Article Critique -three pages APA Format in-text citations

401

Active on the Job—Proactive in Change How Autonomy at Work Contributes to Employee Support for Organizational Change

Severin Hornung Technical University of Munich

Denise M. Rousseau Carnegie Mellon University

Two separate samples of a hospital’s employees (N = 166 and 133) obtained at different points in time are used to test a structural equation model of the developmental and socializing effects autonomy at work has on employees’ proactivity and its resultant impact on their support for organizational change. In both samples, autonomy positively affected employee role breadth self-efficacy and personal initiative, which in turn had positive though differential relationships with employee responses to change. The model fitting each cross-sectional sample is supported in a third separate, longitudinal sample (N = 74) of participants in both surveys. Results indicate that the proactivity characteristic of workers enjoying on-the-job autonomy promotes their positive responses to structural change. Promoting worker autonomy itself can be a critical precursor to successful implementation of certain forms of organizational change.

Keywords: organizational change; job autonomy; proactive behavior; socialization; change support

A nontraditional style of active employee behavior characterizes effective con- temporary firms. This behavior is proactive, problem oriented, flexible, and cus- tomer focused (Crant, 2000; Frese & Fay, 2001; Kanter, 1993; Parker, 2000). It transcends compliance with orders, formal rules, and narrow roles, contrasting with long-standing corporate notions of more constrained employee conduct (Whyte,

THE JOURNAL OF APPLIED BEHAVIORAL SCIENCE, Vol. 43 No. 4, December 2007 401-426 DOI: 10.1177/0021886307307555 © 2007 NTL Institute

1956). Worker self-determination and initiative underlie this active style as evident in such contemporary workplace concepts as psychological empowerment (Spreitzer, 1996), organizational citizenship (Organ, 1988), contextual performance (Motowidlo & Van Scotter, 1994), taking charge (Morrison & Phelps, 1999), con- structive deviance (Galperin, 2005), personal initiative (PI; Frese & Fay, 2001), flex- ible work orientations (Parker, Wall, & Jackson, 1997), and role breadth self-efficacy (RBSE; Parker, 1998). Such active behavior is particularly critical to organizations implementing structural changes. In such conditions, employees need to revise exist- ing psychological schemata (Bartunek & Moch, 1987; Rousseau, 2001) and partici- pate in the remaking of their work roles and organizational practices.

Despite the prevalence of change in contemporary organizations, only recently have empirical studies addressed such change-promoting employee beliefs and behaviors as adaptivity (Griffin, Neal, & Parker, 2007), program commitment (Jansen, 2004; Neubert & Cady, 2001), commitment to change (Cunningham, 2006), openness to change (Wanberg & Banas, 2000), readiness to change (Armenakis, Harris, & Mossholder, 1993), and change-oriented citizenship behavior (Choi, 2007). Although the socializing effects of everyday work experiences on the development of proactive work behaviors are well established (e.g., Frese, Kring, Soose, & Zempel, 1996; Parker, 1988), their role in change has received limited attention. Instead, change studies largely draw on individual differences and contextual factors such as leadership, teamwork orientation, or social climate to explain employee support (e.g., Choi, 2007; Neubert & Cady, 2001; Wanberg & Banas, 2000).

The present study seeks to integrate two research streams, the proactivity-enhancing effects of work characteristics and positive employee responses to change. Using separate independent samples obtained within the same firm at different points in time, we replicate and expand research on the role workplace autonomy plays in shaping personal initiative and self-efficacy and examine their resultant contribution to employee support for organizational change. Following previous recommenda- tions for studying employee reactions in organizational transitions (Piderit, 2000; Wanberg & Banas, 2000), we distinguish between two employee change responses, anticipated benefits from the change and commitment to actively support it. In turn, we examine the mutual and differential effects of the two forms of proactivity on these change responses. The model’s hypothesized relationships are then examined in a third separate, longitudinal sample providing data from the same organization at both measurement points. This study differs from previous research by investigating

402 THE JOURNAL OF APPLIED BEHAVIORAL SCIENCE December 2007

Severin Hornung is a researcher at the Institute of Psychology of the Technical University of Munich, Germany, where he received his master’s and doctoral degrees. His research interests include extrarole behavior at work, job characteristics, and organizational change.

Denise M. Rousseau is the H.J. Heinz II Professor of Organizational Behavior and Public Policy at Carnegie Mellon University’s Heinz School of Management and Tepper School of Business. She received her AB, MA, and PhD from the University of California at Berkeley with degrees in psychology and anthropology.

This research was funded by an H.J. Heinz II professorship. We thank Marge Hardt and Jürgen Glaser for their contributions to this project.

the empirical distinctiveness and functional relationship of critical facets of worker proactivity previously only considered separately. Most important, it studies their dynamics and relative contribution to employee responses in the context of organiza- tional change, where active worker support can be vital for the change’s success.

Autonomy as a Core Dimension of Work

Autonomy on the job is perhaps the central work characteristic in shaping worker attitudes, motivation, and behavior (Fried & Ferris, 1987; Hackman & Oldham, 1975). Defined as “the degree to which the job provides substantial freedom, inde- pendence, and discretion to the employee in scheduling the work and in determining the processes to be used in carrying it out” (Hackman & Oldham, 1980, p. 79), autonomy promotes worker responsibility and acceptance of accountability. High decision latitude, a concept very similar to autonomy (Ganster, 1989; Smith, Tisak, Hahn, & Schmieder, 1997), contributes to the development of new, more active behavior patterns (Karasek, 1979, 1989; Karasek & Theorell, 1990). Low decision latitude, on the other hand, reduces ability to cope with stressors, decreases overall activity and problem-solving attempts, resulting in learned helplessness (Seligman, 1975). Organizationally induced helplessness can create performance problems via the passive and maladaptive behaviors low autonomy engenders (Martinko & Gardner, 1982). In a study in a manufacturing setting, autonomy led to more flexi- ble, strategic, and proactive work orientations, which researchers attributed to its intervening effects on employee learning and expanded behavioral repertoires (Parker et al., 1997). To the extent that autonomy motivates workers to undertake more responsibility and initiative, its effects should be visible in the two commonly studied facets of proactivity, PI and RBSE.

Employee Proactivity: Personal Initiative and Role Breadth Self-Efficacy

The basis of employee proactive behavior is a central concern in contemporary firms and in theories of workplace behavior. Employee proactivity is typically con- ceptualized in two ways. First, as a behavioral pattern, PI is characterized by the self- starting, proactive, and persistent pursuit of long-term personal and organizational goals (Frese, Fay, Hilburger, Leng, & Tag, 1997). Persons high on PI engage in inno- vation such as improving organizational procedures. PI differs from other concepts related to self-initiated active worker behavior such as organizational citizenship (Organ, 1988): PI goes beyond mere overfulfillment of existing roles or organiza- tional values (Frese & Fay, 2001; Parker, 1998), manifesting in novel behaviors in pursuit of goals. Although PI has been conceptualized as a relatively stable behavior pattern, as opposed to a situational response, evidence indicates that it is subject to effects from socialization and on-the-job experiences. Job control has been found to engender PI by creating opportunities for decision making, responsibility, and enhanced self-efficacy (Fay & Frese, 2001; Ohly, Sonnentag, & Pluntke, 2006; Speier & Frese, 1997). Similarly, a longitudinal study in East and West Germany demonstrated that occupational socialization promoting job autonomy led to greater worker PI (Frese et al., 1996).

Hornung, Rousseau / AUTONOMY AT WORK 403

Second, as an individual belief-shaping proactive behavior, RBSE is a specific form of self-efficacy (Bandura, 1997). Reflecting the influence of both personality and situational factors, RBSE is “the extent to which people feel confident that they are able to carry out a broader and more active role, beyond traditional prescribed tech- nical requirements” (Parker, 1998, p. 835). Similar to Frese’s reasoning for PI (Frese et al., 1997; Frese & Fay, 2001), Parker (1998, 2000) argues that RBSE results from both the individual’s personality as well as the socializing power of organizational experiences. RBSE increases as a result of autonomy, control, and higher possibilities for influence on the job (Axtell & Parker, 2003; Galperin, 2005; Morgeson, Delaney-Klinger, & Hemingway, 2005; Parker, 1998).

Together, studies of PI and RBSE emphasize the susceptibility of worker self- determination to workplace developmental and socializing experiences. Autonomy and related beliefs regarding on-the-job control and responsibility enhance worker confidence in taking on a broader job role, behaving in novel and innovative ways, in the self-starting pursuit of goal accomplishment. In the present study of proactiv- ity in the context of organizational change, we expect to replicate the findings of pre- vious research treating PI and RBSE separately:

Hypothesis 1a: Autonomy at work will have a positive effect on PI. Hypothesis 1b: Autonomy at work will have a positive effect on RBSE.

