher. The fact that job characterization is a process with information about the observer, also implies that job characterization is a process with infor- mation about the worker and his social context. Characteristics are imputed to jobs, and this imputation may be a consequence of a social process. (Salancik & Pfeffer, 1977, p. 446) The need satisfaction model is also criticized on methodological and empirical grounds. First, the survey methods typically used in studies of job attitudes may be plagued by problems of consistency effects and priming effects. Respondents asked to describe both the characteristics of their job and their attitudes may try to respond to attitude items in a manner consistent with responses to questions about characteristics (consistency effect). Similarly, the mere mention of a job characteristic or attitude in a question could make the attitude or characteristic more salient to the employee and influence responses (priming effect). Also, despite the possi- ble artifactual results of consistency and priming, relatively small effect sizes have been obtained for the relationship between job characteristics and absenteeism, productivity, and attitudes such as job satisfaction and organi- zational commitment. Social information proceSsing-theory. Salancik and Pfeffer (1978) followed up their criticism of the need satisfaction models (Salancik and Pfeffer, 1977) with a theory of job attitudes which is largely communicative in nature. Its conceptual roots can be seen in theories of social construction of reality (Berger & Luckmann, 1967; Neick, 1979), self-perception (Bem, l972), conformity (Kiesler & Kiesler, 1979) and attribution (Kelly, 1967). The theory states that attitudes and needs of an individual have three determinants: (1) the job or task environment characteristics, (2) social information, and (3) the individualis own behaviors. Needs in this theory, like attitudes, are conceptualized as outcomes produced by a person rather than inherent pro- perties of individuals. Characteristics of the job are included in the model 'as determinants of attitudes, but Salancik and Pfeffer state that "individuals develop attitude or need statements as a function of the information available to them at the time they express the attitude or need. The form and content of that expression are affected by the request for the attitude, the purpose for which it is requested, and any other fact that might affect the relative saliency of information relevant to the person deriving the attitude" (1978, p. 226). Thus, the expression of an attitude will not be based on the correspondence between needs and job characteristics, but will be based on all available information -- specifically, the characteristics of the job, the information received, and the individual's own behavior. Salancik and Pfeffer describe the mechanisms through which behavior and social information influence the formation of job attitudes. In the tradi- tion of self-perception theory (Bem, 1972), social information processing theory states that individuals formulate attitudes based on observations of their own behavior. Commitment to an attitude comes from the binding of the individual to behavioral acts which are explicit, irrevocable, of free voli- 'tion, and public (Salancik, 1977). Thus, the attitude-behavior relationship is not unidirectional, as proposed by need-satisfaction theorists, but mutually causal, where behaviors and attitudes influence each other. Salancik and Pfeffer also delineate four ways in which social informa- tion received can influence attitudes. First, evaluative statements received can serve as an aid in interpreting complex cues or can suggest ways in which an employee can conform to the work group. Second, information received can structure attentional processes by making aspects of the environment more or less salient._ For example, receiving cues about information from supervisors could make feedback a salient feature of the environment to an employee. Third, beyond just focusing attention on various aspects of the environment, social information can aid in the interpretation of environmental cues. This is especially true when cues from the environment are equivocal. In these cases, the social definition given to an event will probably prevail. Finally, social information can influence attitudes by aiding workers in the interpre- tation of needs. According to Salancik and Pfeffer, ". , . the comment that a job does not give a person a chance to think implies not only that the job has a certain feature but that the presence or absence of that feature should be important to the person . . . in other words, people learn what their needs, values and requirements should be in part from their interactions with others" (1978, p. 230). Empirical tests of social information processing theory. The social information processing theory has received several direct empirical tests in the area of attitudes toward job characteristics. The theory has also been used as an explanatory tool in the areas of product evaluation (Pincus & Waters, 1977) and organizational commitment (Eisenberg, Monge, & Miller, 1983), but only the direct tests of the theory (O‘Reilly & Caldwell, 1979; White & Mitchell, 1979) will be discussed in detail. Both direct tests of the theory used experimental designs in which stu- dents were placed in ad hoc work groups and task enrichment and social infor- mation were manipulated. White and Mitchell (1979) manipulated job enrich- ment through verbal instructions and task design, and O‘Reilly and Caldwell (1979) manipulated enrichment simply through task design. In both studies, two tasks were designed to differ on the levels of variety, autonomy, iden- tity, significance, and feedback, task characteristics indicated by Hackman and Oldham (1976) as important characteristics influencing job attitudes. In the Nhite and Mitchell study, social information was manipulated through the placement of a confederate in the work groups who made either positive or negative comments about the task (e.g., "this job becomes more meaningful as you do it"). In the O‘Reilly and Caldwell study, social information was manipulated through task description and evaluative statements about the task ostensibly made by past students. The results of both studies indicate that social information plays an important role in shaping attitudes about task characteristics. White and Mitchell found a more substantial effect for job factors, but social cues had a significant impaCt on perceptions of skill variety, job satisfaction, and productivity. O'Reilly and Caldwell found that information was a more potent influence on attitudes than task characteristics, with information significantly influencing skill variety, autonomy, task significance, overall satisfaction, pay satisfaction, and growth satisfaction. Both studies point to the importance of social cues in incongruous job settings in which infor- mation contradicted job characteristics. In these cases, information provided by coworkers played an important role in shaping attitudes. According to O'Reilly and Caldwell, "task dimensions may also provide cues that are ambi- ' guous and difficult for the job holder to interpret . . . Ambiguous cues require a social context for interpretation. This context may be provided by informational cues from others" (1979, p. 163). Although these studies are relatively persuasive in support of the social information processing theory of job attitudes, they have several major weak- ’nesses. First, the experimental nature of the studies may have decreased the level of external validity in terms of both tasks and coworkers. Although subjects in both studies were paid for their work, the short duration of the tasks (approximately one half day) certainly did not resemble work conditions normally encountered in an organization. It seems unlikely that formation of attitudes about a two-hour task is the same process as the formation of attitudes about a permanent eight-hour a day job. Similarly, operationali- .zations of social information do not accurately reflect the effect of com- ‘munication in the workplace. Cues received from a stranger should have a much different impact than those received from coworkers or supervisors in the organization. These studies also suffered from the implicit assumptions made by utilizing task characteristics developed by Hackman and Oldham (1976). Using these task characteristics assumed that the most relevant task char- acteristics to study are variety, autonomy, task identity and feedback. One of Salancik and Pfeffer's original criticisms of the need satisfaction model was the somewhat arbitrary labeling of job dimensions which I'should" be important in influencing perceptions. Yet, in testing the social infor- mation processing theory, O'Reilly and Caldwell and White and Mitchell both fall back on the list of characteristics typically used in the need satis- faction models. Thus, though these studies are internally valid and provide strong support for social information processing theory, it may not be appropriate to generalize these findings to the attitudes of employees working day to day in the same organization. There are four major ways in which improve- ments could be made in testing the theory. First, organizational employees should be tested, rather than students in ad hoc work groups. Second, attitudes about real tasks which are part of the job should be examined. Third, the social information considered should come from coworkers who are part of the work group or, possibly, another organizational source. Finally, an effort should be made to consider characteristics of the job which are relevant to employees, rather than those developed in past need satisfaction models. One set of job attitudes which has received a great deal of attention in industrial and organizational psychology is stress. Like satisfaction, productivity, commitment, and other outcome variables of interest to organi- zational researchers, stress has typically'been studied as a response to environmental characteristics or as a function of personality type. As with empirical work on the need satisfaction model, relationships found between objective job characteristics and stress and anxiety have generally been weak. The research on job characteristics discussed above indicates that a shift in stress research to considering social information as an antecedent to occupational stress and anxiety would be appropriate. Thus, the study discussed in this paper examined the effect of different types of information about a potentially stressful organizational event on the degree of anxiety experienced by employees. In the next section, theory and research on organizational stress and anxiety will be explored. First, the terms stress, anxiety, stressor, and threat will be defined. Next, individual and environmental antecedents to stress and anxiety will be discussed, and the concept of person-role fit will be presented. Finally, social information processing theory will be applied to the area of occupational stress and anxiety, and hypotheses about the role of communication in influencing anxiety about a coming event will be presented. Determinants of Occupational Stress Theory and research in the area of occupational stress has been plagued by inconsistency of definition. This is partially due to the eclectic nature of the field. Researchers in medicine, psychiatry, social psychology, sociology, cultural anthropology, and management have often approached the problem of stress in different manners, and there is little sharing among disciplines. For example, physiological researchers have generally adopted a very broad definition of stress. Selye (1976, 1980) defines biolo- gical stress as a nonspecific bodily response to any demand. Psychological researchers, in contrast, adopt a more actively cognitive definition of stress. Cofer and Appley (1964) define stress as "a state where the well-being (or integrity) of an individual is endangered and he must devote all of his energies to its protection" (p. 463). )0 Researchers within the specific field of occupational stress have also had little luck in reaching consensus on a definition of stress. Some theorists see stress as a reaction to stressors in the environment (Beehr & Newman, 1978; Cooper & Marshall, 1976) and some see stress as an inter- vening variable -- a psychological state which may or may not lead to stress symptoms (Schuler, 1980). Some researchers, however, do not define stress as a reaction to environmental conditions, but rather define stress as the environmental factor which causes strain in an individual (Caplan, Cobb, French, Van Harrison, & Pinneau, 1980). For example, Caplan, et al. define stress as "any characteristic of the job environment which pose a threat to the individual," and strain as "any deviation from normal responses in the person." Other researchers go further in making a distinction between stress and threat (Cummings & Cooper, 1979; Lazarus, 1966) where stresses are current environmental conditions and threats are anticipated stresses. This paper will combine several of these perspectives in the use of four key terms: stress, stressor, threat, and anxiety. Stress will be viewed as an individual's reaction to aspects of the environment (stressors) which cause deviation from normal responses in that person. A person may have a variety of psychological, physiological, and behavioral.responses to stress in the workplace such as tenseness, job dissatisfaction, high blood pressure, susceptibility to heart disease, increased heart rate, in- creased breathing rate, headaches, ulcers, increased or decreased task per- formance, smoking, and extensive use of coffee or alcohol. When a particular coming event serves as a stressor for an individual, that event will be called a threat and the reaction to that threat called anxiety. Thus, a threat is a special kind of stressor -- an anticipated event rather than a 11 part of the current environment, and anxiety is a special kind of stress -- an event-specific reaction to a threat. Research on threats and anxiety in the workplace has been virtually nonexistent. Literature on organizational change is extensive, but generally deals with steps in the change process and strategies which will make organi- zational change easier. Little research has been conducted to investigate the qualities of individual events which will lead to anxiety in employees. There are two exceptions to this generalization. A study of decision-specific stress (Holbrook & Ryan, 1982) indicated that decisions involving infor- mation overload or conflict were particularly stressful. Naismith (1975) found that organizational change did not influence levels of general stress among managers. However, neither of these studies addressed the issue of anxiety felt