Abstract
Despite frequent references to “the changing nature of work,” little empirical research has investigated proposed changes in work context perceptions. To address this gap, this study uses a cross-temporal meta-analysis to examine changes in five core job characteristics (e.g., task identity, task significance, skill variety, autonomy, and feedback from the job) as well as changes in the relationship between job characteristics and job satisfaction. An additional analysis of primary data is used to examine changes in two items related to interdependence. On average, workers perceived greater levels of skill variety and autonomy since 1975 and interdependence since 1985. In contrast, the results of a supplemental meta-analysis did not support significant changes in the association between the five core job characteristics and satisfaction over time. Thus, although there is some evidence for change in job characteristics, the findings do not support a change in the value placed on enriched work. Implications for researchers and organizations navigating the modern world of work are highlighted.
Keywords
“The changing nature of work” has received considerable attention among management scholars (Howard, 1995), business leaders (Maney, 2009), and the popular press (Godin, 2009). Some suggest that technology has ushered in a new industrial revolution—the age of digitalization (National Academy of Sciences, 1999). Others propose that the line between work and nonwork has blurred (Allan, Loudoun, & Peetz, 2007; Patterson, 2001), with traditional jobs and work hours replaced by fluid work roles and flexible schedules (Davis, 1995; Motowidlo & Schmit, 1999). According to Oldham and Hackman (2010: 465), “the world of work is now different than it was then, perhaps fundamentally so.” Indeed, assumed changes in the nature of work are often explicitly referenced as rationale for the importance of studying a variety of organizational phenomenon, such as: person-organization fit (Caldwell, Herold, & Fedor, 2004), organizational commitment (Solinger, van Olffen, & Roe, 2008), changes in selection systems (Murphy, 1999), alternations in performance assessment (Motowidlo & Schmit, 1999), organizational citizenship behaviors (Randall, Cropanzano, Borman, & Birjulin, 1999), as well as job design (Grant, Fried, Parker, & Frese, 2010).
Yet, scarce empirical research has sought to substantiate proposed changes (National Academy of Sciences, 1999). Much of the extant literature is anecdotal (e.g., Canton, 2006) or reflects sociological/macroeconomic interpretations of the underlying causes of presumed changes (e.g., changes in technology, globalization, or the workers themselves) rather than direct measures of psychological change from the perspective of workers using constructs common to the management literature. This is unfortunate, as work context is thought to underlie organizational behavior and attitudes (Johns, 2006), and therefore any changes to work context could necessitate changes in talent management processes. From this perspective, documenting changes in work context is critical to understanding how, if at all, talent management approaches should be adapted for 21st-century work. The current article represents the first empirical analysis of changes in employees’ perceptions of work characteristics. We address this question in three studies.
First, cross-temporal meta-analysis (Konrath, O’Brien, & Hsing, 2011; Twenge & Im, 2007) is employed to provide an empirical analysis of changes in the constructs included in Hackman and Oldham’s (1975) job characteristics model (JCM) since 1975 in studies using the associated measure of job characteristics, the Job Diagnostic Survey (JDS). By summarizing means from the population of available studies and examining change in means over time, cross-temporal meta-analysis affords robust and generalizable inferences relative to single sample comparisons (Twenge, 2011; Twenge & Campbell, 2010). Using cross-temporal meta-analysis, we investigate whether (a) mean-level perceptions of the five core job characteristics have changed over time, (b) changes differ across occupational groups, and (c) mean-level changes in job characteristics differ for samples with a higher proportion of females.
In the second study, we examine two items related to social context using a large primary sample. Despite the popularity of the JCM, many authors (Grant, Fried, & Juillerat, 2011; Grant & Parker, 2009; Humphrey, Nahrgang, & Morgeson, 2007), including its creators, have acknowledged that the original formulation of the JCM is deficient in describing the modern work context (Oldham & Hackman, 2010). Specifically, the JCM does not include variables relevant to social context (Grant, Fried, & Juillerat, 2011; Grant & Parker, 2009; Humphrey et al., 2007), such as the interpersonal aspects, interactions, and social structure of work (Johns, 2006). To allow for an analysis of social context, we located primary data to supplement our cross-temporal meta-analysis. Two items regarding the cooperation of coworkers and the provision of support from outside workgroups from a large national database (i.e., WorkTrendsTM employee opinion database; Kowske, Rasch, & Wiley, 2010) are used to assess changes in interdependence between 1985 and 2011.