We also extend previous research by investigating their interrelationship, antecedents, and consequences. As products of workplace developmental and social- izing experiences, PI and RBSE represent two facets of employee proactivity expected to be empirically related yet conceptually distinct. The nature of this rela- tionship can be inferred from the underlying composition of each construct. Although PI is conceptualized as a behavior, persistent action to accomplish both organizational and personal goals, RBSE is a self-perception, the confidence a worker places in his or her own efforts to perform in ways exceeding formal duties. Insofar as beliefs regarding one’s capacity to demonstrate a particular behavior tend to increase the likelihood of its occurrence (Vroom, 1964), it is no surprise that gen- eral self-efficacy has been found to be an antecedent of PI (Frese & Fay, 2001; Speier & Frese, 1997). We postulate that a parallel effect exists where RBSE increases the PI workers manifest:

Hypothesis 2: RBSE will have a positive effect on PI.

Support for an Organizational Change

Employee support is a prerequisite for many types of successful change (Kanter, 1983; Piderit, 2000). Focusing on employee support as an antipode to the somewhat blurry and, according to several authors, outdated concept of resistance (Dent & Goldberg, 1999; Merron, 1993; Piderit, 2000), Herscovitch and Meyer (2002) define commitment to change as a mind-set “that binds individuals to a course of action deemed necessary for the successful implementation of a change initiative” (p. 475). According to Wanberg and Banas (2000), openness to change, a similar concept,

404 THE JOURNAL OF APPLIED BEHAVIORAL SCIENCE December 2007

comprises both the willingness to support change as well as positive affect toward the change’s outcomes. However, as openness is a weak expression for genuine pos- itive attitudes, we prefer the term employee support to characterize positive beliefs regarding the change’s potential outcomes and behavioral intentions to actively par- ticipate in it (Armenakis et al., 1993; Miller, Johnson, & Grau, 1994).

Cognitive appraisal of a change’s potential consequences, that is, gains and losses, is an especially vital aspect of employee support and an important precursor of intentions to actively partake in the change (Bartunek, Rousseau, Rudolph, & Depalma, 2006; Rousseau, 1995). When employees believe they have nothing to gain from an organizational transition, denying support is a rational strategy. Following Piderit’s (2000) recommendations for studying employee responses to change as multidimensional attitudes with cognitive, emotional, and intentional components, the present study differentiates behavioral commitment to support the change from the cognitive–emotional evaluation of its anticipated benefits, examining how each relates to employee proactive orientations.

Uncertainty is inherent in organizational change, both in terms of how workers should act with respect to the change and the outcomes they anticipate from it (Rousseau, 1995). Changes that challenge existing authority or role structures create ambiguities regarding appropriate and effective action (Wheatley, 1992), requiring enactment to transform old patterns into a new order. This enactment comes about as employees make sense of and act in ways that uncover previously unrecognized opportunities for flexibility and create new courses of action, which in turn trans- form organizational processes and structures (Weick, 1995). Such opportunities are particularly likely in the context of structural change such as that studied here.

Structure-based changes include interventions introducing decentralization, employee participation structures, or a new compensation system (Legge, 1984). It contrasts with other change efforts where top management might decide to enter a new market (i.e., strategic change) or build another facility in a new location (i.e., capacity change) without necessarily affecting current employees. In many cases, structural change is undertaken to enhance interdependent outcomes such as cus- tomer service and employee development. Where structural change targets employee quality of work life as well as business goals, workers are more likely to view change as positively aligned with their own interests (e.g., offering them opportunities to shape their work environment to better suit personal needs; Hackman, 1970; Wrzesniewski & Dutton, 2001).

Although every structural change may not be positive from an employee per- spective, any successful implementation ultimately alters worker in-role behavior. Structural changes can be difficult to implement all at once, making supportive and sustained employee action critical to their success (Weick & Quinn, 1999). Employees whose work roles have prepared them to behave proactively are better positioned to respond to structural change’s ambiguities and opportunities than are those less used to proactivity. Where the particular implementation of structural change has the potential to serve employee interests as well as the firm’s, employ- ees’ autonomy at work can contribute substantially to their effective change support.

Hornung, Rousseau / AUTONOMY AT WORK 405

Worker initiative takes the form of active change support when it incorporates the change’s features into the adoption of a broader role than current job descriptions specify (Morrison & Phelps, 1999). In an organizational intervention to promote customer satisfaction, for example, an employee might look for new ways to respond to customer needs. Frese and colleagues (1996) emphasize PI’s relevance to organi- zational transitions, where workers high on PI are more likely than their low PI counterparts to proactively support a constructive change and its goals. Similarly, insofar as workers confident in their ability to go beyond their traditional roles are better positioned to support an organizational change and realize its benefits, RBSE should also promote employee change support. RBSE has been found to promote constructive deviant behavior, that is, acts of an innovative and creative nature, chal- lenging existing norms and rules in a manner beneficial to the organization (Galperin, 2005). Given RBSE’s and PI’s postulated positive role in change, we hypothesize that both facets of proactivity will be positively related to change sup- port and perception of its benefits.

Hypothesis 3a: RBSE will have a positive effect on commitment to support change. Hypothesis 3b: RBSE will have a positive effect on anticipated benefits from change. Hypothesis 4a: PI will have a positive effect on commitment to support change. Hypothesis 4b: PI will have a positive effect on anticipated benefits from change.

Despite the parallel effects postulated earlier, differences in the composition of RBSE and PI also suggest that they function somewhat distinctly in an organizational transition. Specifically, we postulate that the goal attainment versus role-expanding instrumentality of PI and RBSE imply different emphases on the two components of attitudes toward organizational change. As a behavior pattern involving the active and persistent pursuit of goals, PI is closely related to situational control. Emphasizing PI’s goal-oriented nature, Frese (1989) defines control as being able to “have an impact on the conditions and on one’s activities in correspondence with some higher- order goal” (p. 107). Perceived controllability, associated with high levels of PI, rep- resents a coping resource, which leads to a more positive cognitive evaluation of situations, framing them as an opportunity for realizing beneficial outcomes rather than a threat (Lazarus & Folkman, 1984). In an organizational transition, high levels of PI imply expectations of turning the situation to a personal advantage and finding ways of accomplishing personal as well as organizational goals (Frese et al., 1996). Perceiving themselves as being able to actively influence change outcomes in benefi- cial ways, individuals with high levels of PI tend to adopt a favorable view of struc- tural change that promotes employee interests, interpreting the transition as an opportunity for improvement. Due to its goal- and control-oriented focus, PI is expected to relate more strongly to an optimistic view regarding a change’s antici- pated benefits than to commitment to support the change for its own sake. Note, how- ever, that when a change clearly threatens employee interests, we might expect high PI people to be among its most active opponents.

In contrast, RBSE is by definition role oriented, reflecting personal confidence in the ability to take on duties that increase the responsibility and complexity in an

406 THE JOURNAL OF APPLIED BEHAVIORAL SCIENCE December 2007

individual’s role. As active participation in an organizational change represents an exemplary form of broader role adoption, high RBSE is likely to promote acceptance of changes that provide such opportunities. As RBSE lacks the outcome-oriented nature of PI, however, we postulate that it will be more strongly related to commit- ment to participate in the change as a role-expanding process and less strongly asso- ciated with a positive evaluation of change’s other potential outcomes.

In sum, PI and RBSE vary in the extent that they are outcome versus process focused. Thus, we hypothesize that the strength of their effects on employee change support differs between the support facets, change commitment and expected bene- fits, studied here.

Hypothesis 5a: The direct effect of PI on anticipated benefits of the change will be stronger than its effect on commitment to support the change.

Hypothesis 5b: The direct effect of RBSE on commitment to support the change will be stronger than its effect on anticipated benefits of the change.

On a theoretical level, empirically confirming these postulated differential effects would provide a strong argument against any conceptual redundancy between PI and RBSE, a common problem in organizational research (e.g., Morrow, 1983; Van Dyne, Cummings, & Parks, 1995). From a practical point of view, it would indicate that change support on the part of individuals with higher PI or RBSE is motivated by different reasons (cf. Rousseau & Tijoriwala, 1999).

Finally, with regard to the two facets of change support, we hypothesize a posi- tive effect of expected benefits on intentions to support the change. This typical instrumentality hypothesis (Vroom, 1964) is in line with previous work on the importance of anticipated outcomes for promoting employee support of organiza- tional change (e.g., Bartunek et al., 2006; Piderit, 2000).

Hypothesis 6: Anticipated benefits of the change will have a positive effect on commitment to support change.

METHOD

Organizational Setting

The setting is a private 154-bed general hospital with approximately 350 employees located in the Northeastern United States. About 18 months before the first survey, a structural change in the form of a Shared Leadership intervention (Porter-O’Grady, 1992) was initiated by the hospital’s top management. This change program was initiated by the hospital’s chief operating officer (and later CEO) in response to a regionally competitive health care market. Program goals were to gen- erate competitive advantage through improved quality of care, patient satisfaction, employee morale, and professionalism by decentralizing decision making to the point of service and creating a more supportive and empowering work environment.

Prior to the change, the hospital was organized functionally and largely managed in a centralized, top-down fashion. Nonunionized, it had few formal channels for employee input or participation in decision making. Supervisors had little previous

Hornung, Rousseau / AUTONOMY AT WORK 407

managerial training, typically rising from the ranks in their particular department and directing the activities in their departments based on hospital rules and job spec- ifications. Little cross-departmental communication or cooperation existed, unless related to direct patient care.