toward a specific coming event. In contrast to the area of threat and anxiety, however, there have been a great many studies of general stress in the workplace. The next three sections of this paper will briefly review the literature on workplace antecedents to stress, individual predispositions to occupational stress, and the concept of person-role fit. More comprehensive reviews of the occupational stress literature can be found in Cooper and Marshall (1976), McGrath (1976), and Schuler (1980). Environmental stressors. Most of the research relating stress to work- place factors has concentrated on the load of information or work an employee has, and the way a worker defines his/her role within the organization. Other variables studied have included responsibility, participation in decision-making, and social relationships at work. A distinction has been made in the stress literature between quanti- tative and qualitative work load (French & Caplan, 1973). Quantitative 12 overload refers to having "too much" work and qualitative overload is having to deal with work that is "too difficult." Both quantitative and qualitative overload have been empirically linked to stress among workers. It has been suggested that qualitative and quantitative underload are also stressful, but no research has been conducted to explore this idea. Quantitative overload has been shown to have a positive relationship with a variety of physiological, psychological, and behavioral stress symp- toms. In a retrospective study, Breslow and Buell (1960) found that the number of hours worked was positively related to death from coronary diseases, and French and Caplan (1973) found that overload was positively related to cigarette smoking. Margolis, Kroes, and Quinn (1974) found that work over- load was directly related to escapist drinking, absenteeism, low motivation to work, lowered self esteem, and an absence of suggestions to employers. However, the correlation coefficients obtained for these relationships were quite small, ranging from .O6 to .26. Thus, as Cooper and Marshall (1976) have indicated, Quantitative overload is obviously a potential source of occupational stress, as other studies (Porter & Lawler, 1965; Quinn, Seashore, & Mangione, 1971) also indicate, but on the evidence available it may not, by itself, be a main factor in occupational ill health. (p. 15) The relationship between qualitative overload and stress appears to be occupation-specific. French, Tupper, and Mueller (1965) found that quali- tative overload was significantly linked to low self-esteem in university professors but not in university administrators. Qualitative overload and stress symptoms have also been positively linked for medical students (Dreyfuss & Czackes, 1959) and accountants (Friedman, Rosenman, & Carroll, 1958). 13 Role conflict and role ambiguity have also been examined as potential antecedents to stress. Role conflict can be defined as the simultaneous occurence of two (or more) sets of pressure such that compliance with one would make compliance with the.other more difficult, and role ambiguity can be defined as a lack of knowledge or inconsistent knowledge about role expec- tations, activities, and consequences (Kahn, Wolfe, Quinn, Snoek, & Rosenthal, 1964). In a large-scale study involving questionnaires and intensive inter- views, Kahn, et al. found that greater degrees of role conflict and role ambi- guity lead to higher levels of stress in workers. This finding has been supported by French and Caplan (1973), Margolis, Kroes, and Quinn (1974), and Shirom, Eden, Silberwasser, and Kellerman (1973). A third job characteristic which has been linked to work stress is responsibility. Many researchers have differentiated between responsibility for people and responsibility for things, hypothesizing that a high level of responsibility for people will lead to more stress symptoms than respon- sibility for things. This hypothesis was supported in several studies (French & Caplan, 1973; Pincherle, 1972; Wardwell, Hyman, & Bahnson, 1964). . Stress has also been linked to social relationships at work and parti- cipation in decision-making.‘ Empirical investigations have found that poor relationships with fellow workers are positively related to stress symptoms (Buck, 1972; French & Caplan, 1973; Kahn, et a1., 1964). Also, Buck (1972) and French and Caplan (1973) found that workers who participated in organi- zational decision-making felt less pressure and more job satisfaction. The research just reviewed indicates that there are five areas of job characteristics which influence stress in workers -- work load, role defi- nition, responsibility, social interaction, and participation in decision- making. Heavy research emphasis has been placed on information overload, l4 role conflict, and role ambiguity. Unfortuately, while these characteristics have received fairly consistent support from research, the strength of rela- tionship between environmental factors and stress symptoms has generally been fairly weak. Individual characteristics. Although few stress researchers would suggest that the characteristics of an individual alone can "cause" stress, many studies have been conducted to determine if individuals with specific psychological characteristics are more susceptible to stressful organizational conditions than others. A general predisposition to anxiety has been ex- plored in two areas of stress research. First, Endler and his colleagues (Endler & Okada, 1975; Endler & Shedletsky, 1973) have researched the concept of trait anxiety (A-Trait) as "a relatively unfluctuating condition of the individual, a latent disposition to respond with state anxiety (A-State). under stress" (Endler & Okada, 1975, p. 319). Endler and Shedletsky (1973), Hodges (1968), and Sales (1978) have found that high A-trait individuals will show more physiological and psychological stress symptoms over a variety of situations, though this finding was not supported by Endler and Okada. In the organizational context, general predisposition to stress has been considered in research on Type A and Type B behavior. Persons exhi- biting Type A behavior patterns tend to be more competitive and aggressive than Type 8 persons and are more likely to develop fatal and nonfatal coronary heart disease (Jenkins, Rosenman, & Zyanski, 1974; Jenkins, 1976; Rosenman, Friedman, & Strauss, 1964, 1966). Although originally treated as only a behavioral pattern, Jenkins (1971) states that it is also "a deeply ingrained, enduring trait." Stress as person-role fit. A final line of research in the area of occupational stress is that of person-role fit.) This view, originally 15 suggested by Kahn, et a1. (1964), has been given extensive recent attention by Caplan, Cobb, French, Van Harrison, and Pinneau (1980) and by French, Caplan, and Van Harrison (1982). These theorists suggest that it is not specific factors of the job or role which will lead to stress in individuals, but rather the discrepancy between qualities of the job and a person‘s tolerance of these qualities. In particular, a great deal of attention has been given to the degree of person-role fit in the areas of role conflict, role ambiguity, and information load. It is suggested that different em- ployees find different levels of conflict, ambiguity, and load satisfactory, and only when levels of these variables on the job exceed personal tolerance will stress be induced. .Studies utilizing the person-role fit perspective have generally concluded that fit is a better predictor of job stress than objective characteristics of the job alone. However, correlation coefficients relating fit with stress symptoms still generally range from .10 to .30, and insignificant results were found for many dimensions of organizational life which were hypothesized to be stressful. (See Caplan, et al., 1980, and French, et al., 1982, for a complete review of research findings on person-role fit and stress.) The person-role fit approach to job stress is clearly similar to the need satisfaction models of job characteristics. Both approaches assume that people have relatively stable underlying needs. Both also assume that the job has relatively stable characteristics. In the need satisfaction models the characteristics of interest have typically been feedback, autonomy, task significance, and task variety. Person-role fit theorists have concen- trated on role conflict, role ambiguity, and work load. In both approaches, though, psychological responses (satisfaction in need models or stress in person-role fit models) are proposed to be a function of the correspondence 16 between the relatively stable characteristics of the job and the individual. Many of the criticisms made of the need satisfaction model can also be levied against the person-role fit approach to job stress. First, it is unlikely that individual needs and job characteristics are stable phenomena. This is especially true of the job characteristics of interest to stress researchers. Clearly, the levels of work load, conflict, and ambiguity fluctuate as a response to changing responsibilities and changes in the environment. Yet person-role fit theorists study these characteristics as stable aspects of the job. Also, as with need satisfaction models, the unidirectional causal flow from characteristics to attitudes seems unwar- ranted, as employees can often impute ambiguity or conflict to their jobs. Methodologically, research on person-role fit has been stronger than research based on need satisfaction models. Employees have not typically been asked to define both objective job characteristics and attitudes about those characteristics. Rather, job features have generally been defined by coworkers of the “focal person.” Hence, this research has probably not suffered from "consistency“ and “priming" effects like need satisfaction models. However, person-role fit approaches have still had methodological problems stemming from initial assumptions about what types of job charac- teristics will be stressful. Almost all research has been based on the characteristics of work load, role conflict, role ambiguity, and, to a lesser extent, level of responsibility. Little effort has been made to explore sources of stress in the job beyond the job characteristics which were ini- tially identified in the literature. Finally, like need satisfaction models, the person-role fit approach has typically received weak, though consistent, empirical support. The correspondence between individual needs and role characteristics typically l7 accounts for very little variance in the prediction of occupational stress. In summary, then, the study of occupational stress and anxiety has been dominated by approaches which emphasize objective characteristics of jobs, individual predispositions, or the interaction of these antecedents. Unfor- tunately, this line of research has often been ineffective in identifying the processes in organizations which lead to the development of stress and anxiety in employees. In an effort to deal with the shortcomings of current approaches to job stress, the next section of this paper will present an application of Salancik and Pfeffer’s (1978) social information processing theory to the area of occupational anxiety. Social Information Processing Approach to Job Anxiety Salancik and Pfeffer's (1978) social information processing theory states that job attitudes are a function of three forces within the work- place. These are (1) characteristics of the job, (2) the individual, and (3) social information about the job. This theory has traditionally been applied to the areas of job satisfaction, motivation, and organizational commitment, but the extension to the area of occupational stress is straight- forward. That is, it is not only the characteristics of the job or the correspondence of these characteristics to an individual’s needs which lead to stress, but also information received from coworkers or other organiza- tional sources. Expanding on Salancik and Pfeffer, social information could influence the development of stress in several ways: (1) By aiding in the interpretation of needs for a worker ("Taking a break twice a day shouldn't be necessary in a job like yours"). (2) By highlighting characteristics of the environment or work role ("I've never seen a person who has to deal with so many conflicting requests from management"). 18 (3) By aiding in interpretation and reaction to environmental or role cues ("I'd be nervous as a cat if I had to work in the middle of all this noise and confusion"). Social information of these three types will serve to alter perceptions about what the job is like, what a person needs, and how a person should react. The information, then, serves as an antecendent which, together with objective job characteristics and the individual, will influence the level I of stress felt on the job. This framework can also be extended to the explanation of the anxiety employees feel about a coming event (threat). Again, social information serves as an influence in the interpretation of needs, environmental characteristics, and proper reaction, but the information is about a coming event rather than current work activities. In fact, social information may well play a particularly crucial role in the formation of anxiety because the information received cannot be compared to the experiences of the indivi- dual. The following theoretical model (see Figure 1) presents the key compo- nents and interrelationships in the development of anxiety from this perspec- tive. It should be noted that this model does not fully incorporate all of the relationships suggested in Salancik and Pfeffer‘s social information processing theory. In particular, the role of an individual's own behavior in influ- encing job attitudes is not included. The issue of mutual causality between behavior and attitude is an important one which should be further explored in organizational research. This study, however, concentrates on the addi- tion of different types of social information as a determinant of the job attitude of anxiety. Thus, the proposed model will allow for a valuable partial test of social information processing theory in a new conceptual Figure 1. Theoretical model of employee anxiety. Legend for the model. X1 = social information >< ll 2 job characteristics .