In the third study, we use psychometric meta-analysis to examine changes in the relationship between job satisfaction and the five core job characteristics. Whereas the first two studies are useful in understanding changes in employee perceptions of the job characteristics, this meta-analysis will be useful to examine changes in workers themselves. The association between the five core job characteristics and job satisfaction is well established (Humphrey et al., 2007). However, to the extent that these associations have changed over time, evidence would be provided that job characteristics are differentially valued by respondents in more recent years.
In light of the evidentiary and theoretical basis substantiating context as a focal determinant of employee attitudes and behaviors (e.g., Bamberger, 2008; Dierdorff & Ellington, 2008; Dierdorff & Morgeson, 2007; Dierdorff, Rubin, & Morgeson, 2009; Fried, Grant, Levi, Hadani, & Slowik, 2007; Hackman, 2003; Johns, 2001, 2006; Juillerat, 2010; Morgeson, Dierdorff, & Hmurovic, 2010; Morgeson, Johnson, Campion, Medsker, & Mumford, 2006), a clearer understanding of changes in perceptions of the work context has several potential implications for organizations, managers, and management researchers. Given that human behavior is a function of the interaction between the employee and their environment (Lewin, 1951), identifying changes in employees’ perceptions of context can provide insight into potential trends in employee attitudes and behaviors and clearly focus researchers on specific contexts that require integration into theory (Johns, 2006). Similarly, focusing attention on key contextual changes is critical to the development of strategies to help organizations navigate the changing nature of work (Rousseau & Fried, 2001; Ryan, Schmitt, & Johnson, 1996). Moreover, doing so using a large-scale analysis, including meta-analysis and a large national dataset, enhances the applicability of these insights across areas of inquiry.
Proposed Changes in Work and Workers
Johns’ (2006) model of work context proposes that the omnibus context (e.g., technological advances, globalization, flattened organizational hierarchies) shapes the local, or discrete, context (e.g., task characteristics, social characteristics), which in turn shapes employee attitudes and behaviors. Time is an informative omnibus context variable because it provides “a surrogate for environmental stimuli occurring when the research is conducted” (Johns, 2006: 392). Omnibus contextual trends are routinely highlighted to explain predicted changes in work context and in workers themselves. Indeed, similar time-based changes in omnibus context such as “globalization, communication and information technology, [and] economics” (Myers & Sadaghiani, 2010: 235) are commonly invoked to predict both changes in discrete context (e.g., Fried et al., 2007; George & Jones, 2000; Mitchell, 1997) and changes in worker characteristics (e.g., Myers & Sadaghiani, 2010; Twenge, Campbell, Hoffman, & Lance, 2010).
Changes in workers
Management research investigating changes in organizationally relevant variables has almost exclusively focused on changes in workers (e.g., Cennamo & Gardner, 2008; Costanza, Badger, Fraser, Severt, & Gade, 2012; Hansen & Leuty, 2012; Twenge et al., 2010). Both popular and academic sources have proposed that employees increasingly value the opportunity for individual growth, skill development, flexibility, voice in important organizational decisions, and time spent completing significant and meaningful work (e.g., Cennamo & Gardner, 2008; Eby, Butts, & Lockwood, 2003; Fonner & Stache, 2012; Grant, 2007; Inkson, 2008; Myers & Sadaghiani, 2010; Strauss & Howe, 2000; Twenge et al., 2010; Winograd & Hais, 2011). In essence, this research proposes that mean workforce values have shifted toward a desire for more enriched work (Myers & Sadaghiani, 2010; Strauss & Howe, 2000; Winograd & Hais, 2011).