Increased employee support and participation in decisions formerly made by managers were implemented via several interrelated efforts: hospital-wide advisory councils, comparable department-based committees, and worker and supervisory training focused on improved problem-solving processes. One hospital-wide coun- cil focused on clinical issues (e.g., quality of care) and another on work–life issues (e.g., employee rewards). In each hospital-level council, elected employee represen- tatives from each department meet monthly to discuss current problems and make recommendations for improvements. Creation of these councils was accompanied by training and development activities to help workers more effectively share infor- mation, participate in decisions, and solve problems. Employees who were not council members could participate by making suggestions, providing input into council deliberations, becoming involved in departmental committees, and becom- ing more active in on-the-job problem solving. From our own observations, the councils acted as both decision-making bodies as well as a clearinghouse, channel- ing information, feedback, and change modification decisions throughout the hospi- tal. The presence of hospital leadership on the councils served as a “power promoter” (Hauschildt & Kirchmann, 2001), broadening the impact of employees on the change’s content and process. In this vein, the Shared Leadership councils rep- resented a form of employee comanagement, resembling the legally institutionalized “works councils” found in many European countries (e.g., Rogers & Streeck, 1995).

Samples

Our samples comprise health care personnel (e.g., nurses, physical therapists), technical personnel (e.g., radiology technicians), administration (e.g., clerical staff, managers), and workers in dietary, security, and other support areas in this hospital. For the first sample, 166 out of 350 questionnaires were returned, representing a 47% response rate. The second sample was obtained from the same organization approximately 18 months after the first survey. During this time the hospital had annual turnover of approximately 20% while its staffing level had increased to 400 employees. In total, 207 participants (approximately 52% of the whole staff) com- pleted questionnaires at Time 2. In both surveys, respondents were asked to provide the last four digits of their social security number as a matching code. More than 90% (150) of the participants in the first survey and 95% (196) in the second survey complied with this request. Initial matching using this identification generated lon- gitudinal data for 65 respondents. Comparing departmental and demographic infor- mation (e.g., gender, age, tenure, education) and handwriting, another 9 participants were independently identified as repeat participants by two separate coders, and the matching was validated by a third researcher. After careful examination of the data and verifying social security numbers with information provided by the human resource department, we concluded that 133 persons participated only in the second survey, whereas 74 respondents had taken part in both surveys.

408 THE JOURNAL OF APPLIED BEHAVIORAL SCIENCE December 2007

To keep the two cross-sectional samples independent, repeated participants were excluded from Sample 2; instead, they comprised Sample 3, a longitudinal panel of 74 participants. Demographic data for the three samples are displayed in Table 1. Chi-square tests compared the categorical demographic variables among the three samples. The only statistically significant difference exists in terms of organizational tenure, with employees participating in both samples having a higher tenure at Time 2 than those who only participated in the second survey. Excluding those cases from Sample 2 therefore leads to Sample 2 showing a lower average tenure than Sample 1. Apart from tenure, the samples were demographically comparable. In addition, at both Time 1 and Time 2, respondents and nonrespondents were compared using hos- pital archival records. No statistically significant differences were detected.

Measures

Identical measures were used in all samples. All measures except for the demo- graphics were 5-point Likert-type scales ranging from not at all (1) to to a very great extent (5).

Autonomy. Autonomy was measured using the three-item scale from the Job Diagnostic Survey (Hackman & Oldham, 1980). Sample items are “The job gives me considerable opportunity for independence and freedom in how I do the work” and “My job permits me to decide on my own how to go about doing the work.”

Personal initiative. PI was assessed with Frese et al.’s (1997) seven-item scale tapping proactive behaviors. Sample items are “Whenever something goes wrong, I search for a solution immediately” and “I use opportunities quickly in order to attain my goals.”

Role-breadth self-efficacy. RBSE was measured using Parker’s (1998) 10-item scale assessing confidence in one’s ability to demonstrate change-related behaviors. Items refer to individual confidence in representing the work unit in meetings, help- ing to set target goals, contributing to the strategy of the organization, and so on.

Hornung, Rousseau / AUTONOMY AT WORK 409

TABLE 1

Demographic Properties of the Three Samples

Sample 1 Sample 2 Sample 3 (N = 166) (N = 133) (N = 74)a

Female 146 (88.0%) 112 (84.2%) 62 (83.8%) Median age 46 to 50 years 41 to 45 years 46 to 50 years Median tenure 6 to 10 years 4 to 5 years 6 to 10 years High school 43 (25.9%) 36 (27.1%) 17 (23.0%) Registered nurse diploma 23 (13.9%) 15 (11.3%) 9 (12.2%) Associate degree 46 (27.7%) 46 (34.6%) 26 (35.1%) Bachelor degree 42 (25.3%) 19 (14.3%) 18 (24.3%) Master’s degree 8 (4.8%) 10 (7.5%) 1 (1.4%)

a. Referring to measurement Point 2.

Sample items are “I would feel confident designing new procedures for my depart- ment” and “I would feel confident analyzing a long-term problem to find a solution.”

Anticipated benefits. Expected benefits from the change were measured using three items developed specifically for the context of our study based on discussion with top management and council members: “Shared Leadership will make it more pleasant to work at [organization],” “Shared Leadership will enable me to participate in decisions formerly made by management,” and “I will have more control over my work environment through Shared Leadership.”

Change commitment. Commitment to support and participate in the change was assessed using three items developed via discussion with top management and coun- cil members: “I am personally committed to bringing issues to the attention of the Councils,” “I am personally committed to speaking up at the Councils when requested,” and “I am personally committed to discussing my work–life issues with Council members.”

Demographics. Items with prespecified categories were used to assess gender, age (11 categories, ranging from younger than 21 to older than 65), organizational tenure (9 categories each, ranging from a year or less to more than 30 years), and education (high school, registered nurse diploma, associate, bachelor, or master’s degree).

RESULTS

Scale Analyses

With Cronbach’s alphas ranging from .75 to .94, reliabilities are satisfactory for all samples (see Tables 2 and 3). Following a two-step approach to structural equation modeling (SEM), confirmatory factor analyses (CFAs) were conducted for the scales. We considered the ratio of chi-square (χ2) to the degrees of freedom to evaluate model fit. Acceptable values for this indicator usually are below 3.00. We also report the incremental fit (IFI), non-normed fit (NNFI), and comparative fit indices (CFI), which by convention should be above .90. In anticipation of our results we note that these fit indices usually range from 0 to 1.00 but can also be greater than 1, which indicates per- fect model fit for this particular criterion, or lower than 0, which would be interpreted as a sign of severe model misspecification. For the root mean square error of approxi- mation (RMSEA), an estimate of .00 indicates perfect fit, whereas values up to .09 indicate acceptable model fit (e.g., Byrne, 1998; Kline, 1998).

In the two cross-sectional data sets, a measurement model with two factors for PI and RBSE was compared to a model with all PI and RBSE items loading on one factor. The two-factor model indicated acceptable fit in both cases, whereas the one- factor model was rejected (Table 4), confirming that PI and RBSE constitute two related yet empirically distinct constructs. Subsequently, items were randomly aggregated into three parcels per construct (i.e., two two-item and one three-item

410 THE JOURNAL OF APPLIED BEHAVIORAL SCIENCE December 2007

parcels for PI and two three-item and one four-item parcels for RBSE; assignment of items to parcels is identical across all groups). There are several reasons for using item parceling procedures; among them are reduction of non-normality in the distri- bution of data, higher reliability of parcels than of individual items, improvement of the ratio of estimated parameters to sample size, greater likelihood of achieving a proper model solution, and better model fit (e.g., Bandalos & Finney, 2001). Item parceling is a widely used and legitimate strategy in SEM if the factor structure is supported at the item level (Coffman & MacCallum, 2005; Hall, Snell, & Foust, 1999). The two-factor solution with three-item parcels per construct yielded satis- factory fit in both cross-sectional samples. Compared to the disaggregated measure- ment model, the fit indices improved slightly in Sample 1 and more so in Sample 2. For the six items measuring anticipated benefits and change commitment, the two- factor solution also shows acceptable fit in both cross-sectional samples, signifi- cantly better than the one-factor solution (see Table 4). Because we measured autonomy using a well-established measure with only three items, a just-identified case in SEM, no CFA was performed.