< ll 1 perceived helpfulness of information employee needs -< ll 3 employee anxiety 20 area, employee anxiety, and in a real organizational context. This model has four key components: (1) social information and the perceived helpfulness of the information, (2) individual needs, (3) job/role characteristics, and (4) anxiety about change. An important feature of this model is that it deals with anxiety about a specific event, and only varia- bles relevant to that event are considered as influences of anxiety. That is, only needs, job characteristics, and social information which directly deal with the change under consideration are inputs to the model. If the threat were a switch in company policy from flextime to a rigid nine to five schedule, relevant job characteristics might be autonomy and interdependence. If the threat were a job transfer from home office sales assistant to regional director of sales in the outback of Australia, relevant needs might be the 'need for feedback or social interaction. In any case, there is not a pre- defined list of needs and job characteristics in this model; they are deter- mined by the threat for which anxiety is being predicted. Anxiety is directly influenced by three components in the model: needs, job characteristics, and social information. The three types of social information relevant to the formulation of anxiety are those involving (1) needs related to the change, (2) descriptions of the change, and (3) interpretations of descriptions in terms of needs. In essence, the first tells people what they want, the second tells people what they're getting, and the third tells people why what they're getting is (or is not) what they want.- A meSsage with more complex information should have a greater influence on anxiety than one that provides simpler information. Thus, interpretive information should have a greater impact on anxiety than either descriptive or needs information. The distinction between the effect of description 21 and need information is less clear. However, stress research has generally found a stronger relationship between environmental/job characteristics and stress than between individual characteristics and stress (Cooper & Marshall, 1976) so it is likely that descriptive information will have a greater impact on anxiety than need information. Social information and job characteristics are also causes of perceived needs in this model, as Salancik and Pfeffer suggest that need statements will be based on all inputs available to an indi- vidual. “Formal hypotheses for this model can be specified at both the macro and micro levels. There are two macro level hypotheses: (1) a hypothesis of dif- ference between the proposed model and the null model, and (2) a hypothesis of goodness of fit of the model to the data. The micro level hypotheses are hypotheses of the significance of individual relationships in the model. The research hypotheses in each of these areas are stated below. Macro level hypotheses: H1: There is a significant difference between the proposed model and the null model of no relationships among the variables (M0 7 MA). The proposed model provides an accurate representation of the relationships found in the data (gA = S). Micro level hypotheses: H3: An increase in positivity of information about organiza- tional change will lead to a decrease in anxiety about organizational change (B31< 0; Y31< 0). H4: There is a significant relationship between social infor- mation and perceived needs (821 f O) 5: There is a significant relationship between job charac- teristics and perceived needs (v22 f O). 22 H6: There is a significant relationship between perceived needs and anxiety about organizational change (832 f 0). 7: There is a significant relationship between job characteristics and anxiety about organizational change (v32 f 0). It should be noted that the direction of relationship in hypotheses 4, 5, 6, and 7 is a function of the particular needs and_job characteristics relevant to the organizational change. CHAPTER II METHODOLOGY AND PROCEDURES The proposed model of anxiety antecedents was tested with a field experiment examining employees anticipating a move to a new environment. The workers, employees of a state agency, were working in traditional work settings with floor to ceiling offices, but were planning to move to offices in a new building. The new offices used "open landscaping," characterized. by space without interior walls, work stations separated by partitions, and higher levels of visibility than in traditional offices. For almost all employees, the move represented a shift in work environment; one which would radically alter processes for getting work accomplished. This chapter will describe the methodology and procedures used in this study. First, the subjects in the research will be discuSSed,then the, procedures and research design will be described. (Next, instrumentation of variables under consideration will be discussed. Finally, the analytic model to be tested will be presented, and procedures for testing the model as a system of structural equations will be described. Subjects Subjects for this study were drawn from a state department of education in the midwest. Three intact service areas (departments) were used for the . study, a financial aid area employing 90 people, a research area employing 23' 24 25 people, and a vocational-technical education area employing 65 people. These service areas were chosen primarily for pragmatic reasons. Administra- tors in these service areas were interested in the proposed research and willing to allot time in the work day for research sessions. There were a total of 180 participants. Both males and females were well-represented in the sample, and all organizational levels were represented. A brief discussion of the social climate of this organization is in order for a better understanding of the subjects participating in the research. The agency employs approximately 1,000 workers at various levels of status and income. All workers are civil service employees, and most workers view themselves as professionals. The agency, like many governmental agencies, is a relatively structured organization. There is a central office which oversees four bureaus and several offices, Bureaus are then divided into functional service areas. Although there are relatively rigid reporting channels within the agency, many employees feel that their service areas operate with a fair degree of autonomy. The pre-move decentralized physical layout of the agency reinforced this feeling of divisional independence. At the time of the move, the agency was experiencing other forms of stressfiwhich could have served to overshadow worries about the changed work environment. Because of economic instability, the agency was involved in a reduction of force, and many workers were worried about layoffs. The agency was also going through a structural reorganization, but the reorgani- zation was only a concern to upper-level management. Research Design and Procedures In order to assess the impact of information on employee attitudes, six informational conditions were considered: (1) information about employee 25 needs reflecting positively on the move, (2) information about employee needs reflecting negatively on the move, (3) positive descriptions of the new offices, (4) negative descriptions of the new offices, (5) positive interpretations of the move, and (6) negative interpretations of the move. A control condition of no information about the move was also used. 'Because of organizational constraints, only quasi-random assignment to informational conditions was possible. A contact person in each service area set up a schedule of two to six 45 minute research sessions. Employees could then sign up for a time which was convenient for their work schedule. Because this study was part of a series of research programs with the organi- zation, employees were relatively cooperative about signing up for and attending research sessions. Employees were told when signing up that they would be completing a questionnaire and would also receive feedback from researchers about earlier phases of the research project. At the research session, the researcher gave a brief presentation about previous results from the research project. The presentation included the six informational manipulations; the presentation in the control group did not include any previous results relevant to the move. Employees were told before the presentation that departmental management was considering including the information in an upcoming newsletter and that they would be asked their opinion of the informations helpfulness in the questionnaire. After the researcher's presentation, questionnaires were distributed. The first page of the questionnaire was a cover letter which contained a brief version of the information given in the researcher‘s presentation. Employees were asked to fill out the questionnaire and were thanked for their cooperation. The researcher stayed in the room to answer any questions employees might have. After research procedures were completed in all three service areas, 26 participating employees were sent a letter explaining the research. Instrumentation The model derived from social information processing theory includes four components: information, needs, job characteristics, and anxiety. The operationalization of these four model components will be discussed in this section. The questionnaire items used are included in Appendix A and the informational manipulations used are included in Appendix 8. Information. The information component of the model was included in the research design with two variables. Information about the move was mani- pulated through the research presentation described in the procedure section and through a message included in the questionnaire. In addition to the manipulation, a measure was included in the instrument to assess the adequacy of information which had already been received about the move. The manipulation of social information about the move required the design of two forms each of six messages. Messages_for the oral presentation and cover letter were constructed for the positive/needs, negative/needs, poSitive/ description, negative/description, positive/interpretation, and negative/ interpretation conditions. It was important to control several dimensions of these messages. The most important of these was_the_topjcharea considered in the message. The topic had to be relevant to the move and salient to the employees. It was alSo necessary that the topic be capable of both posi- tive and negative treatment. The topic area chosen was access. It was clear from in-depth employee interviews and questionnaire responses from previous phases of the study that access to others in the organization -- and the extent to which others had access to them -- was a topic of great interest to organizational employees. This topic is also intimately linked to a change to open-scape offices. In 27 addition, it is possible to emphasize either positive or negative issues of access in open-scape offices. Positive messages emphasized the impor- tance of access to other people and more information, and the degree to which open offices opened lines of communication. Negative messages empha- sized the point that more access meant less privacy and the idea that con- fidential converSations were more difficult in open offices. A second dimension of the manipulation which had to be taken into con- sideration was the perceived source of the information. Although Salancik and Pfeffer (1978) suggest that information from a variety of sources can be influential in shaping employee attitudes, the central portion of their theory involves social information received from colleagues at work. Thus, the message received in the study had to be perceived as originating, at least in part, with employees of the organization. To do this, each mes- sage included references to the interviews and survey results from previous stages of the study. An effort was also made to control for the length and wording of the informational manipulations. Each written message constructed for the cover letter consisted of a single paragraph. The oral messages were each approxi- mately four minutes long with similar topic ordering and word choice. In addition to the manipulation, three items were included in the in- strument to assess the adequacy of previous information received about the move. These items asked if the information received had been timely, useful, and had adequately answered questions about the move. This scale had been used in an earlier large-scale study (N = 450) at the same organization and had proved to be reliable (<1= .86). A nine-point response format ranging from strongly agree to strongly disagree was used. Only the endponts of the scale were anchored. 28 Individual needs. Needs of the individual were operationalized through items asking about the level of privacy the individual needed to do his/her job, and the level of interdependence needed. These needs are directly related to the topic of the information provided and the organiza- tional change the workers would experience. Three privacy items and three interdependence items were included. Although the scales have not been sub- ject to rigorous tests of reliability and validity, forms of these questions were used in a previous study of 450 employees at the organization studied. Both scales were shown to be quite reliable (for the privacy scale,c1= .86, for the interdependence scale,'<1= .75). Again, a nine-point response for- mat was used. Job characteristics. A variety of job characteristics might impact on anxiety about a move to open offices. The most general of these is whether your job is one that will be changed a great deal by the move. For many lower level employees (i.e., clerks and secretaries) the change to the new office would not involve much adjustment in work procedures. They were “in the open" already, so change would be minimal. However, as one moved up the hierarchy, the effect of the move was greater. Educational consultants would move from plush offices to mid-sized cubicles. They would have to make major adjustments in work processes. The same held true for managers and super- visors. Thus, job characteristics (as they impact on the move) were opera- tionalized as job level. As the organization being studied was a government agency, civil service classifications were available. These classifica- tion numbers provided a clear index of the job level and job type. Anxiety. As questions about anxiety must necessarily deal with a very specific event, no standard items were available for this variable. However, previous stress research (Kahn, et al., 1964, French, et al., 1982) was 29 used as a basis for the development of questions tapping anxiety about the move. A three-question scale was developed. The three keywords in the questions ("anxious," "worry," "concern") all attempted to tap the concept of a future event which is causing anxiety for the individual. This variable was also operationalized using a nine-point response format. Information helpfulness. Three items were developed to measure the perceived helpfulness of information received. Using a nine-point response scale, employees were asked whether information received about the move was positive, favorable, and appropriate for publication in a department-wide newsletter. Information helpfulness items followed all other items in the instrument to guard against consistency and priming effects. Analysis Based on the theoretical model developed in Chapter I and the instru- mentation discussed above, a full model including theoretical and measure- ment components was derived.