These propositions are consistent with a supplies-values (S-V) perspective on person-environment (P-E) fit, where fit is a function of the extent to which organizations, vocations, and jobs supply the employee with valued work characteristics and rewards (Cummings & Cooper, 1979; Edwards, 1992; French, Caplan, & Harrison, 1982). To the degree that worker values have changed, the S-V fit perspective proposes that organizations should “create job offerings and work environments that are more likely to engage and retain millennial workers” (E. Ng, Schweitzer, & Lyons, 2010: 281). In other words, changes in worker values should be used to inform organizations how to change organization and job characteristics (Hansen & Leuty, 2012; Smola & Sutton, 2002; Swanberg 2005; Swanberg & Simmons, 2008) because these changes in values impact satisfaction (Locke, 1976). Two conditions are implicit in the S-V fit interpretation of the changing nature of work. First, employee values reflect a shift toward more enriched jobs in more recent years. Second, because of the proposed shift in workforce values, enriched jobs are expected to be more satisfying in more recent years. In the third study, we report a meta-analysis of changes in the relationship between the core job characteristics and job satisfaction to determine if enriched work is more satisfying to workers in recent years.
Changes in work
Work context has been defined as “situational opportunities and constraints that affect the occurrence and meaning of organizational behavior as well as functional relationships between variables” (Johns, 2006: 386). Similar to alleged changes in workers, temporal changes in omnibus context are routinely invoked when proposing changes to the discrete work context (e.g., Cappelli et al., 1997; Grant & Parker, 2009; Howard, 1995; National Academy of Sciences, 1999; Parker, Wall, & Cordery, 2001). For instance, omnibus changes in globalization, increased competition, and the rise of the knowledge-based industry are often implicated in changes in discrete work context (e.g., Fried, Levi, & Laurence, 2008; Grant et al., 2010; Grant & Parker, 2009; Juillerat, 2010; Oldham & Hackman, 2010; O’Toole & Lawler, 2006; Parker et al., 2001).
Proposed changes in context are more consistent with demands-abilities (D-A) P-E fit (Dierdorff et al., 2009; Edwards, 1996; Kristof, 1996), where fit is a function of the worker’s ability to meet contextual demands. Specifically, whereas S-V fit processes are thought to be driven by employee values, D-A fit processes are thought to be driven by contextual demands (Livingstone, Nelson, & Barr, 1997). When discussing the changing nature of work, work context researchers typically emphasize the influence of proposed contextual changes on the knowledge, skills, and abilities (KSAs) needed to perform effectively, whereas generational research recommends changing organizational structure and processes on the basis of changing employee values. In this way, the overarching implication of the D-A fit perspective is that segments of the workforce with the requisite KSAs to meet changing contextual demands will be increasingly competitive (Kichuk & Wiesner, 1998; Latham & Sue-Chan, 1998; Stevens & Campion, 1994) and that organizations should actively target individuals with these KSAs. In Study 1, we use cross-temporal meta-analysis to examine the extent to which perceptions of job characteristics have changed since 1975; and in Study 2, we examine whether perceptions of interdependence have changed over time using a large-scale primary study.
Study 1: Changes in the Core Job Characteristics
According to the JCM, levels of the five core job characteristics (task identity, task significance, skill variety, feedback, and autonomy) act as a gauge for the motivational enrichment of jobs (Hackman & Oldham, 1975). We focus on the JCM for a few reasons. First, the corresponding JDS measure is among the oldest measures of job characteristic perceptions and “remains the dominant model of job design today” (Grant, Gino, & Hofmann, 2011: 421). The continued popularity of this framework and measure provides a large database and a standardized measure with which to examine the temporal changes needed to provide generalizable inferences (Konrath et al., 2011; Twenge, 2011; Twenge & Im, 2007). Second, past research has frequently and explicitly proposed changes in the core job characteristics (e.g., Fried et al., 2007, 2008; Morgeson & Campion, 2003; Parker et al., 2001), yet there has been no empirical substantiation of these propositions. The JCM will be useful in the examination of changes in work context perceptions because of the popularity of this model and measure and the relevance of these characteristics to outcomes (Humphrey et al., 2007). In addition to the pragmatic value, work context researchers have explicitly adopted the JCM as a framework when proposing contextual changes, supporting the relevance of the JCM in examining changes. Below we describe temporal changes in the omnibus context proposed to yield increases in each of the five discrete core job characteristics.