CFAs were performed separately for each construct in the longitudinal sample due to its smaller sample size. Each CFA included three items—respective parcels— from measurement Point 1 and three from measurement Point 2. Both a two-factor model, which differentiates between measurement points, and a one-factor model,

Hornung, Rousseau / AUTONOMY AT WORK 411

TABLE 2

Descriptive Statistics and Correlations for Samples 1 (N = 166) and Sample 2 (N = 133)

Alpha M SD 1 2 3 4 5 6 7 8

1. Gender S1 — — — S2

2. Agea S1 — — — –.01 S2 –.09

3. Tenurea S1 — — — –.12 .46** S2 .06 .41**

4. Educationa S1 — — — –.19* –.12 –.01 S2 –.07 –.00 –.08

5. Autonomy S1 .82 3.56 0.98 .04 .07 –.06 .08 S2 .81 3.72 0.93 –.04 .10 .01 .16

6. Role breadth S1 .94 3.11 0.96 –.05 –.06 –.10 .24** .36** self-efficacy S2 .93 3.17 0.94 –.11 .07 .01 .39** .38**

7. Personal initiative S1 .85 3.70 0.64 .04 .11 –.03 .04 .30** .53** S2 .87 3.85 0.63 –.13 .13 .01 .13 .40** .53**

8. Anticipated benefits S1 .92 2.98 1.10 –.15 .07 –.15 .10 .26** .37** .41** S2 .82 2.97 1.01 –.17 –.06 –.18* .17 .38** .33** .33**

9. Change commitment S1 .90 3.15 1.15 –.09 .08 .08 .24** .26** .54** .32** .55** S2 .92 3.11 1.17 –.04 .05 –.04 .28** .21* .59** .30** .43**

NOTE: S1 = Sample 1; S2 = Sample 2. a. Age, tenure, and education are categorical variables. *p < .05. **p < .01. (two-tailed)

412

T A

B L

E 3

D es

cr ip

ti ve

S ta

ti st

ic s

an d

C or

re la

ti on

s fo

r S

am p

le 3

( N

== 74

)

A lp

h a

M S

D 1

2 3

4 5

6 7

8 9

1 0

1 1

1 2

1 3

1. G

en de

r —

— —

2. A

ge a

— —

–. 12

3. T

en ur

ea —

— –.

02 .4

9* *

4. E

du ca

ti on

a —

— –.

13 .0

6 .0

9 5.

A ut

on om

y T

im e

1 .8

1 3.

67 0.

96 .0

7 .0

8 –.

13 .0

0 6.

A ut

on om

y T

im e

2 .7

5 3.

76 0.

90 –.

03 .1

5 .1

3 .1

2 .3

0* 7.

R ol

e br

ea dt

h se

lf -e

ff ic

ac y

T im

e 1

.9 2

3. 31

0. 84

–. 04

–. 02

–. 10

.2 1

.3 4*

* .1

1 8.

R ol

e br

ea dt

h se

lf -e

ff ic

ac y

T im

e 2

.9 1

3. 36

0. 86

.2 2

.0 2

.0 3

.2 3

.3 2*

* .1

2 .6

4* *

9. P

er so

na l

in it

ia ti

ve T

im e

1 .8

4 3.

87 0.

57 –.

07 .1

0 –.

06 –.

09 .3

4* *

.1 6

.5 0*

* .1

7 10

. P

er so

na l

in it

ia ti

ve T

im e

2 .8

6 3.

82 0.

61 –.

14 .1

7 .0

1 .1

0 .3

1* *

.2 7*

.3 9*

* .3

9* *

.4 6*

* 11

. A nt

ic ip

at ed

b en

ef it

s T

im e

1 .9

2 3.

15 1.

05 –.

29 *

.1 7

–. 13

.0 8

.2 4*

.2 0

.3 3*

* .1

0 .2

4* .2

1 12

. A nt

ic ip

at ed

b en

ef it

s T

im e

2 .8

9 3.

00 1.

10 –.

20 .2

8* .1

3 .0

4 .2

6* .3

0* *

.2 2

.2 1

.1 0

.3 4*

* .4

6* *

13 .

C ha

ng e

co m

m it

m en

t T

im e

1 .9

1 3.

30 1.

11 –.

17 .1

4 .0

3 .1

4 .3

0* .1

4 .4

8* *

.1 8

.1 8

.0 8

.5 5*

* .0

3 14

. C

ha ng

e co

m m

it m

en t

T im

e 2

.9 0

3. 07

1. 10

–. 08

–. 11

.0 9

.0 5

.2 2

.0 2

.3 0*

.4 8*

* .0

6 .2

5* .1

4 .3

6* *

.2 4*

a. A

ge ,

te nu

re ,

an d

ed uc

at io

n ar

e ca

te go

ri ca

l va

ri ab

le s

an d

ar e

re fe

rr in

g to

m ea

su re

m en

t P

oi nt

2 .

*p <

.0 5.

* *p

< .0

1. (

tw o-

ta il

ed )

413

T A

B L

E 4

M ea

su re

m en

t M

od el

s in

S am

p le

1 a

n d

S am

p le

2

R o o t

M ea

n In

cr em

en ta

l N

o n -N

o rm

ed `

C o m

p a ra

ti ve

S q u a re

E rr

o r

o f

N χ²

df χ²

/d f

F it

I n d ex

F it

I n d ex

F it

I n d ex

A p p ro

xi m

a ti

o n

R ol

e br

ea dt

h se

lf -e

ff ic

ac y

(R B

S E

) an

d pe

rs on

al i

ni ti

at iv

e (P

I) S

am pl

e 1:

on e

fa ct

or 16

6 53

1. 60

** 11

9 4.

47 .7

6 .6

8 .7

5 .1

5 S

am pl

e 1:

tw o

fa ct

or s

16 6

28 6.

82 **

11 8

2. 43

.9 0

.8 7

.9 0

.0 9

S am

pl e

1: tw

o fa

ct or

s w

it h

th re

e it

em p

ar ce

ls 16

6 17

.3 3*

8 2.

17 .9

8 .9

6 .9

8 .0

8 S

am pl

e 2:

on e

fa ct

or 13

3 43

5. 11

** 11

9 3.

66 .7

7 .7

0 .7

7 .1

4 S

am pl

e 2:

tw o

fa ct

or s

13 3

22 6.

09 **

11 8

1. 92

.9 2

.9 0

.9 2

.0 8

S am

pl e

2: tw

o fa

ct or

s w

it h

th re

e it

em p

ar ce

ls 13

3 8.

86 8

1. 11

.9 9

.9 9

.9 9

.0 3

A nt

ic ip

at ed

b en

ef it

s an

d ch

an ge

c om

m it

m en

t S

am pl

e 1:

on e

fa ct

or 16

6 20

9. 67

** 9

23 .3

0 .7

2 .3

4 .7

2 .3

7 S

am pl

e 1:

tw o

fa ct

or s

16 6

14 .0

4 8

1. 76

.9 9

.9 8

.9 9

.0 7

S am

pl e

2: on

e fa

ct or

13 3

13 4.

85 **

9 14

.9 8

.7 4

.3 7

.7 3

.3 3

S am

pl e

2: tw

o fa

ct or

s 13

3 14

.2 0

8 1.

78 .9

9 .9

7 .9

9 .0

8

*p <

.0 5.

* *p

< .0

1. (

tw o-

ta il

ed )

which does not, were tested (Table 5). Overall, the fit of all two-factor measurement models are acceptable. In two cases—anticipated benefits and change commitment— RMSEA slightly exceeded the proposed cutoff of .09. In contrast, all one-factor models produced completely unacceptable fit indices. The superiority of the two- factor models signifies the distinctness of the results at different measurement points and indicates that common method variance (Podsakoff, MacKenzie, Lee, & Podsakoff, 2003) poses little problem.

Descriptive Statistics and Correlations

Descriptive statistics and correlations for all three samples are displayed in Tables 2 and 3. We note that means, standard deviations, and correlations in the two cross-sectional samples are comparable. Demographic attributes had little influ- ence on the variables studied. However, education correlates positively with RBSE but not PI, further supporting the distinctness of the two constructs. It also suggests that RBSE might be influenced not only by job content but also by the actual occu- pation individuals hold. For the longitudinal sample, pairwise t tests assessed whether the means had changed significantly over time (no significant differences were observed). At both points in time, mean levels were comparable for all key variables studied here.

Model Testing

For hypothesis testing, measurement models were integrated into a structural model with three manifest indicator variables for each latent construct. Hypotheses 1a, 1b, 2, 3a, 3b, 4a, 4b, and 6 postulate direct effects involving autonomy, the two forms of employee proactivity, and the two change support facets. These eight hypotheses are represented by direct paths in the structural model. Hypotheses 5a and 5b postulate that each construct of employee proactivity relates more strongly to one facet of change support than to the other; namely, PI is postulated to have a stronger positive effect on anticipated benefits than on change commitment, although RBSE is expected to relate more strongly to the latter. These hypothesized differential effects were assessed by comparing the unstandardized regression weights for the respective paths. Pairwise t tests assessed whether differences in effect size attain statistical significance.

The structural model was first tested in Sample 1 and then cross-validated in Sample 2. Results were stable across the two samples and thus are reported together. Fit indices are similar for both samples (see Figure 1) and indicate good fit between observed and hypothesized relationships. Of the eight hypothesized structural paths, six turned out to be significant at the level of p < .01 in Sample 1. In Sample 2, five paths are significant at p < .01, whereas one path is significant at the p < .05 level. Two paths are not significant in either sample. Most significant coefficients in the model fall between .30 and .50, which according to Cohen (1988) constitute medium-size effects. For a better understanding of the results, we also calculated

414 THE JOURNAL OF APPLIED BEHAVIORAL SCIENCE December 2007

415

T A

B L

E 5

M ea

su re

m en

t M

od el

s in

S am

p le

3

R o o t

M ea

n In

cr em

en ta

l N

o n -N

o rm

ed C

o m

p a ra

ti ve

S q u a re

E rr

o r

o f

N χ²

df χ²

/d f

F it

I n d ex

F it

I n d ex

F it

I n d ex

A p p ro

xi m

a ti

o n

A ut

on om

y S

am pl

e 3:

on e

fa ct

or s

74 94

.1 9*

* 9

10 .4

7 0.