(See Figure 2). This section will discuss the statistical data analyses performed to analyze this model. The theoretical model was tested using a two-step analytical technique. First, the measurement models were analyzed using the confirmatory factor analysis subroutine of the PACKAGE computer program (Hunter, Cohen, & Nicol, 1982). After unidimensional measurement model§_were confirmed, the struc- tural equation model for the factors was analyzed using the LISREL V computer program (Joreskog & Sorbom, 1981). A two-step process of this sort is recom- mended by McPhee and Babrow (1983). In comparing the maximum likelihood estimation methods of LISREL/and the sequential confirmatory factor analysis, and regression procedure/of Hunter and Gerbing (1982), McPhee and Babrow state that: Figure 2. 3O Combined theoretical and measurement model of employee anxiety. Legend for combined model. 51 :2 53 01 02 03 04 X1 x2 X3 Y1 Y2 Y3 Yu Y5 Y6 Y7 Y8 Y9 Y10 Y11 Y12 Manipulated Information Previous Information Job Level Information Helpfulness Need for Privacy Need for Anxiety multiple multiple multiple multiple multiple multiple multiple multiple multiple 'multiple multiple multiple multiple multiple multiple Interdependence indicator indicator indicator indicator indicator indicator indicator indicator indicator indicator indicator indicator indicator indicator indicator of of of of of of of of of of of of of of previous previous previous information information information information helpfulness information helpfulness information helpfulness need for need for need for need for need for need for anxiety anxiety of anxiety privacy privacy privacy interdependence interdependence interdependence 31 .xuowxcm mozoPQEw eo Ponce ucosmezmmme new uwuoeomnp umcwnsoo mu mu hm v a p 0% m% 5% ./~\. m: 4/ mu .N weamwu 32 Generally, the arguments presented by all parties . seem to support the following general statement: LISREL is unsurpassed in degree of “leverage" and in efficiency of estimation, but is also the most dependent on theoretical and statistical assump- tions for its validity. Where there is substan- tial ambiguity about the causal structure or measurement validity, PLS [partial least squares] or PACKAGE should be used initially, followed if successful by LISREL (1983, p. 7). The confirmatory factor analysis subroutine of PACKAGE allows the researcher to specify an a priori factor structure. The program then pro- vides a matrix of correlations among and between cluster items, correlations between factors, and factor loadings, using communalities in the diagonal. The unidimensionality of the proposed factors can then be assessed using three criteria proposed by Hunter (1977): homogeneity of item contentjflin- ternal consistency within clusters: and parallelism of items in a cluster with outside variables. The-LISREL V computer program provides estimates of parameters in linear g'structural equations through the method of maximum likelihood. LISREL V also provides a ngoodness-of-fit test for models which are over-identified. : identified, and the fit 0f the model can be tested. The hypotheses presented earlier were tested in the following manner. The first hypothesis stated that the proposed model would be signifi- cantly different from the null model. The null model examined was one of no relationship among the theoretical variables (that is, all the Y and 8 parameters set to zero). The hypothesis of difference between the two models was tested with a test for the significance of the difference between x2 values for each of the models (see Bentler & Bonnett, 1980). This x2d is itself distributed as>dzwith degrees of freedom equal to the difference in degrees of freedom for the two models. The critical value for this test 33 was set at a = .05. The second hypothesis stated that the proposed model would be a good fit to the data. This was tested using the x2 goodness-of-fit test which determines whether the structure imposed by the restrictions of the model "matches" the structure that created the variances and covariances in the data. The logic of this test is one of looking for similarity rather than difference; hence, an insignificant value for x2 is desirable. The critical ‘ value for this test was also set at a = .05. The fit of the model was also considered through the examination of residual covariances. The last set of hypotheses considered the significance of individual structural coefficients in the model. These hypotheses were tested with the tyvalues of the coefficients (coefficients divided by standard error). The critical value for t was also set at a = .05. CHAPTER III RESULTS This chapter will describe the results of analyses performed to test the proposed social information model of employee anxiety. First, prelimi- nary descriptive analyses will be described. Next, analyses used to examine the measurement models involved will be discussed. Finally, the test of the proposed model will be described and revisions to the model will be presented. Preliminary Analyses Preliminary descriptive analyses were performed using the Statistical Package for the Social Sciences (Nie, Hull, Jenkins, Steinbrenner, & Bent, 1975). 146 employees participated in research sessions and returned question- naires. Seventeen of these were employed in the research service area, 81 were employed in financial assistance, and 48 were employed in vocational technical education. 16 employees received positive interpretation informa- tion, 24 received positive description information, 21 received positive needs information, 21 received negative interpretation information, 12 received negative description information, 24 received negative needs information, and 28 employees were in the control condition. All levels of the organization 34 35 were well-represented in the sample. Descriptive statistics revealed one anomaly in the data. Respondents in the control situation (no information about the move) were not responding neutrally to the information helpfulness variable -- i.e., between the positive and negative conditions -- but were instead responding even more negatively than respondents in the negative.interpretation condition. It appeared that any information about the move, even negative information, was perceived as being more positive or helpful than receiving no information about the move. To reflect this situation in the analyses, the seven condi- tions were coded as follows: 1 = control condition, 2 = negative interpre- tation, 3 = negative description, 4 = negative needs, 5 = positive needs, 6 = positive description, 7 = positive interpretation. Measurement models Analyses of the measurement models in this study were performed using the confirmatory factor analysis subroutine of PACKAGE (Hunter, Cohen, & Nicol, 1982). The model involved seven variables: manipulated information, previous information, job level, perceived helpfulness of information, need. for privacy, need for interdependence, and anxiety about the move. One of these variables was manipulated, and another had only one indicator (job level). The other five theoretical variables each had three indicators, and these variables were submitted to confirmatory factor analysis. 4 Hunter (1977) has proposed three criteria (and theorems) for assessing the unidimensionality of hypothesized factors. These criteria were used to evaluate the proposed measurement models. The first criterion for unidimen- .sionality is homogeneity of item content. Items within each factor were derived from standard scales or written with the goal of_tapping a sjggle concept. Thus, the items in each cluster appear to be homogeneous in 36 content. The second criterion for unidimensionality is internal consistency within each item cluster, and the third is parallelism among cluster items with outside variables. Tables 1 through 10 present the items, factor loadings, deviations for internal consistency (observed minus predicted) and deviations for parallelism (observed minus predicted) for the five hypothesized factors. Tables 1 and 2 describe analyses of the previous information factor, Tables 3 and 4 summarize the information helpfulness factor, Tables 5 and 6 sum- marize the privacy needs factor, Tables 7 and 8 summarize the interdependence needs factor, and Tables 9 and 10 summarize the factor for anxiety about open offices. The factor clusters for previous information, helpfulness of informa- tion, privacy needs, and interdependence needs all have relatively high and consistent factor loadings. The lowest factor loading for this set of fac- tors is .44 for item 4 of the privacy needs factor. The highest factor loading is .98 for item 16 of the information helpfulness factor. The average factor loading for these four factors is .78. In addition to these favorable factor loadings, the deviations for both internal consistency and parallelism are all within a range that would be expected from sampling error. However, the measurement model for the anxiety factor does not appear to be unidimensional. In particular, item-#3 ("I feel anxious about the move to the new building") has a factor loading of only ng,_while the other two items (“I am not concerned about working in open offices” and "The thought of working in an open office worries me") have factor loadings of .64 and .81, respectively. A re-examination of item content provides a possible explanation for the low loading of item #3. This item asks 37 Table 1. Items and factor loadings for previous information factor. Item Content Loading VAR06 The information I have received about the .86. new building has been timely. VAR07 The information I have received about the new bu1lding has been useful. .89 VAR08 The information I have received about the new building has adequately answered my ° questions about the move. .80 38 Table 2. Deviations of observed from expected correlations for internal consistency and parallelism. Previous information factor. Items 6 . .7. , 8 6 _ 7 .00 - 8 -.01 01 - 9 - 04 .04 ll 14 - 10 -.03 02 4 02 .05 11 11 -.06 -.05 .ll 13 -.05 -.02 .00 5 - 04 .03 02 10 .00 .01 .OO 12 -.06 .03 .02 15 -.O4 .03 -.15 16 -.04 .09 -.08 39 Table 3. Item and factor loadings for information helpfulness factor. Item Content Loading VAR15 The information provided was positive. , .93 VAR16 The information provided was favorable. .98 VAR17 The information provided was suitable for .89 publication in FYI. 40 Table 4. Deviations of observed from expected correlations for internal consistency and parallelism. Information helpfulness factor. Items » 15 . . 16 .,. 17 15 - 16 .OO - 17 -.02. . .00. - 9 - 13 -.O6 - O6 14 .06 .10 .11 4 - 04 -.08 - 03 11 - 05 -.07 - O4 13 .07 .09 .13 5 - 06 -.02 - 09 10 .05 .04 .08 12 - 01 .01 - 01 6 - O4 -.04 - 01 7 .03 .09 .14 41 Table 5. Item and factor loadings for privacy needs factor. Item Content . . Loading VARO4 I often have to discuss private matters at .44 work without being overheard. VARll It's not important that I have privacy to get .65 my job done. VAR13 I need a quiet work area to get my job done. .75 42 Table 6. Deviations of observed from expected correlations for internal consistency and parallelism. Privacy needs factor. Items 4 . ll . ,. 13 4 _ ll -.02 - 13 4 .OO .01 - 9 - O7 .03 - 14 14 .07 .03 .08 5 ll -.05 04 10 -.O6 -.11 -.Ol 12 -.Ol -.06 .14 6 02 -.06 - 05 7 O9 -.05 - 02 8 .ll -.06 .00 15 - 04 -.05 O7 16 - 08 -.07 O9 43 Table 7. Item and factor loadings for interdependence needs factor. Item Content Loading VAROS I often have to meet or check with other people .58 in the department in order to do my job. VARlO I often have to cooperate directly with other .79 people in the department in order to do my job. VAR12 My job requires me to work cloSely with others .80 employed at MOE. 44 Table 8. Deviations of observed from expected correlations for internal consistency and parallelism. Interdependence needs factor. Item 5 . ..lO . .12 5 - 10 —.01 - 12 4 won .01 - 9 _ - 03 -.16 oo 14 .12 .oo .09 4 ' .11 -.06 4.01 11 - os -.11 - 05 13 .04 - 01 14 o - 04 .oo - O6 7 .03 .01 .03 8 .02 .oo .02 15 - 06 -.05 - 01 16 -.02 04 01 45 Table 9. Item and factor loadings for anxiety factor. Item Content . . { Loading VAR03 I feel anxious about the move to the new .19 building. VARO9 I am not concerned about working in open .64 offices. VAR14 The thought of working in an open office -. .81 worries me. 46 Table 10. Deviations of observed from expected correlations for internal consistency and parallelism. Anxiety factor. Item 3 A 9 14 3 .. 