Task identity
Task identity is described as “the degree to which the job requires completion of a ‘whole’ and identifiable piece of work—that is, doing a job from beginning to end with a visible outcome” (Hackman & Oldham, 1975: 161). Changes in the omnibus context such as globalization, the shift from manufacturing to service, and technological advancements are proposed to yield increases in task identity (Cappelli et al., 1997; National Academy of Sciences, 1999). Specifically, to meet the demands of global competition and do more with less (Royal & Agnew, 2011), bureaucratic organizations have increasingly restructured into leaner hierarchies (Howard, 1995), and previously distinct jobs have merged into more fluid work roles (National Academy of Sciences, 1999). As noted by Nadler, Hackman, and Lawler (1979), combining multiple tasks to produce one “whole” task should yield increased perceptions of task identity and ownership. With the aid of technology, modern workers are able complete multiple, previously separate aspects of a given project more quickly and with less help then in years past (National Academy of Sciences, 1999; Van der Spiegel, 1995). Additionally, as the service industry grew, a greater proportion of jobs required interaction with customers and clients. It is possible that because employees are responsible for providing “whole” services to a given customer, task identity perceptions have increased. Based on these trends, we hypothesize:
Hypothesis 1: Mean levels of task identity perceptions have increased since 1975.
Task significance
Task significance describes “the degree to which the job has substantial impact on the lives or work of other people—whether in the immediate organization or in the external environment” (Hackman & Oldham, 1975: 161). Changes in the omnibus context such as the shift from manufacturing to service and knowledge-based work as well as the proposed demand for greater social interactions and work interdependence have potentially resulted in an increase in task significance (Grant, 2008; Morgeson & Campion, 2003; Turban & Greening, 1997). Fried et al. (2008) proposed that the rise of the service economy would result in greater contact between service providers and consumers, with employees increasingly likely to experience the direct impact of their work on others. Consistent with this, those with greater contact with beneficiaries of their work (e.g., customers; Grant, 2007) and increased interaction with others outside their organization (Morgeson & Humphrey, 2006) tend to report higher levels of significance. Next, employees increasingly work on teams (National Academy of Sciences, 1999) to complete larger, more complex projects. The interdependence associated with working on a team might also be expected to yield an increase in task significance perceptions (Cummings, 1978; Hackman, 1987). Additionally, due to flattened organizational structures, lower level employees ostensibly have greater decision latitude and more authority, potentially resulting in the perception that one’s work is more important (Humphrey et al., 2007). On the basis of increased contact with others and greater authority and responsibility across organizational levels, we predict:
Hypothesis 2: Mean levels of task significance perceptions have increased since 1975.
Skill variety
Skill variety concerns “the degree to which a job requires a variety of different activities to carry out the work, which involve the use of a number of different skills and talents of the employee” (Hackman & Oldham, 1975: 161). Based on the emergence of the knowledge economy and the expansion of employees’ roles to meet competitive demands, increases in skill variety are often proposed (Fried et al., 2008; Howard, 1995; Morgeson & Campion, 2003). In the midst of the information age, “knowledge and information drive economic growth” (Felin, Zenger, & Tomsik, 2009: 555). According to the National Academy of Sciences (1999), the cognitive demands associated with modern work, and in particular with knowledge-based jobs, are much greater than previous decades. Along with increased cognitive complexity, there is also a greater demand for “soft skills” at work. Specifically, team-based structures and increases in service jobs are expected to yield increased interpersonal interactions with customers and colleagues. In turn, modern employees must rely on cognitive and interpersonal skills to complete assignments (Fried et al., 2008; Grant & Parker, 2009), yielding a potential increase in skill variety perceptions. Finally, with trends toward flat organizational hierarchies, fewer employees perform heavier workloads compared to years past, requiring a more diverse skill set for effective performance (Davis, 1995; Fried, Hollenbeck, Slowik, Tiegs, & Ben-David, 1999). Supporting these propositions, employees since the 1970s have increasingly sought higher educational attainment (U.S. Census Bureau, 1992, 2012). Together, these trends suggest that the modern workplace requires employees to use and seek out more diverse skills than in years past. Thus, we predict:
Hypothesis 3: Mean levels of skill variety perceptions have increased since 1975.