56 –.

11 0.

53 .3

6 S

am pl

e 3:

tw o

fa ct

or s

74 8.

07 8

1. 01

1. 00

.9 9

1. 00

.0 1

R ol

e br

ea dt

h se

lf -e

ff ic

ac y

(R B

S E

) S

am pl

e 3:

on e

fa ct

or 74

78 .9

0* *

9 8.

77 0.

77 .4

3 0.

76 .3

3 S

am pl

e 3:

tw o

fa ct

or s

74 12

.5 6

8 1.

57 0.

99 .9

6 0.

98 .0

9 P

er so

na l

in it

ia ti

ve (

P I)

S am

pl e

3: on

e fa

ct or

74 46

.2 2*

* 9

5. 14

0. 76

.3 9

0. 74

.2 4

S am

pl e

3: tw

o fa

ct or

s 74

12 .0

0 8

1. 50

0. 97

.9 3

0. 97

.0 8

A nt

ic ip

at ed

b en

ef it

s S

am pl

e 3:

on e

fa ct

or 74

11 6.

48 **

9 12

.9 4

0. 67

.2 0

0. 66

.4 0

S am

pl e

3: tw

o fa

ct or

s 74

13 .6

5 8

1. 71

0. 98

.9 5

0. 98

.1 0

C ha

ng e

co m

m it

m en

t S

am pl

e 3:

on e

fa ct

or 74

15 4.

38 **

9 17

.1 5

0. 54

–. 13

0. 52

.4 7

S am

pl e

3: tw

o fa

ct or

s 74

13 .9

4 8

1. 74

0. 98

.9 5

0. 98

.1 0

** p

< .0

1. (

tw o-

ta il

ed )

Y 1

Y 2

S 1 : .4

5 **

S

2 : .2

8 **

S 1 : .2

5 **

S

2 : .2

6 **

S 1 : .4

2 **

S

2 : .4

2 **

S 1

: .5

2 **

S 2 : .5

3 **

S 1 : .1

6

S 2 : .1

4

S 1 : .4

4 **

S

2 : .6

4 **

S 1 : .4

0 **

S

2 : .3

5 *

S 1 : – .0

7

S 2 :

– .1

9

A u

to n

o m

y

Y 7

Y 8

Y 9

R o

le B

re a

th S

e lf-

E ff

ic a

cy

Y 4

Y 5

P e

rs o

n a

l In

iti a

tiv e

X 1

X 2

Y 1 0

Y 1 1

C h

a n

g e

C o

m m

itm e

n t

Y 1 2

A n

tic ip

a te

d

B e

n e

fit s

X 3

Y 6

Y 3

F IG

U R

E 1

: T

es t

of t

h e

H yp

ot h

es iz

ed S

tr u

ct u

ra l

M od

el i

n S

am p

le 1

a n

d S

am p

le 2

N O

T E

: S

am pl

e 1

(S 1)

: N

= 16

6; χ

² =

13 9.

18 **

; d f

= 82

; χ²

/d f

= 1.

70 ;

in cr

em en

ta l

fi t

in de

x =

.9 7;

n on

-n or

m ed

f it

i nd

ex =

.9 5;

c om

pa ra

ti ve

f it

i nd

ex =

.9 7;

r oo

t m

ea n

sq ua

re e

rr or

o f

ap pr

ox im

at io

n =

.0 6.

S am

pl e

2 (S

2) :

N =

13 3;

χ ²

= 13

2. 39

** ;

d f

= 82

; χ²

/d f

= 1.

62 ;

in cr

em en

ta l

fi t

in de

x =

.9 6;

n on

-n or

m ed

f it

i nd

ex =

.9 4;

c om

pa ra

- ti

ve f

it i

nd ex

= .9

6; r

oo t

m ea

n sq

ua re

e rr

or o

f ap

pr ox

im at

io n

= .0

7. **

p <

.0 1.

416

indirect and total effects based on the significant paths (Table 6) and tested the main mediating effects in the model using Sobel tests (Sobel, 1982).

Hypotheses 1a and 1b are both supported. In both samples, autonomy manifests significant paths to PI (βS1 = .25, p < .01; βS2 = .26, p < .01) and RBSE (βS1 = .42, p < .01; βS2 = .42, p < .01). The difference in the magnitude of these effects vanishes if the indirect effects of autonomy on PI mediated by RBSE are considered (Table 6). In accordance with Hypothesis 2, RBSE has a positive effect on PI (βS1 = .52, p < .01; β S2 = .53, p < .01). Although Hypothesis 3a and Hypothesis 4b are supported by significant paths from RBSE to change commitment (βS1 = .44, p < .01; βS2 = .64, p < .01) and from PI to anticipated benefits (βS1 = .40 p < .01; βS2 = .35, p < .01), Hypothesis 3b and Hypothesis 4a are not, with no significant direct effects of RBSE on anticipated benefits or of PI on change commitment. In the case of PI, the effect direction is negative, which though not significant, is contrary to expectation. However, there are indirect positive effects of PI on change commitment mediated by anticipated benefits (βS1 = .18 p < .01; βS2 = .10, p = .055) as well as of RBSE on anticipated benefits via PI (βS1 = .21 p < .01; βS2 = .19, p < .05), which we tested for statistical significance (Sobel, 1982). Results indicate that the nonsignificant direct effects are attributable to full mediation of PI’s positive effect on change commit- ment by anticipated benefits as well as RBSE’s effect on the latter via PI. These mediation effects are consistent with the theory underlying our model (see Figure 1 and Table 6). Consistent with Hypothesis 6, anticipated benefits and change com- mitment are positively related (βS1 = .45, p < .01; βS2 = .28, p < .01).

The hypothesized differential effects of the two facets of employee proactivity on change responses (Hypothesis 5a and Hypothesis 5b) are supported by the signifi- cant effects of PI on anticipated benefits and of RBSE on change commitment and the nonsignificant effects of PI on change commitment and of RBSE on anticipated benefits. In addition, pairwise t tests indicate significant differences in the size of path coefficients from PI to anticipated benefits and change commitment (tS1 = 2.86, p < .01; tS2 = 2.68, p < .01) as well as from RBSE to anticipated benefits and change commitment (tS1 = 2.29, p < .05; tS2 = 3.90, p < .01) in both cross-sectional samples. Taken together, these results consistently support Hypotheses 5a and 5b.

Tests of Causal Direction

To address concerns regarding causality and common method variance associated with cross-sectional samples, the causal directions of observed relationships were examined using the longitudinal sample. As the small sample size limits the number of parameters to be estimated reliably at the same time and with sufficient statistical power (e.g., Bentler & Chou, 1987), the six significant paths in the model were each tested separately using simple SEM regressions with two latent and six indicator variables. First, the hypothesized dependent variable measured at Time 2 was regressed on the independent variable measured at Time 1. Then, testing for reverse causality, we reversed the order of measurement and the causal direction of vari- ables. We used model fit and significance of path coefficients to assess the likelihood of our hypothesized causal relationships in comparison to their reversed counterpart. For a more rigorous test of causality, pairwise t tests compared the unstandardized

Hornung, Rousseau / AUTONOMY AT WORK 417

regression coefficients of the postulated and the reverse causal models. This test examines whether the postulated and reverse regression weights are different from each other as well as being significantly different from zero.

Overall, the fit indices for many of these regression models were extraordinarily good, a pattern partially attributable to low degrees of freedom and model simplicity, each known to inflate fit (Table 7). In the case of autonomy and RBSE, the regression model with the hypothesized causal relationship shows good model fit and a signifi- cant path from autonomy at Time 1 to RBSE measured at Time 2 (β = .41, p < .01), whereas fit for the reverse causal model is worse and its path coefficient is not signif- icant. For autonomy and PI, the extraordinary good fit indices do not allow for dis- crimination between the adequacy of the models. However, only the path in the hypothesized causal direction is statistically significant (β = .38, p < .05), roughly cor- responding to the size of the absolute effects found in the cross-sectional data. The hypothesized causal relationship of RBSE with PI is supported by a comparatively better model fit and a significant path coefficient (β = .44, p < .01). The regressions of RBSE and change commitment again show no difference in model fit, while only the

418 THE JOURNAL OF APPLIED BEHAVIORAL SCIENCE December 2007

TABLE 6

Direct, Indirect, and Total Effects in the Model

Role Breadth Personal Self-Efficacy Initiative Anticipated

R2 Autonomy (RBSE) (PI) Benefits

S1 S2 S1 S2 S1 S2 S1 S2 S1 S2

RBSE .18 .18 Total effects 0.42 0.42 Direct effects 0.42 0.42 Indirect effects 0.00 0.00 PI .33 .35 Total effects 0.47 0.48 0.52 0.53 Direct effects 0.25 0.26 0.52 0.53 Indirect effects 0.22 0.22 0.00 0.00 Sobel test (z) 3.57** 3.27** — — Anticipated benefits .16 .12 Total effects 0.19 0.17 0.21 0.19 0.40 0.35 Direct effects 0.00 0.00 ns ns 0.40 0.35 Indirect effects 0.19 0.17 0.21 0.19 0.00 0.00 Sobel test (z) 2.25*a 1.84a 3.08**a 2.27*a — — Change commitment .40 .49 Total effects 0.27 0.30 0.53 0.69 0.18 0.10 0.45 .28 Direct effect 0.00 0.00 0.44 0.64 ns ns 0.45 .28 Indirect effects 0.27 0.30 0.09 0.05 0.18 0.10 0.00 .00 Sobel test (z) 3.19**b 3.28**b — — 2.95** 1.92*c — —