9 -.02 - 14 .03 _ —.01 K V - 4 03 -.03 06 11 - 12 .11 03 13 5.13 -.06 .07 5 14 -.07 06 1o .14 -.21 -.07 12 .05 -.05 .02 6 03 -.08 - 12 7 10 .oo - 06 3 09 .06 - 02 15 .01 -.16 .oo 16 .04 -.09 .10 17 .oo -.08 .12 47 about “the move to the new building" while the other two items ask about "working in an open office." While the move to the new building is a move to open offices, the move also encompasses other changes (e.g., new location, parking changed, new equipment). These aspects of the move could explain the low loading of a move-specific item on the anxiety factor. **"i:39 Because of thefilow item loading, a second confirmatory factor analysis was performed withgonly two itemsgin the anxiety factor. The factor load- ings and deviations for parallelism are presented in Tables 11 and 12. A two item factor prohibited the examination of internal consistency. When item #3 is removed from the factor, the other two items each have factor loadings of .74, and the deviations for parallelism are all within a range that would be expected from sampling error. Thus, it appears that these items represent a 90191T§9§19041 facton measuring anxiety about open offices. . In summary, then, confirmatory factor analyseinndicated that unidimeng sional factors for privacy needs, interdependence needs, helpfulness of information, and previous information were being measured by the three item scales included in the questionnaire. The original scale for anxiety was unsatisfactory, as one item had an extremely low factor loading. When this item was dropped, though, ayunidimensional_facto: for anxiety was obtained. ’ Theoretical Structural Equation Model Because unidimensionalAmeasuremegtflmgggls were confirmed for all of the theoretical variables involved, the correlations between factors were used as input for the analysis of a theoretical structural equation model with the LISREL V computer program (Joreskog & Sorbom, 1981). The matrix of correlations among theoretical variables is presented 48 Table 11. Item and factor loadings for revised anxiety factor. Item Content. Loading VARO9 I am not concerned about working in open .74 offices. VAR14 The thought of working in an open office .74 worries me. 49 Table 12. Deviations of observed from expected correlations for parallelism. Revised anxiety factor. Item 9 l4 4 -.07 .07 11 .03 .03 13 -.14 .08 5 -.03 12 10 -.16 oo 12 .oo .09 6 -.04 - 1o 7 .04 - 03 8 .10 .01 15 -.13' 06 16 -.06 1o 50 in Table 13. Several aspects of the zero-order correlations among theo- retical variables are worth mentioning before examining the structural equa- tion models analyzed. First, there is an extremely high correlation (.78) - between the privacy needs factor and the anxiety factor. This could indicate that a higher order factor exists and that all five items are indicators of this factor. However, the content of the items are clearly distinct, and . it seemed conceptually important to maintain these two factors as separate entities in spite of the high correlation. Second, the factor for need for interdependence does not correlate highly with any other variables in the model (the average correlation is .12). This variable was one of two need variables included in the original model as individual variables influencing anxiety. As it was not related to any other variables in the model, need for interdependence was dropped from the theoretical model and need for privacy was used as the only individual need variable. The theoretical structural equation model representing the theory deve- loped in Chapter I is presented in Figure 3. This model indicates four causal indicators of anxiety -- job level, privacy needs, helpfulness of informa- tion, and the manipulation itself. Helpfulness of information is caused by the manipulation and previous information, and privacy needs are deter- mined by information helpfulness and job level. The structural coefficients for the causal links are indicated in Figure 3. The equations for the endogenous variables in this model, and_RE coeffi- cients, are presented in Table 14. Standard errors are indicated below the structural coefficients. The squared multiple correlations for these equa- tions are .172, .260, and .619, respectively. The average residual for the model was .058. This model was evaluated using several criteria. First, the individual 51 Table 13. Correlation matrix among theoretical variables. l 2 3 4 5 l. Manipulated - Information 2. Previous -.02 - Information 3. Job -.14 -.04 - bevel 4. Information .28 .30 -.13 - Helpfulness 5. Need for -.14 ' -.19 .42 -.12 - Pnivacy 6. Need for -.02 .ll .22 -.10 .28 Interdependece 7. Anxiety -.20 -.13 .42 -.20 .78 52 Figure 3. Theoretical model with structural coefficients. Lengend for the theoretical model. X 1 manipulated information X2 = previous information X3 = job level Y1 = information helpfulness Y2 = need for privacy .< II 3 anx1ety 53 Table 14. Structural equations for theoretical model. Standard errors indicated below structural coefficients. Equation 5? - 2 - v1 - .286x1 + .306x2 + C] 3v] - .172 (.076) (.076) _ 2 - Y2 — -.055Y1 + .503X3 + C2 .3 Y2 - .26 (.072) (.072) _ 2 _ Y3 - -.089Y.I + .752Y2 - .086X1 + .015X3 + C3 .3 Y3 - .619 (.054) (.060) (.054) (.060) 54 structural coefficients were tested for significance using tyvalues. ‘-fi ‘4‘ 7:) Second, the theoretical model was compared to the null model (usingxzd ( values) to determine if the two models were significantly different. Third, the x2 goodness-of—fit test was used to determine if the model proposed was a good match to the data. It should be noted that a significant de is required for rejection of the null hypothesis and a nonsignificant x2 is appropriate to assess the adequacy of fit between the actual and reconstructed correlation matrices. Finally, two increment of fit indices were examined to determine the extent to which the model accounted for variance in the data. Eight structural coefficients are included in the theoretical structural equation model. An examination of tfvalues for these coefficients indicated that five of the eight were statistically significant. The three coefficients that were not significant were the links between job level and anxiety (.015), the manipulation and anxiety (-.O68), and helpfulness of information and privacy needs (-.055). All other coefficients were significant and ranged from -.089 (information helpfulness predicting anxiety) to .752 (privacy needs predicting anxiety).1 In comparing the proposed model to the null model, the x2d value was 213.56 with 8 degrees of freedom. This value is significant, indicating that the proposed model is different from the null. The x2 value for the proposed M_ “— theoretical model was 8.12 with 4 degrees of freedom. This value is not significant, indicating a good fit of the model to the data. Two increment of fit indices were also computed. These indices can range from 0 to 1.00. A value of 1.00 represents an accounting for all of the variance in the data. The theoretical model yielded a value of .941 for D and .963 for A . Both. of these increment-of—fit indices indicate an excellent degree of fit for the model. 55 Model Modifications The proposed theoretical model proved to be a good model in several ways. The model was significantly different from the null model of no relation- ship, it was a good fit to the data, and it accounted for a large proportion of the variance. However, three of the eight links in the model were not significant. Because of these insignificant links, an effort was made to find a model which better fit the data and included only significant links, but which was still conceptually consistent with the theoretical framework provided by social information processing theory. ( “"“~The revised models considered were tested using a procedure designed g 15‘ :2. by Bentler and Bonnett (1980) for testing hierarchically arranged structural}: ){J’ ;';A equation models. de values are used to compare models which place different I 3 degrees of restriction on the data in order to determine which model does ; the best job of fitting the data with the most parsimonious structure pos- 1*},sible. ‘ Revisions to the model were made in two ways. First, the insignificant links in the original model were removed in a step-by-step process in order to make the model more parsimonious yet retain a good fit to the data. The first causal link removed was the link between information helpfulness and need for privacy (821). Next, the direct link between job status and anxiety (v33) was removed. Finally, the direct link between the manipulation and anxiety (Ygl) was removed. In addition to removing links for a more parsimonious model, an effort was made to add any links which were theoretically consistent and would improve the fit of the model to the data. The modification indices produced by the LISREL V computer program indicate the increment in x2 value which can be obtained by adding a structural coefficient to the model. The 56 modification indices suggested that adding a link between previous informa- tion about the move and perceived need for privacy might significantly improve the fit of the model to the data. As a link between information about open offices and perceived need for privacy seemed conceptually con- sistent with social information processing theory, a model with this link (v22) was also compared to the original model and the revised models. In sum, then, four revised models were compared to the original model. Three of these were hierarchically arranged variations of the original model, with insignificant links dropped one at a time. The fourth model adds one link (suggested by the modification indices) to the most parsimonious of the hierarchical models. A comparison among these four models and the ori- ginal in terms of goodness of fit, increment of fit, and difference between models is presented in Table 15. As can be seen from Table 15, the original model (Model 1) is not a significantly better fit to the data than any of the hierarchically arranged revised models (Model 2, Model 3, and Model 4). Because there is not a significant difference in fit for any of these models, it is reasonable to choose the model which is most parsimonious, i.e., the one including the fgw§§t_structural coefficients. This is Model 4. However, an addition of the link between previous information and need for information ggg§_provide a significant improvement in fit for the model (as indicated by a significant xzd figure for Model 4 minus Model 5). Thus, the model which provides a good fit to the data in the most parsimonious manner is Model 5: This model, along with structural coefficients, is presented in Figure 4. All of the structural coefficients for Model 5 are significant (as indi- cated by tfvalues). The equations for the dependent variables in ModelS. are presented in Table 16L The squared multiple correlation coefficients 57 Table 15. Model comparisons. Model . de g:_ p A Comparison Null minus Model 1 213.56*' 8 .941 .963 Null minus Model 2 212.98* 7 .958 .961 Null minus Model 3 212.92* 6 .974 .961 Null minus Model 4 211.28* 5 .972 .953 Null minus Model 5 217.03* 6 1.013 .979 Model 2 minus Model 1 .58 l ‘ .017 .003 Model 3 minus Model 2 .06 l .016 .000 Model 4 minus Model 3 1.64 l -.002 .007 Model 4 minus Model 5 I 5.75* l .041 .026 *significant value at p <..05 Legend for Table 15. Model 1 3 Original theoretical structural equation model Model 2 = Model 1 minus insignificant link between information helpfulness and privacy needs. Model 3 = Model 2 minus insignificant link between job level and anxiety. Model 4 = Model 3 minus insignificant link between manipulated information and anxiety. Model 5 = Model 4 plus link between previous information and privacy needs. Figure 4. 58 .286 -.108 .306 .767 - 170 .503 Revised model with structural coefficients. Legend for revised model. X1 = manipulated information X2 = previous information 3 X job level .< ll 1 information helpfulness - need for privacy Y3 = anxiety 59 Table 16. Structural equations for revised model. Standard errors indicated below structural coefficients. . 2 Equation .3 _ 2- Y1 - .286X1 + .306X2 + :1 .3 Y] - .17 (.076) (.076) _ 2- Y2 - -.l70X2 + .503X3 + :2 .3 Y2 - .289 (.071) (.071) _ 2- Y3 - -.108Y1 + .767Y2 + c3 .3 Y3 - .617 (.052) (.052) 60 for these equations are, respectively, .172, .289, and .617. The average residual for the model was .02. The increment of fit indices indicate that this model provides an excellent fit to the data (0 = 1.013; A = .979). However, this model should be tested and validated with new data before it is fully accepted. CHAPTER IV DISCUSSION The present research was undertaken to test the social information pro- cessing theory by examining the determinants of employee anxiety about organi- zational change. This chapter will discuss the results of the field study conducted to test Salancik and Pfeffer's theory of job attitudes. First, the obtained model will be discussed in terms of the theory from which it was generated. This discussion will include consideration of the determinants of anxiety, needs, and information helpfulness.- Next, the small impact of interdependence needs will be considered, and implications for traditional need satisfaction models will be discussed. Third, the perceptions by employees of different types of information will be examined, with attention given to the "no information" control group.- Next,linfitations to the study will be presented. Finally, managerial implications and directions for ' future research will be explored. Social Information Processing Theory_and the Obtained Madel Salancik and Pfeffer (1978) proposed their social information processing theory of job attitudes as a response to prevailing need satisfaction theo- ries which conceptualized job attitudes as a function of the fit between individual needs and objective job characteristics. Instead, Salancik and 61 62 Pfeffer suggested that job attitudes are also influenced by social information received about objective characteristics of the organizational environment, employee needs, and interpretations of the environment in terms of needs. Thus, there are three major determinants of job attitudes: individual needs, job characteristics, and social information. Salancik and Pfeffer also suggest that an individual‘s own behavior will influence attitudes. Tests of social information processing thebry to this point have involved experimental settings and have provided relatively strong support for the theory (O'Reilly & Caldwell, 1979; White & Mitchell, 1979). These studies, however, have not yet established the external validity of the theory. In addition, only traditional job attitudes identified by need satisfaction theorists have been considered. The study described in this paper extended social information processing theory to the study of employee anxiety about organizational change and con- stituted a partial test of the theory in a realistic organizational setting. Anxiety (job attitude) was hypothesized to be a function of individual needs, job characteristics, and social information, as proposed by Salancik and Pfeffer. Employees facing an actual organizational change (move to open offices) were studied. Unfortunately, the impact of the individual's own behavior on anxiety, included as a part of the theory, could not be assessed. The structural equation model derived from Salancik and Pfeffer's social information processing theory was significantly different from the null model and provided a good fit to the data. In addition, increment of fit indices indicated that the model accounted for a great deal of the variance in the data. However, several structural coefficients were insignificant, and there was evidence that another link might be significant. Hence, model modifications were made in an attempt to make the model even better. A 63 new model was found which provided a better fit for the data and only included significant causal links. Most of the following discussion will be in reference to this revised model (see Figure 4). However, it should be emphasized that this model should be validated with new data before it is fully accepted. The revised model provided relatively strong support f0r Salancik and Pfeffer's social information processing theory in a new conceptual area, employee anxiety. Social information (bothmanipulated and previous information) had an impact on anxiety, as did the individual variable of need for privacy. Hewever, the impact of job level on anxiety was totally mediated by perceived need for privacy. Thus, the job attitude of anxiety about open offices was indeed determined by the three factors suggested by Salancik and Pfeffer: social information, individual needs, and job charac- teristics (mediated by need for privacy). This model will be explored further by examining three areas: the relative impact of the determinants of anxiety, the determinants of need for privacy, and the determinants of perceptions of information helpfulness. Determinants of Anxiety: Relative Impact. It was noted above that infor- mation, needs, and job level all influenced anxiety, thus supporting social information processing theory. However, an examination of the standardized structural coefficients in the model indicate clearly that the three factors did not have an equal impact on anxiety about open offices. The coefficient linking information helpfulness and anxiety was -.108, while the coefficient for the link between privacy needs and anxiety was .767. Thus, though both coefficients are statistically significant, it is clear that need for privacy had a much greater impact on anxiety than social information. There are several possible explanations for the importance of needs 64 over social information. First, of course, it's possible that needs are a more important determinant of job attitudes than Salancik and Pfeffer‘s theory would suggest. This could indicate that the need satisfaction model should not be rejected as whole-heartedly as Salancik and Pfeffer advocate, but should just be revised to consider more pertinent needs. The organiza- tional situation considered could also account for the strong impact of needs. The amount of privacy was clearly a central issue in the move to the new building. Individual and role-related needs could well have a smaller impact on other more general job attitudes. Finally, the discrepancy in effect size could be a function of the strength of manipulation. A stronger message, or information repeated at several points in time, could have had a more sub- stantial impact on anxiety about open offices. Determinants of needs. Although the dependent variable of the most interest in this study was anxiety, it is also valuable to consider the causal influences of other variables in the model. Perceived need for privacy .503), and was also ll was strongly influenced by job level in the model (Yza influenced by previous information about the move (Yzz -.l70). There was no causal relationship between information helpfulness and privacy needs, as proposed in the original model. Salancik and Pfeffer suggest that indi- vidual needs are a function of social information, and the obtained model provides partial support for this contention. However, the informational variable with an impact on needs was previous information -- not the manipu- lation itself or the helpfulness of information variable. It appears that an accumulation of information over time is needed to influence perceptions of needs which might be strongly held. The strong influence of job level on perceived needs seems intuitively obvious, but it is a relationship not entirely accounted for by social 65 information processing theory or need satisfaction theories of job atti- tudes. These theoretical bases see needs as a mediating factor which will determine the nature of the functional relationship between the job charac- teristics and job attitudes. However, the need satisfaction model suggests that needs are relatively stable traits and not greatly influenced by the job, and social information processing theory places the major causal influ- ence of needs on information. The model deve10ped in this study suggests that a more direct relationship between needs and job characteristics should be considered. Determinants of information helpfulness. The variable of information helpfulness was designed as a quasi-manipulation check. Because the study was dealing with a ”real" situation, previous information received was also included as a cause of this variable. It seemed unlikely that a question about information helpfulness would only be influenced by the manipulation when employees had been receiving information about the move for several months. Both the manipulation and previous information had a moderate im- pact on information helpfulness, which has interesting implications. Salancik and Pfeffer state that the most salient information available will be used in formulating attitude statements. They also suggest that recency will in large part determine the salience of information. The results of this study contradict that portion of social information processing theory. The manipulation and previous information had virtually equal impact on information helpfulness, the manipulation did ngt_have a direct impact on anxiety, and previous information gjg_have a direct impact on perceived privacy needs. This suggests that recently obtained information (at least of the strength used in this study) will not have an over-riding impact on statements of job attitudes. The accumulation of information 66 appears to be as important a determinant of attitudes as recent information. Moreover, in the statement of "strongly held needs, the accumulation of information seems to be more important than recently obtained information. In summary, then, the model obtained provides relatively strong support for social information processing theory. The model was a good fit to data obtained from organizational employees facing an upcoming change. However, the model indicated a stronger effect for privacy needs than social infor- mation. Also, job level was not a direct cause of anxiety; the relationship was mediated by privacy needs. Finally, accumulated information appeared to be as important, if not more important, than recent information in the formation of job attitudes. The next two sections will discuss important findings of this study not directly related to the accepted model. Impact of Interdependence Needs Perceived need for interdependence was originally proposed as one of two individual needs which would be affected by social information and in turn positively influence anxiety about open offices. Because the zero- order correlations of this factor with all other variables were so low, the interdependence needs variable was dropped from the model. Further analyses were performed using need for privacy as the sole individual need variable. The lack of impact of interdependence needs has several implications. First, it should be noted that some of the informational manipulations dealt with the subject of interdependence. The messages were designed in part to reflect the needs which would be important in the move. Because_ these messages dealt with a subject that was not salient to the employees, they may not have had as strong an impact as possible. The lack of influence of interdependence needs provides a possible explanation for the relatively small effect of information on anxiety. 67 Second, the low impact of interdependence needs emphasizes a point made in the first chapter -- traditional job characteristics/needs may not mean much in the prediction of specific job attitudes. Interdependence is one of the job characteristics originally identified by Hackman and Lawler (1971) as a determinant of job attitudes. Because there seemed to be a strong conceptual link between interdependence and open offices, modified items from the Job Description Survey (Hackman & Oldham, 1974) were used as an individual need variable in the model. The failure of this factor to fit into the model points again to the inadequacy of these traditional variables. One possible explanation is that all of the traditional job characteristics as measured by the 005 are general indicators of job satis- faction. Hunter (1983, Note 1), in a reanalysis of the data used to develop the 005 (Hackman & Oldham, 1974), found that there was a second order factor model in which all of the subscales were indicators of one general factor. This could explain why one subscale Variable of the JDS was not effective in predicting a specific job attitude such as anxiety. Impact of Manipulation A second issue outside the model which is worth considering is one raised during preliminary data analysis. It was originally proposed that employees would respond to no information about the move (control condition) neutrally, i.e., more negatively than those receiving positive information and more positively than those receiving negative information. This was not the case. Individuals receiving no move-related information had perceptions of the information which were even more negative than the negative interpretation condition. The manipulation was recoded to reflect this discrepancy in the model. The implications for this result are clear. It appears that individuals 68 felt that any information -- even information which emphasized the negative .aspects of the move -- was more helpful than no information about the move. This emphasizes the high need employees apparently had to reduce uncertainty about the move to the new building. This result, then, could be limited to attitudes which, like anxiety, deal with reactions to the unknown. When dealing with attitudes about the Current work environment, negative informa- tion could well be perceived more negatively than no information, as there is less need to reduce uncertainty about the unknown. It should also be noted that a stronger negative manipulation (i.e., you're going to hate working in open offices) may have yielded different results. In this Case, the positive aspects of reducing uncertainty could be outweighted by the sheer negativity of the message. The low ranking of the "no information" group would be of little sur- prise to scholars in the area of diffusion of innovations. In considering resistance to change, these researchers point to the importance of a continued and repetitive flow of information about an innovation, even if some of that information could be perceived negatively. Jones (1972) has suggested that sheer education about the innovation will reduce resis- tance to change. Similarly, in a study of an educational innovation, Eicholz and Rogers (1972) concluded that a major reason for rejection of the innovation was ignorance. Limitations This study had several limitations which should be mentioned before managerial implications and directions for future research are explored. Several of these have been mentioned in previous sections. First, it should be emphasized that this study was a partial test of Salancik and Pfeffer's social information processing theory. The theory suggests that 69 job attitudes will be influenced to a great degree by an individual's own behavior. This study did not explore such a self-perception propo- sition. Second, organizational constraints necessitated only quasifrandom selection into experimental groups. Thus, it is possible that indivi- duals self-selected themselves into groups with friends or people they were similar to in terms of job level or task. These relationships could have influenced responses to questionnaire items in a way not related to the informational manipulations. Finally, social informa- tion processing theory is based on the notion of a constant informational environment created by coworkers. Clearly, the informational manipula- tions could not duplicate this environment, though an effort was made to make the manipulations strong and give the appearance of coworker support. Managerial Implications The results of this study have several implications for organizational managers interested in improVing employees' attitudes toward work or the organization or helping them adjust to organizational change. First, managers should be cognizant of the multiple determinants of job attitudes. They should be aware that feelings about work are a joint function of indi- vidual characteristics, objective job characteristics, and social information. The traditional remedy for dissatisfaction with work is job enrichment. How- ever, this study indicates that changes in the task may not be enough. Even taking the additional step of matching individuals to tasks, as the person- role fit literature would advocate, might not be effective if negative information about the job was still being received. Rather, managers should attempt a three way fit of individual needs, job characteristics, and infor- mation flow as a possible means of reducing anxiety and stress, improving worker satisfaction, or influencing other job attitudes. 70 Second, this research indicates that any information is better than no information, so managers facing change should try to provide employees with as much information as possible in an effort to reduce uncertainty about the unknown. Third, this study showed a very strong influence of privacy needs on anxiety. Managers should attempt to isolate needs relevant to the change and see that the needs are being met in the new situation and help employees to know that their needs are being considered. Managers should also be aware that individual needs can vary as a function of the job and should design change strategies accordingly. These suggestions may also apply to situations beyond the general organizational change situation considered in this study. On the indivi- dual level, organizational entry represents a time of maximal uncertainty. This study and social information processing theory suggest that reducing this uncertainty through information on needs, objective Characteristics of the job, and the interpretation of new experiences could ease the trauma of organizational entry. Louis (1980) makes a similar suggestion in her discussion of organizationa entry as'a "sense-making" activity. ...it seems particularly important for newcomers to have insiders who might serve as sounding boards and guide them to important background information for assigning meaning to events and surprises. Insiders are seen as a potentially rich source of assistance to newcomers in diag- nosing and interpreting the myriad surprises that may arise during their transition into new settings (p. 243). Directions for Future Research This study provided clear support for Salancik and Pfeffer's social information processing theory for the explanation of employee anxiety about organizational change. A model was proposed which provided a good fit to the data and which explained a great deal of the variance in the formation of anxiety about moving to an open office environment. The research also 71 suggests several avenues for future research which should be of interest to organizational communication scholars studying the impact of communicative activities on the formation of job attitudes. First, the findings of this research should be replicated with new data on employee anxiety. The original model in this study was revised in an attempt to find a better fit to the data. This modified model should be. validated with new data before it can be accepted with full confidence. Second, the social information processing theory should be tested with a variety of outcome variables to determine if the theory provides a viable explanation for the formation of a wide range of job-related attitudes. O'Reilly and Caldwell (1979) and White and Mitchell (1979) have looked at the theory in terms of traditional