NOTE: S1 = Sample 1; S2 = Sample 2. a. Mediation via PI. b. Mediation via RBSE. c. p = .055. *p < .05. **p < .01. (two-tailed)

419

T A

B L

E 7

L on

gi tu

d in

al A

n al

ys is

o f

C au

sa l

E ff

ec ts

a n

d R

ev er

se C

au sa

li ty

i n

S am

p le

3

R o o t

M ea

n N

o n

S q u a re

In cr

em en

ta l

N o rm

ed C

o m

p a ra

ti ve

E rr

o r

o f

T im

e 1

→ T

im e

2 β

b S

E χ²

df χ²

/d f

F it

I n d ex

F it

I n d ex

F it

I n d ex

A p p ro

xi m

a ti

o n

A ut

on om

y →

R B

S E

.4 1*

* .5

6 .2

0 10

.6 0

8 1.

33 0.

99 0.

97 0.

99 .0

7 R

ol e-

br ea

dt h

se lf

-e ff

ic ac

y (R

B S

E )

→ A

ut on

om y

.1 3

.0 7

.0 8

14 .7

4 8

1. 84

0. 97

0. 92

0. 97

.1 1

A ut

on om

y →

P I

.3 8*

.4 2

.1 7

4. 25

8 0.

53 1.

00 1.

00 1.

00 .0

0 P

er so

na l

in it

ia ti

ve (

P I)

→ A

ut on

om y

.2 6

.1 9

.1 1

7. 80

8 0.

98 1.

00 1.

00 1.

00 .0

0 R

B S

E →

P I

.4 4*

* .3

8 .1

2 8.

75 8

1. 01

0. 99

0. 99

0. 99

.0 4

P I

→ R

B S

E .2

5 .3

2 .1

8 16

.9 3

8 2.

12 0.

96 0.

88 0.

95 .1

2 R

B S

E →

C ha

ng e

to c

om m

it m

en t

.3 3*

* .5

4 .2

0 5.

56 8

0. 70

1. 00

1. 00

1. 00

.0 0

C ha

ng e

to c

om m

it m

en t

→ R

B S

E .1

8 .1

2 .0

9 6.

81 8

0. 85

1. 00

1. 00

1. 00

.0 0

P I

→ A

nt ic

ip at

ed b

en ef

it s

.1 4

.2 1

.2 0

9. 35

8 1.

17 0.

99 0.

98 0.

99 .0

5 A

nt ic

ip at

ed b

en ef

it s

→ P

I .2

2 .1

7 .1

0 15

.1 4

8 1.

89 0.

97 0.

92 0.

97 .1

1 A

nt ic

ip at

ed b

en ef

it s

→ C

ha ng

e to

c om

m it

m en

t .0

9 .1

1 .1

5 10

.3 4

8 1.

29 0.

99 0.

98 0.

99 .0

6 C

ha ng

e to

c om

m it

m en

t →

A nt

ic ip

at ed

b en

ef it

s .0

8 .0

6 .1

0 7.

16 8

0. 90

1. 00

1. 00

1. 00

.0 0

** p

< .0

1. (

tw o-

ta il

ed )

predicted causal direction manifests a significant effect (β = .33, p < .01). No unequiv- ocal results could be attained from the analysis of the causal relationships between PI and anticipated benefit and between the latter and change commitment.

Results of the t test, however, indicate significant differences in hypothesized and reverse direction unstandardized path coefficients only for the regressions of auton- omy and RBSE and of RBSE and change commitment (t = 2.22, p < .05 and t = 1.91, p < .05). These results provide evidence for our argument that job autonomy causes self-efficacy in the context of one’s role, which in turn gives rise to employee com- mitment to support an organizational change.

DISCUSSION

Organizational changes with the potential to modify work roles and alter perfor- mance goals invite employees to behave in ways that go beyond their existing duties and responsibilities. Active employee support is essential to fill the white spaces accompa- nying structural change, from underspecified new roles to unclear paths to new goals. Structural change’s ultimate success often depends on the cumulative changes active, self-determining employees introduce. On the other hand, changes can also yield inse- curity and potential losses as employees are challenged to cope with new demands (Rousseau, 1995). The present study provides support for the socializing power of everyday work experience, where employees used to autonomy are better positioned to support organizational change than their less advantaged counterparts.

Autonomy enhances confidence in taking on a broader job role and performing in new ways. Replicating the findings of preceding research on the constructs separately (Axtell & Parker, 2003; Frese et al., 1996; Galperin, 2005; Morgeson et al., 2005; Parker, 1998), autonomy fitted best as an antecedent of both PI and RBSE in our cross- sectional models. The results of the longitudinal analysis support the postulated causal direction fully for RBSE and at least partially for PI. We conclude that the autonomy a job offers promotes its occupant’s development of proactive, self-starting behaviors and self-efficacy, enabling acceptance of a broader organizational role.

However, findings for PI’s causal relationships are more equivocal, as no differ- ences were found in its effect size for postulated and reverse causal directions. We need therefore to consider alternative explanations. If reverse causality is the case, selection effects may exist such that proactive individuals are more likely to choose jobs high in autonomy and influence (e.g., Bateman & Crant, 1993; Crant, 1995, 2000; Frese et al., 1996). Autonomy thus would be the result of employee proac- tivity rather than the other way around. Highly proactive individuals may also make efforts to actively redefine their tasks and “craft” their jobs to gain more autonomy and discretion (Wrzesniewski & Dutton, 2001). Combined with occupational socialization through work tasks, these effects can lead to a form of reciprocal determination with employee proactivity and autonomy having a mutual influence on each other (Bandura, 1982; Frese & Fay, 2001).

Our results go beyond previous research and indicate that not only are PI and RBSE empirically distinct but that their impact on change-related responses differs. As a

420 THE JOURNAL OF APPLIED BEHAVIORAL SCIENCE December 2007

function of their ability to persistently pursue goals even in the face of obstacles, employees high on PI evaluate the potential outcomes of the change more positively than their low PI counterparts. Their commitment to support the change can be instru- mental in attaining potential benefits. On the other hand, RBSE relates directly to change commitment, suggesting that changes that broaden their role may be welcomed by workers confident of their ability to perform successfully in the new capacity.

Organizations anticipating the need for future structural changes may lay the groundwork in advance by fostering workforce flexibility and proactivity. Such interventions might involve job rotations, developing broader skills, and/or sharing information about the larger context in which one’s work is performed. Previous studies have identified several antecedents of employee support for change, includ- ing innovative climate, shared information, and participation (Choi, 2007; Wanberg & Banas, 2000). Our results suggest that employee proactivity is an important medi- ator in the relationship between these workplace socializing experiences and change support. Importantly, antecedents of change support are not limited to contextual factors in the organization but also include aspects of the individuals’ work tasks. Active work tasks (Karasek, 1979) offering high levels of employee discretion and control may be crucial to developing proactive employees whose response to orga- nizational change is to shape their surroundings in ways beneficial for themselves and their organization.

Limitations

Several methodological and contextual limitations warrant mention. The scales used for measuring anticipated benefits from and commitment to support the change were designed specifically for the organizational change studied here. Our use of customized measures was motivated by the limits of existing measures, including openness to change (Griffin et al., 2007; Miller et al., 1994), program commitment (Jansen, 2004; Neubert & Cady, 2001), and the three-component commitment to change scale (Herscovitch & Meyer, 2002), none of which differentiates between expected outcomes and commitment to support the change. In undertaking the present study, we noted that Wanberg and Banas (2000) report that their adapted unidimensional measure of openness to change empirically split in two factors, rep- resenting the two facets of employee support we studied here. Moreover, the afore- mentioned scales operationalize responses to change in very general terms, whereas instead we focused on the specific potential benefits and concrete behavioral support intentions relevant to the change we studied. In doing so, we followed Parker’s (1998) approach of customizing measures to the change, which nonetheless raises concerns regarding generalizability that must be addressed through research in other change settings. To this end we concur with Piderit (2000) that explicitly taking into account employee assessment of consequences for explaining supportive, ambiva- lent, or resistant responses to change represents an important yet largely neglected perspective in change research.