task-related attitudes. Work with other traditional outcome variables such as job satisfaction, organizational commitment, and job involvement could provide further support for Salancik and Pfeffer's theory. An extension of the model to the area of day-to-day job stress and, as mentioned earlier, organizational socialization, could also be valuable. Finally, the social information processing model could have valuable applications in the area of perceptions of and attitudes toward fellow workers. Research in this area could focus on both the work group and supervisor-subordinate relationships. In future studies of all these outcome attitude variables, care should be taken to insure the external validity of the research. The formation of work stress should be studied with employees facing the challenges of their every day work, not with college sophomores receiving conflicting or over- loading work assignemtsn in a two-hour task. Laboratory studies can be very valuable in providing control of extraneous variables and isolating research problems. However, conclusions about the effect of information in organi- zations can best be drawn from data considering communication in an organi- 72 zational setting. Future research in this area should also strive to consider elements of the work environment which are most relevant to the job attitude or outcome under investigation. The high cost of doing research in organizations has made scholars wary of collecting any data using untried measurement instru- ments. This is good; researchers should strive for valid and reliable measures of the variables in their theories and models. However, organiza— tional research has almost reached a point where the phrase "job character- istics" is synonymous with "autonomy, variety, task identity, interdependence, and feedback." This has proved to be a hindrance in the study of job atti- tudes, as there has been little search beyond these variables for factors relevant in the production of specific job attitudes. Researchers should give careful consideration to the problem at hand before using the Job Description Survey for the variable "job characteristics" in their theory. Two final areas of research could be very useful in developing and testing Salancik and Pfeffer's social information processing theory of .job attitudes. First, this study pointed to the contrast between accumulated and recently received information. Salancik and Pfeffer suggest that recent information will be most salient in influencing job attitudes, but this study casts some coubt on thatposition. Valuable insight could be drawn from a study which simultaneously tracked the introduction of informa- tion and the formation of job attitudes in an organization. Such research could address the critical questions of how much accumulated information is necessary to influence attitudes, how lasting the effects of social informa- tion are, and whether or not there is a lag time between the provision of information and a change in job attitudes. Finally, future research should attempt to explore the difference in the impact of formal and informal information. Salancik and Pfeffer's theory 73 emphasizes the importance of the informal communication among coworkers in influencing attitudes and gives little consideration to formal information provided to employees by management in an organization. Research comparing these two informational sources could help to isolate the qualities of infor- mation which make the largest impact on employees, and provide further insight into the process of job attitude formation in organizations. FOOTNOTES Analysis of variance comparing the experimental groups on the infor- mation helpfulness variable and the anxiety variable confirmed the results of this structural equation model. Results were significant for the information helpfulness variable (F = 3.87; p <1.001) and insignificant for the anxiety variable (F = 1.35; p = .24). 74 REFERENCE NOTES 1. Hunter, J. E.. From lecture on "Longitudinal Panel Analysis: Measure- ment Issues." Department of Psychology, Michigan State University, May 25, 1983. 75 REFERENCES Aldefer, C. P. Human needs in organizational settings. New York: The Free Press of Glencoe, 1972. Beehr, T. A., & Newman, J. E. Job stress, employee health, and organiza- tional effectiveness: A facet analysis, model, and literature review. Personnel Psychology, 1978, 31, 665-699. Bem, D. J. Self-perception theory. In L. Berkowitz (Ed.), Advances in experimental social psychology, Vol. 6. New York: Academic Press, 1972. Bentler, P. M., & Bonett, D. 6. Significance tests and goodness of fit in the analysis of covariance structures. ‘Psychological Bulletin, 1980, 88, 588-606. Berger, P. L., & Luckmann, T. The social constructiOn of reality. Garden City, N. Y.: Doubleday, 1967. Breslow, L., & Buell, P. Mortality from coronary heart disease and physical activity of work in California. Journal of Chronic Diseases, 1960, 11, 615-626. - Buck, V. Working Under pre55ure. London: Staples, 1972. Caplan, R. D., Cobb, 5., French, J. R. P. Relationships of cessation of smoking with job stress, personality and social support. Journal of Applied Psychology, 1975, 69, 211-219. Caplan, R. 0., Cobb, 5., French, J. R. P., Jr., Harrison, R. V., & Pinneau, S. R., Jr. Job demands and worker health:‘ Main effects and occupational differences. Ann Arbor, Mich.: Institute for Social Research, 1980. Cofer, C. B., & Appley, M. H. Motivation: Theory and research. New York: Wiley, 1964. Cooper, C. L., & Marshall, J. Occupational sources of stress: A review of the literature relating to coronary heart disease and mental ill health. Journal of OcCupational PSycholOgy, 1976, 49, 11-28. 76 77 Cummings, T. 6., & Cooper, C. L. A cybernetic framework for studying occupational stress. HUman Relations, 1979, 32, 395-418. Dreyfuss, F., & Czackes, J. W. Blood cholesterol and uric acid of healthy medical students under stress of examination. Archives of Internal Medicine, 1959, 193, 708. Eichholz, G.,'& Rogers, E. M. Resistance to the adoption of audiovisual aids by elementary schoolteachers: Contrasts and similarities to agricultural innovation. In G. Zaltman, P. Kotler, & I. Kaufman (Eds.), Creating social change. New York: Holt, Rinehart and Winston, 1972. Eisenberg, E. M., Monge, P. R., & Miller, K. I. Involvement in communication networks as a predictor of organizational commitment. Human Communication Research, 1983, in press. Endler, N. S., & Okada, M. A multidimensional measure of trait anxiety: The S-R inventory of general trait anxiousness. ‘Journal of Consulting and Clinical Psychology, 1975,'4§, 319-329. Endler, N. S., & Shedletsky, R. Trait vs. state anxiety, authoritarianism, and ego threat vs. physical threat. Canadian J0urnal of Behavioural Science, 1973, 5, 347-361. French, J. R. P., & Caplan, R. 0. Organizational stress and individual strain. In A. J. Marrow-(Ed.), The failure of Success. New York: AMACOM, 1973. Friedman, M. Rosenman, R. H., & Carroll, V. Changes in serum cholesterol and blood clotting time in men subjected to cyclic variations of occu- pational stress. Circulation, 1958, 17, 852-861. Hackman, J. R., & Lawler, E. E., III Employee reactions to job character- istics. Journal of Applied PSyphology, 1971, 55, 259-286. Hackman, J. R., & Oldham, G. R. Development of the job diagnostic survey. Journal of Applied PsychOlogy, 1975, 69, 159-170. Hodges, W. F. Effects of ego threat and threat of pain on state anxiety. Journal Of‘Personaljty and Social E§yCholpgy, 1975, 69, 159-170. Holbrook, M. 8., & Ryan, M. J. Modeling decision-specific stress: Some methodological considerations. AdministratiVe Science Quarterly, 1982, 27, 243-258. Hunter, J. E. Cluster analysis: Reliability, construct validity, and the multiple indicators approach to measurement. Unpublished manuscript, Department of Psychology, Michigan State University, 1977. Hunter, J. E., Cohen, S. H., & Nicol, T. S. PACKAGE: A system of routines to do correlational analysis, including path analysis, confirmatory factor analysis, and exploratory factor analysis. Unpublished manu- script, Department of Psychology, Michigan State University, 1982. 78 Hunter, J. E., & Gerbing, D. W. Unidimensional measurement, second order factor analysis, and Causal models, ReSearCh in OrganizatiOnal Behavior, 1982, 4, 267-320. Jablin, F. M. Organizational communication theory and research: An over- view of communication climate and network research. In D. Nimmo (Ed.), Communication yearbook 4. New Brunswick, N. J.: Transaction Books, 1980. Jenkins, C. D. Psychologic and social precursors of coronary disease. 959. England Journal of Medicine, 1971, 294, 244-255. Jenkins, C. 0. Recent evidence supporting psychological and social risk factors for coronary heart disease. New England Journal of Medicine, 1976, 294, 987-994, 1033-1038. ' Jones, G. N. Strategies and tactics of planned organizational change: Case examples in the modernization process of traditional societies. In G. Zaltman, P. Kotler, & I. Kaufman (Eds.), Creating social change. New York: Holt, Rinehart and Winston, Inc., 1972. Joreskog, K. G., & Sorbom, D. LISREL V: Analysis of linear structural relationShips by maximum likelihood and least square methods. Chicago: International Education Services, 1981. Kahn, R. L., Wolfe, D. M., Quinn, R. P., Snoek, J. D., & Rosenthal, R. A. Organizational stress. New York: Wiley, 1964. Kelley, H. H. Attribution theory in social psychology. In D. Levine (Ed.), Nebraska symposium 0n motivation. Lincoln, Nebraska: University of Nebraska Press, 1967. Kiesler, C. A., 8 Kiesler, S. C0nformity. Reading, Mass.: Addison-Wesley, 1969. Lazarus, R. S. Psychological stress and the coping process. New York: McGraw-Hill, 1966. Louis, M. R. Surprise and sense-making: What newcomers experience in enter- ing unfamiliar organizational settings. Administrative Science Quarterly, 1980, 25, 226-251. Margolis, B. K., Kroes, W. H., & Quinn, R. P. Job stress: An unlisted occupational hazard. Journal of OcCupational Medicine, 1974, 15, 659-661. Maslow, A. H. A theory of human motivation. Psychological Review, 1943, 59, 370-396. McGrath, J. E. Stress and behavior in organizations. In M. Dunnette (Ed.), Handbook of industrial and organizational psychology. Chicago: Rand McNally, 1976. 79 McPhee, R. D., & Babrow, A. S. Causal modeling in speech communication research: Use, disuse, and misuse. Paper presented at annual conven- tion of International Communication Association, Dallas, Texas, May, 1983. Naismith, D. C. Stress among managers as a function of organizational change (Doctoral dissertation, George Washington University, 1975). Dissertation Abstracts International, 1975, 36, 40l-A. Newman, J. E. Understanding the organizational structure-job attitude rela- tionship through perceptions of the work environment. Organizational Behavior and Human PerfOrmance, 1975, 44, 371-397. O'Reilly, C. A., & Caldwell, D. F. Informational influence as a determinant of perceived task characteristics and job satisfaction. Journal of Applied Psychology, 1979, 54, 157-165. Pincherle, G. Fitness for work. Proceedings Report of SoCial Medicine, 1972 g _6__5_, 321-324 . Pincus, S., & Waters, L. K. Informational social influence and product quality judgments. Journal Of Applied Psychology, 1977, 52, 615-619. Porter, L. W. Job attitudes in management: I. Perceived deficiencies in need fulfillment as a function of job level. Journal of Applied Psychology, 1962, 45, 375-384. Porter, L. W. Job attitudes in management: II. Perceived importance of needs as a function of job level. Journal of Applied Psychology, 1963, 42, 141-148. Porter, L. W., & Lawler, E. E. III Properties of organization structure in relation to job attitudes and job behavior. Psychological Bulletin, 1965, 54, 23-51. Quinn, R. P., Seashore, S., & Mangione, I. Survey of working conditions. Washington, D. C.: U.S. Government Printing_0ffice, 1971. Redding, W. C. Organizational communication theory and ideology: An overview. In D. Nimmo (Ed.), Communication yearbook 3. New Brunswick, N. J.: Transaction Books, 1979. Richetto, G. M. Organizational communication theory and research: An overview. In B. Ruben (Ed.), CommUnication yearbook 1. New Brunswick, N. J.: Transaction Books, 1977. Salancik, G. R. Commitment and control of organizational behavior and belief. In B. M. Staw and G. R. Salancik (Eds.), New directions in organizational behavior. Chicago: St. Clair Press, 1977. Salancik, G. R., & Pfeffer, J. An examination of need-satisfaction models‘ of job attitudes. Administrative Science Quarterly, 1977, 22, 427-456. 80 Salancik, G. R., & Pfeffer, J. A social information processing approach to job attitudes and task design. Administrative Science Quarterly, 1978, 23, 224-253. Schuler, R. S. Definition and conceptualization of stress in organizations. Organizational Behavior and Human Performance, 1980, 25, 184-215. Selye, H. The stress of life (Rev. Ed.). New York: McGraw-Hill, 1976. Selye, H. Stress without distress. New York: J. B. Lippincott, 1980. Shirom, A., Eden, 0., Silberwasser, S., & Kellerman, J. J. Job stresses and risk factors in coronary heart disease among occupational categories in kibbutzim. SoCial Science Medicine, 1973, 2, 875-892. Wardwell, W. I., Hyman, M. M., & Gahnson, C. B. Stress and coronary disease in three field studies. Journal of Chronic Diseases, 1964, 24, 453-468. Weick, K. The socialpsychology of organizing (2nd Edition). Reading, Mass.: AddiSon-Wesley, 1979. White, S. E., & Mitchell, T. R. Job Enrichment versus social cues: A comparison and competitive test. J0urna1 of Applied PsyCholOgy, 1979, 64, 1-9. (1) (2) (3) (1) (2) (3) (1) (2) (3) APPENDIX A QUESTIONNAIRE ITEMS Anxiety I feel anxious about the move to the new building. I am not concerned about working in open offices. The thought of working in an open office worries me. Need for Privagy» I often have to discuss private matters at work without being overheard. It's not important that I have privacy to get my job done. I need a quiet work area to get my job done. Need for Interdependence I often have to meet or check with other people in the department in order to do my job. I often have to cooperate directly with other people in the department in order to do my job. My job requires me to work closely with others employed at MDE. PreviOus Information The information I have received about the Ottawa Street Building has: been timely been useful c. adequately answered my questions about the move 81 82 Information Helpfulness (l) The information provided was: a. positive b. favorable c. appropriate for publication in FYI APPENDIX B INFORMATIONAL MANIPULATIONS ORAL AND WRITTEN Positive’Needs WRITTEN INFORMATION For example, research shows that open communication is important among offices within a large organization. Different work environments can help the flow of important communication among people in work groups and between work groups which must work together. Our interviews with MDE employees show that people working at the Department of Education want to have easier access to people and information in other service areas. ORAL INFORMATION One thing we've been trying to do with the results of our research is provide information throughout the department about what to expect in the move. This information will be distributed through FYI in a column that the research team is preparing. A variety of move-related issues will be covered such as open office etiquette and communication. We're presenting some of these ideas to groups around the department before publication in the newsletter. One crucial issue related to the move is flow of information. Researchers in the field of organizational behavior and organizational communication have long recognized the importance of a free flow of work information. This access to information can influence both how effective you are in getting your work done and how satisfied you are with your work. ‘ There are two ways in which the flow of information is important to pe0p1e on the job. First, it's important for individuals within a work group to have easy access to information. In studies of communication networks, individuals who are well-connected into a network of work communication tend to be more productive and more satisfied. The work group as a whole also needs to be well-connected with other work groups -- a group which has ready access to other people and other information in the organization will probably be more effective. 