Maintaining our methodological latent variable approach, the size of the longitu- dinal sample permitted only simple regression models to be tested without control- ling for other variables. As such, potential confounding factors may exist, though our

Hornung, Rousseau / AUTONOMY AT WORK 421

cross-sectional analyses suggest that these are unlikely. Another concern is the selected interval between the two surveys. Although measurement points in longitu- dinal studies should correspond to the time lag needed for the causal variables to affect the outcome variables, recommendations for how to do so are scarce and inconsistent (Taris, 2000). As one of our aims was to show the medium- to long-term stimulating and socializing effects of autonomy on proactivity, the chosen time lag of 18 months seems appropriate for this purpose (Taris & Kompier, 2003). For other tested relationships (e.g., proactivity and change support) however, a shorter time span may be suitable. To more thoroughly study the psychological processes under- lying posited causal relationships, multiphase longitudinal studies are needed. As discrete events may play an important role in shaping employee support for organi- zational change, diary studies (e.g., Conway & Briner, 2002) may be a suitable alter- native approach to studying change.

The present study used autonomy, representing one of the most important and widely researched job characteristics as a measure of worker control. Other job content mea- sures might also be pertinent (e.g., Jackson, Wall, Martin, & Davids, 1993; Parker, 1998). For instance, job complexity (Frese et al., 1996; Frese & Fay, 2001; Ohly et al., 2006) can also promote employee proactivity during change. Thus, we recommend broader investigation of how other job facets and worker control (e.g., participation in manager- ial decisions, access to company financial information; cf. Rousseau & Shperling, 2003; Wanberg & Banas, 2000) jointly influence worker responses to change.

Regarding the interrelation of the two proactivity constructs, our results suggest that RBSE represents an antecedent to the behavioral pattern of PI rather than the other way around. RBSE shows higher stability coefficients than PI across the two points in time (.64 vs. .46), raising questions regarding PI’s stability. Viewing it as a stable personal- ity trait is not readily reconciled with the socializing effects of work and organizational variables on proactivity that numerous studies report. Along the same lines, Parker and Sprigg (1999), using the construct of proactive personality as a moderator, report its positive correlation with job control. They state that “In reality proactive personality is likely to be an outcome as well . . . passive employees whose jobs are redesigned to have greater autonomy will become more proactive” (p. 938). In our study, RBSE mediated the relationship between autonomy and PI, consistent with findings Speier and Frese (1997) report for generalized self-efficacy. However, we used Frese’s (Frese et al., 1997) widely applied PI measure, whereas Parker and Sprigg use only a subset of items developed by Bateman and Crant (1993) in their cross-sectional study. As such, our findings regarding the interrelationship of PI and RBSE underscore the need for careful attention to the measures used to operationalize proactivity and the partic- ular facets of this construct studied (i.e., trait, state, or behavior).

The specific change context studied here raises another limitation. The hospital experienced budget restrictions and financial pressures at the time of the second survey. In the face of these fiscal challenges, its managers and employees were somewhat pressed to attend to financial concerns at the expense of the Councils’ goals for improving work–life quality and clinical care. Such conditions may have reduced employee beliefs regarding their influence over the organization. Interviews with hospital management and employee representatives to the advisory councils at

422 THE JOURNAL OF APPLIED BEHAVIORAL SCIENCE December 2007

Time 2 confirmed concerns regarding the Councils’ capacity to act on all the issues employees raised. It is possible that such contextual factors contributed to our fail- ure to confirm the role of expected benefits as an antecedent of commitment to sup- port the change. This interpretation is plausible given the decline in the effect of anticipated benefits on change commitment from Time 1 to Time 2 as evident in our otherwise stable cross-sectional models.

Supporting this view of employees as proactive and goal-oriented “crafters” of their work environment, anticipated benefits predicted change commitment in our cross-sectional models (Rousseau, 1995). Failure to replicate these findings in the longitudinal sample and the observable decrease in the cross-sectional effect size from Time 1 to Time 2 indicate changes in employee attitudes toward the organiza- tional transition that might be attributable to the specific context of our study.

Conclusions

This study provides evidence that a more active, flexible, and problem-oriented approach to work can be fostered by workplace experiences where workers have opportunity to exert influence over their jobs. Whereas our findings concerning employee responses to an organizational change must be interpreted in light of our particular context (i.e., the change’s impact on the organization and members), results regarding autonomy’s socializing effects on employee proactivity were con- firmed. Recognition of the developmental and socializing effects work experiences have on employees, in particular their acquisition of the capacity and confidence to undertake new roles and responsibilities, may help resolve lingering questions about the nature of so-called worker change resistance (Piderit, 2000). Unless their work experiences have enabled them to act effectively in times of change, the apparent resistance workers manifest is likely due at least in part to two related factors: (a) lack of confidence in their own ability to participate effectively in the change and (b) inexperience with the proactive, self-starting, and persistent pursuit of goals. Workplace experiences are critical to reversing two conditions in contemporary workplaces that can make organizational change especially difficult for workers.

Autonomy plays a paradoxical role in change management: It is not only a poten- tial change outcome (e.g., Porter-O’Grady, 1992) but also a prerequisite for employee ability, confidence, and willingness to support even those changes that would increase their empowerment. Participatory organizational change, intended to create a more efficient and supportive work setting, can trigger an upward spiral of activity by rein- forcing individual control opportunities and aspirations—the opposite of the vicious cycle of learned helpless that pervades many organizations (Martinko & Gardner, 1982). Efforts to enhance autonomy thus can be a critical step in a successful organi- zational change by positioning employees to effectively respond to the further demands and opportunities change creates. Critical to successful structural change may be the enactment of new ways of organizing by proactive employees, whose capabilities, though often present, are not well recognized. As Margaret Wheatley (1992) says, “We have not invented our strategies for change; we have merely discov- ered them” (p. 96).

Hornung, Rousseau / AUTONOMY AT WORK 423

REFERENCES

Armenakis, A. A., Harris, S. G., & Mossholder, K. W. (1993). Creating readiness for organizational change. Human Relations, 46, 681-703.

Axtell, C. M., & Parker, S. K. (2003). Promoting role breadth self-efficacy through involvement, work redesign and training. Human Relations, 56, 113-131.

Bandalos, D. L., & Finney, S. J. (2001). Item parceling issues in structural equation modeling. In G. A. Marcoulides & R. E. Schumacker (Eds.), New developments and techniques in structural equa- tion modeling (pp. 269-296). Mahwah, NJ: Lawrence Erlbaum.

Bandura, A. (1982). Self-efficacy mechanism in human agency. American Psychologist, 37, 122-147. Bandura, A. (1997). Self-efficacy: The exercise of control. New York: Freeman. Bartunek, J. M., & Moch, M. K. (1987). First-order, second-order, and third-order change and organization

development interventions: A cognitive approach. The Journal of Applied Behavioral Science, 23, 483-500.

Bartunek, J. M., Rousseau, D. M., Rudolph, J., & Depalma, J. (2006). On the receiving end: Sensemaking, emotion, and assessments of an organizational change initiated by others. The Journal of Applied Behavioral Science, 42, 182-206.

Bateman, T. S., & Crant, J. M. (1993). The proactive component of organizational behavior: A measure and correlates. Journal of Organizational Behavior, 12, 103-118.

Bentler, P. M., & Chou, C. (1987). Practical issues in structural modeling. Sociological Methods and Research, 16, 78-117.

Byrne, B. M. (1998). Structural equation modeling with LISREL, PRELIS, and SIMPLIS: Basic concepts, applications, and programming. Mahwah, NJ: Lawrence Erlbaum.

Choi, J. N. (2007). Change-oriented organizational citizenship behavior: Effects of work environment characteristics and intervening psychological processes. Journal of Organizational Behavior, 28, 467-484.

Coffman, D. L., & MacCallum, R. C. (2005). Using parcels to convert path analysis models into latent variable models. Multivariate Behavioral Research, 40, 235-259.

Cohen, J. (1988). Statistical power analysis for the behavioral sciences. Hillsdale, NJ: Lawrence Erlbaum. Conway, N., & Briner, R. B. (2002). A daily diary study of affective responses to psychological contract

breach and exceeded promises. Journal of Organizational Behavior, 23, 287-302. Crant, J. M. (1995). The proactive personality scale and objective job performance among real estate

agents. Journal of Applied Psychology, 80, 532-537. Crant, J. M. (2000). Proactive behavior in organizations. Journal of Management, 26, 435-462. Cunningham, G. B. (2006). The relationships among commitment to change, coping with change, and

turnover intentions. European Journal of Work and Organizational Psychology, 15, 29-45. Dent, E., & Goldberg, S. (1999). Challenging “resistance to change.” The Journal of Applied Behavioral

Science, 35, 25-41. Fay, D., & Frese, M. (2001). The concept of personal initiative: An overview of validity studies. Human

Performance, 14, 97-124. Frese, M. (1989). Theoretical models of control and health. In S. Sauter, J. Hurrel, & C. Cooper (Eds.),

Job control and worker health (pp. 107-128). Chichester, UK: Wiley. Frese, M., & Fay, D. (2001). Personal initiative: An active performance concept for work in the 21st

century. In B. M. Staw & R. M. Sutton (Eds.), Research in organizational behavior (Vol. 23, pp. 133-187). Amsterdam: Elsevier Science.

Frese, M., Fay, D., Hilburger, T., Leng, K., & Tag, A. (1997). The concept of personal initiative: Operationalization, reliability and validity in two German samples. Journal of Occupational and Organizational Psychology, 70, 139-161.

Frese, M., Kring, W., Soose, A., & Zempel, J. (1996). Personal initiative at work: Differences between East and West Germany. Academy of Management Journal, 39, 37-63.