83 84 Interviews with departmental employees clearly indicate that flow of infor- mation is an important feature of an office environment. Employees in a variety of service areas and at many levels feel a need for free access to information, both for the individual and for the work group as a whole. PositiveiDesCription WRITTEN INFORMATION For example, research shows that open offices help to increase the amount of job-related communication among employees. In an open environment, there is greater access to information needed to complete tasks, both within the work group and between work groups which must work together. Our interviews with MDE employees who have worked in open offices also point to the increased level of communication in an open office environ- ment. ORAL INFORMATION One thing we've been trying to do with the results of our research is provide information throughout the department about what to expect in the move. This information will be distributed through FYI in a column that the research team is preparing. A variety of move-related issues will be covered such as open office etiquette and communication. We're presenting some of these ideas to groups around the department before publication in the newsletter. One crucual issue related to the move is flow of information. This is an issue which has been studied a great deal by researchers in organiza- tional behavior and organizational communication. Researchers have looked at the kinds of information people need to send and receive, the effect of different levels of information flow, and types of office environment which facilitate information flow. It is clear that an open office facilitates the flow of information in an organization. Closed offices are replaced by partitioned work space which makes communication among employees simpler. Communication is improved both in terms of easing the flow of work-related information among employees and in maintaining more personal ties among people who work together. There are two ways in which the move to the new office will improve your access to people and information at MDE. The largest change will be yoUr access to people in other service areas. Employees who were housed in 17 buildings will now be under one roof, making meetings and general informa- tion exchange much simpler. It is also likely that communication within the work group will improve as a function of the open office environment. Many employees at the Department of Education have worked in open offices in the past. Interviews with these employees support the findings of academic researchers. An open office environment improves the flow of information in an organization -- both between work groups and among indi- viduals you work with every day. 85 Positive’Interpretation WRITTEN INFORMATION For example, research shows that open offices help improve the flow of communication that is necessary to get the job done. A work environment should provide easy access to information in each work group and between groups which must work together; an open offices provides such an environ- ment. Our interviews with MDE employees show that improved information flow is a clear advantage of the new open office environment. ORAL INFORMATION One thing we've been trying to do with the results from our research is provide information throughout the department about what to expect in the move. This information will be distributed through FYI in a column that the research team is preparing. A variety of move-related issues will be covered such as open office etiquette and communication. We're presenting some of these ideas to groups around the department before publication in the newsletter. One crucial issue related to the move is flow of information. Researchers in the field of organizational behavior and organizational communication have long recognized the importance of a free flow of work information. This access to information can influence both how effective you are in getting your work done and how satisfied you are with your work. For example, in studies of communication networks, individuals who are well- connected into a network of work communication tend to be more productive and more satisfied. The same thing holds for work groups as a whole -- a group which has ready access to other people and other information in the organization will probably be more effective. The move to the new office building will be a big change in terms of access to information and people. The largest change will be your access to people in other service areas. Workers who were housed in 17 different buildings will now be under one roof, making meetings and general informa- tion exchange much simpler. It is likely that communication within the work group will also increase as a function of the open office environment. Interviews with departmental employees clearly indicate that greater access is seen as a major advantage to the move. This view was echnoed through- out the Department. "I won't spend so much time traveling." "It will be nice to be together." "Communication should definitely improve." Although there are worries about the move, employees at all levels of the NOE are looking forward to the freer flow of information in the new building. Negative Needs WRITTEN INFORMATION For example, research shows that privacy and confidentiality of information are very important in a large organization. A worker who can talk about 86 work-related matters without the threat of being overheard will probably be more productive and satisfied. Our interviews with MDE employees show that people in many service areas feel that privacy is a crucial aspect' of the job. ORAL INFORMATION One thing we've been trying to do with the results of our research is provide information throughout the department about what to expect in the move. This information will be distributed through FYI in a column that the research team is preparing. A variety of move-related issues will be covered such as open office etiquette and communication. We're presenting some of these ideas to groups around the department before publication in the newsletter. One crucial issue related to the move is privacy. Researchers in the fields of organizational behavior and organizational communication have long recognized that many jobs and workers require a certain level of privacy in order to accomplish their tasks. The level of privacy in an office can influence both how effective workers are in accomplishing tasks on the job, and how satisfied they are with their work and the organization they work for. Of course, different jobs require different amounts of privacy. A psychologist probably feels a much greater need for privacy than workers on an assembly line. There are two types of privacy that many people feel they need on the job. The first of these simply involves the level of noise and ability to see others in an office. Many workers feel that they need aural and visual privacy in order to do their jobs. Seeing and hearing other employees working nearby can be distracting and can make some people less effective in their work. A second way in which many people feel they need privacy on the job is in terms of confidentiality of conversations. For many jobs, it is important to be able to talk to clients and coworkers without the fear of being overheard. The comfort of employees and people outside the organization can be influenced by lack of privacy in an office. Interviews with departmental employees clearly indicate that privacy is an important aspect of work. Employees in a variety of service areas and at many levels feel a need for privacy. This need is particularly pronounced in service areas where confidential relationships with clients are a "tradition" in the field. These people, in addition to desiring aural and visual provacy, want an office where confidential conversations can be held without fear or worry. Neggtive‘Description WRITTEN INFORMATION For example, research shows that there is very little privacy available in many open offices. In an open-scape office, an employee can rarely shut the door for confidential conversations with fellow employees and clients. Our interviews with MDE employees who have worked in open offices also suggest that there is a low level of privacy available in an open 87 office environment. ORAL INFORMATION One thing we've been trying to do with the results of our research is provide information throughout the department about what to expect in the move. This information will be distributed through FYI in a column that the research team is preparing. A variety of mve-related issues will be covered such as open office etiquette and communication. We're presenting some of these ideas to groups around the department before publication in the newsletter. One crucial issue related to the move is privacy. Researchers in the field of organizational behavior and organizational communication have studied the amount of privacy available in different types of office environments and considered the impact that the am0unt of privacy available might have on the way work is done. It is clear that an open office limits the amount of privacy available to employees. Closed offices are replaced by partitioned work space where doors cannot be closed. It is much easier to hear and see other people working in work spaces nearby. There are two ways in which the move to.the new office will effect the amount of privacy available to you. First, it is likely that you will have a lower level of visual and aural privacy available to you. Noise levels will pr0bab1y be higher throughout the office, and you will be able to see others working more easily -- just as they will be able to see you. Second, there will be less opportunity in an open office for confidential conversations. The partitioned style of the office will not allow for con- fidential conversations between coworkers or with clients and others ‘ outside of the organization. Many employees at the Department of Education have worked in open offices in the past. Interviews with these employees support the findings of academic researchers. An open office environment provides workers with less privacy -- both in terms of aural and visual privacy and in terms of the confidentiality of conversations. Negative Interpretation WRITTEN INFORMATION- For example, research shows that open offices often limit the amount of privacy available to employees. Many jobs require that conversations and records be kept in strict confidence; an open office can make this difficult. Our interviews with MDE employees have indicated that the potential lack of privacy is a major worry about the new open office environment. ORAL INFORMATION One thing we've been trying to do with the results of our research is provide 88 information throughout the department about what to expect in the move. This information will be distributed through FYI in a column that the research team is preparing. A variety of move-realted issues will be covered such as open office etiquette and communication. We're pre- senting some of these ideas to groups around the department before publication in the newsletter. One crucial issue related to the move is privacy. Researchers in the fields of organizational behavior and organizational communication have long recognized that many jobs and workers require a certain level of privacy in order to accomplish their tasks. This privacy is necessary for two.reasons: First, so employees can concentrate on their work,.and second, so employees can hold confidential conversations regarding their work. The level of privacy in an office can effect both how effective workers are in accomplishing their tasks and how satisfied they are with their jobs and the organization they work for. Of course, different jobs require different levels of privacy. For example, a psychologist has a high need for privacy in the work place, but privacy is not so crucial for workers on an assembly line. The move to the new office building will be a big change in terms of privacy. Open offices are characterized by open spaces and partitions rather than floor to ceiling offices which can be closed off from other workers. Be- cause of this change in environment, many workers will have to make adjust- ments in the way they do their work and the amount of privacy they can expect in the workplace. Interviews with departmental employees clearly indicate that the level of privacy available at the new building is a major concern. Employees in a variety of service areas at a variety of levels are worried that they won't be able to "close the door" and that they will not have enough privacy to complete work as efficiently as in old work space. Adjustments in the way employees work are to be expected in a new environment, and for employees at MOE, the major area in which changes should be expected from the move is privacy. Control Condition WRITTEN INFORMATION For example, previous phases of this study have already been used in terms of determining issues that are important to MDE employees. Results from our interviews with employees and from surveys have already been used to formulate informational articles in FYI. In addition, sessions have been held with service areas around the MDE to discuss results from specific work groups.