Fried, Y., & Ferris, G. R. (1987). The validity of the job characteristics model: A review and meta-analysis. Personnel Psychology, 40, 287-322.

Galperin, B. L. (2005). Examination of the moderating and mediating effect of role breadth self-efficacy in explaining deviance. Academy of Management Proceedings, 2005, T1-T6.

424 THE JOURNAL OF APPLIED BEHAVIORAL SCIENCE December 2007

Ganster, D. C. (1989). Worker control and well-being: A review of research in the workplace. In S. Sauter, J. Hurrel, & C. Cooper (Eds.), Job control and worker health (pp. 3-23). Chichester, UK: Wiley.

Griffin, M. A., Neal, A., & Parker, S. K. (2007). A new model of work role performance: Positive behavior in uncertain and interdependent contexts. Academy of Management Journal, 50, 327-347.

Hackman, J. R. (1970). Task and task performance in research on stress. In J. E. McGrath (Ed.), Social and psychological factors in stress (pp. 202-237). New York: Holt, Rinehart & Winston.

Hackman, J. R., & Oldham, G. R. (1975). Development of the Job Diagnostic Survey. Journal of Applied Psychology, 60, 159-170.

Hackman, J. R., & Oldham, G. R. (1980). Work redesign. Reading, MA: Addison-Wesley. Hall, R. J., Snell, A. F., & Foust, M. S. (1999). Item parceling strategies in SEM: Investigating the subtle

effects of unmodeled secondary constructs. Organizational Research Methods, 2, 233-256. Hauschildt, J., & Kirchmann, E. (2001). Teamwork for innovation—The troika of promoters. R&D

Management, 31, 41-49. Herscovitch, L., & Meyer, J. P. (2002). Commitment to organizational change: Extension of a three-

component model. Journal of Applied Psychology, 87, 474-487. Jackson, P. R., Wall, T. D., Martin, R., & Davids, K. (1993). New measures of job control, cognitive

demand, and production responsibility. Journal of Applied Psychology, 78, 753-762. Jansen, K. J. (2004). From persistence to pursuit: A longitudinal examination of momentum during the

early stages of strategic change. Organization Science, 15, 276-294. Kanter, R. M. (1983). The change masters. New York: Simon & Schuster. Kanter, R. M. (1993). Men and women of the corporation (2nd ed.). New York: Basic Books. Karasek, R. (1989). Control in the workplace and its health-related aspects. In S. Sauter, J. Hurrel, &

C. Cooper (Eds.), Job control and worker health (pp. 129-159). Chichester, UK: Wiley. Karasek, R., & Theorell, T. (1990). Healthy work. Stress, productivity, and the reconstruction of working

life. New York: Basis Books. Karasek, R. A. (1979). Job demands, job decision latitude, and mental strain: Implications for job

redesign. Administrative Science Quarterly, 24, 285-308. Kline, R. B. (1998). Principles and practice of structural equation modeling. New York: Guilford. Lazarus, R. S., & Folkman, S. (1984). Stress, appraisal, and coping. New York: Springer. Legge, K. (1984). Evaluating planned organizational change. San Diego, CA: Academic Press. Martinko, M. J., & Gardner, W. L. (1982). Learned helplessness: An alternative explanation for perfor-

mance deficits. Academy of Management Review, 7, 195-204. Merron, K. (1993). Let’s bury the term “resistance.” Organizational Development Journal, 11, 77-86. Miller, V. D., Johnson, J. R., & Grau, J. (1994). Antecedents to willingness to participate in a planned

organizational change. Journal of Applied Communication Research, 22, 59-80. Morgeson, F. P., Delaney-Klinger, K., & Hemingway, M. A. (2005). The importance of job autonomy,

cognitive ability, and job-related skill for predicting role breadth and job performance. Journal of Applied Psychology, 90, 399-406.

Morrison, E. W., & Phelps, C. C. (1999). Taking charge at work: Extrarole efforts to initiate workplace change. Academy of Management Journal, 42, 403-419.

Morrow, P. C. (1983). Concept redundancy in organizational research: The case of work commitment. Academy of Management Review, 8, 486-500.

Motowidlo, S. J., & Van Scotter, J. R. (1994). Evidence that task performance should be distinguished from contextual performance. Journal of Applied Psychology, 79, 475-480.

Neubert, M. J., & Cady, S. H. (2001). Program commitment: A multi-study of its impact and antecedents. Personnel Psychology, 54, 421-448.

Ohly, S., Sonnentag, S., & Pluntke, F. (2006). Routinization, work characteristics and their relationship with creative and proactive behaviors. Journal of Organizational Behavior, 25, 257-279.

Organ, D. W. (1988). Organizational citizenship behavior: The good soldier syndrome. Lexington, MA: Lexington Books.

Parker, S. (1998). Enhancing role breadth self-efficacy: The roles of job enrichment and other organiza- tional interventions. Journal of Applied Psychology, 83, 835-852.

Parker, S. K. (2000). From passive to proactive motivation: The importance of flexible role orientations and role breadth self-efficacy. Applied Psychology: An International Review, 49, 447-469.

Hornung, Rousseau / AUTONOMY AT WORK 425

Parker, S. K., & Sprigg, C. A. (1999). Minimizing strain and maximizing learning: The role of job demands, job control, and proactive personality. Journal of Applied Psychology, 84, 925-939.

Parker, S. K., Wall, T. D., & Jackson, P. R. (1997). “That’s not my job”: Developing flexible employee work orientations. Academy of Management Journal, 40, 899-929.

Piderit, S. K. (2000). Rethinking resistance and recognizing ambivalence: A multidimensional view of attitudes toward an organizational change. Academy of Management Review, 25, 783-794.

Podsakoff, P. M., MacKenzie, S. B., Lee, J.-Y., & Podsakoff, N. P. (2003). Common method bias in behavioral research: A critical review of the literature and recommended remedies. Journal of Applied Psychology, 88, 879-903.

Porter-O’Grady, T. (1992). Implementing shared governance: Creating professional organizations. St. Louis, MO: Mosby.

Rogers, J., & Streeck, W. (Eds.). (1995). Works councils: Consultation, representation, and cooperation in industrial relations. Chicago: University of Chicago Press.

Rousseau, D. M. (1995). Psychological contracts in organizations. Understanding written and unwritten agreements. Thousand Oaks, CA: Sage.

Rousseau, D. M. (2001). Schema, promise and mutuality: The building blocks of the psychological contract. Journal of Occupational and Organizational Psychology, 74, 511-541.

Rousseau, D. M., & Shperling, Z. (2003). Pieces of the action: Ownership and the changing employment relationship. Academy of Management Review, 28, 115-134.

Rousseau, D. M., & Tijoriwala, S. A. (1999). What’s a good reason to change? Motivated reasoning and social accounts in promoting organizational change. Journal of Applied Psychology, 84, 514-528.

Seligman, M. E. P. (1975). Learned helplessness. New York: Freeman. Smith, C. S., Tisak, J., Hahn, S. E., & Schmieder, R. A. (1997). The measurement of job control. Journal

of Organizational Behavior, 18, 225-237. Sobel, M. E. (1982). Asymptotic intervals for indirect effects in structural equations models. In S. Leinhart

(Ed.), Sociological methodology (pp. 290-312). San Francisco: Jossey-Bass. Speier, C., & Frese, M. (1997). Generalized self-efficacy as a mediator and moderator between control

and complexity at work and personal initiative: A longitudinal field study in East Germany. Human Performance, 10, 171-192.

Spreitzer, G. M. (1996). Social structural characteristics of psychological empowerment. Academy of Management Journal, 39, 483-504.

Taris, T. W. (2000). A primer in longitudinal data analysis. London: Sage. Taris, T. W., & Kompier, M. (2003). Challenges of longitudinal designs in occupational health psychology.

Scandinavian Journal of Work, Environment, and Health, 29, 1-4. Van Dyne, L., Cummings, L. L., & Parks, J. M. (1995). Extra-role behaviors: In pursuit of construct and

definitional clarity (a bridge over muddied water). In L. L. Cummings & B. M. Staw (Eds.), Research in organizational behavior (Vol. 17, pp. 215-285). Greenwich, CT: JAI.

Vroom, V. H. (1964). Work and motivation. New York: John Wiley. Wanberg, C. R., & Banas, J. T. (2000). Predictors and outcomes of openness to change in a reorganizing

workplace. Journal of Applied Psychology, 85, 132-142. Weick, K. E. (1995). Sensemaking in organizations. Thousand Oaks, CA: Sage. Weick, K. E., & Quinn, R. E. (1999). Organizational change and development. Annual Review of

Psychology, 50, 361-386. Wheatley, M. J. (1992). Leadership and the new science: Learning about organization from an orderly

universe. San Francisco: Berrett Koehler. Whyte, W. H. (1956). The organization man. New York: Simon & Schuster. Wrzesniewski, A., & Dutton, J. E. (2001). Crafting a job: Revisioning employees as active crafters of their

work. Academy of Management Review, 26, 179-201.

426 THE JOURNAL OF APPLIED BEHAVIORAL SCIENCE December 2007

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.