Feedback
Feedback is described as “the degree to which carrying out the work activities required by the job results in the employee obtaining direct and clear information about the effectiveness of his or her performance” (Hackman & Oldham, 1975: 162). Technological advances allow for new ways to track progress and obtain feedback. For instance, modern software programs can identify human errors and prompt the employee to fix mistakes even before the task is completed (Hesketh & Neal, 1999). Similarly, technology can allow for real-time tracking and reporting on employee performance, facilitating more input into feedback. In addition, with the popularization of multirater feedback systems and the emphasis on performance management practices, feedback from others has also potentially increased. Although feedback from the job and others are distinct, they tend to overlap substantially (ρ = .57; Humphrey et al., 2007). Thus, it is possible that proposed increases in feedback from others might also register in respondents’ reports of feedback from the job itself.
Hypothesis 4: Mean levels of feedback perceptions have increased since 1975.
Autonomy
Autonomy refers to “the degree to which the job provides substantial freedom, independence, and discretion to the employee in scheduling the work and determining the procedures to be used in carrying it out” (Hackman & Oldham, 1975: 162). On the basis of omnibus economic changes (e.g., downsizing, doing more with less) and social changes (e.g., women’s roles), employees are working longer hours (Families and Work Institute, 2006) and there are more dual-earner families (U.S. Census Bureau, 2011). At the same time, aided by the increases in information technology, there is often less of a need for workers to report directly to work (Tam, Korczynski, & Frenkel, 2002). Autonomy in the form of flexible work schedules is increasingly important and practical to help employees manage dual roles (Allen, Herst, Bruck, & Sutton, 2000; Halpern, 2005; Hill, Hawkins, Ferris, & Weitzman, 2001) and has been found to be increasing in later years (Matos & Galinsky, 2012). In addition to changes in scheduling autonomy, employees are proposed to experience increased autonomy in on-the-job decision-making in recent years (Fried et al., 1999; Howard, 1995). With global competition and the resulting flattening organizational hierarchies, decision-making authority is proposed to be more diffuse, with lower-level employees granted a greater span of control and more responsibility than in previous decades (Cappelli, 1999; Howard, 1995). Based on these trends, we predict:
Hypothesis 5: Mean levels of autonomy perceptions have increased since 1975.
Finally, we examine the possibility of nonlinear changes in the job characteristics over time. As noted by an anonymous reviewer, even if significant linear effects are not revealed, it is still possible that job characteristics have undergone nonlinear changes. For instance, it is possible that fluctuations in the U.S. economy will register a nonlinear trend. Thus, in order to provide as comprehensive an analysis as possible, we provide an exploratory analysis of nonlinear changes in the five core job characteristics.
Occupational Complexity
Of course, to some extent, changes in job characteristics perceptions are a reflection of changes in the occupational makeup of the broader economy (Levy & Murnane, 2004). This is important in the context of the present study for two reasons. First, it is important to ensure that observed effects are not attributable to differences in the occupations sampled in the literature over the years. To account for the potential differential sampling of occupations, we control for occupational complexity, or the skill-level necessary for successful performance in a given occupation (Hollingshead, 1974). To the extent that occupational characteristics have changed, controlling for changes in occupational complexity provides conservative estimates of the proposed changes in job characteristics. However, there is evidence that changes in job characteristics are thought to occur within and across occupations (Murphy & Welch, 1993). Accordingly, we expect significant changes in the job characteristics since 1975 even when controlling for occupational characteristics.
Second, it is possible that changes have differentially affected different occupational groups. For instance, it is possible that more highly skilled occupations have seen increases in the core job characteristics whereas lower skilled occupations have evidenced decreased or stable level of enrichment. Similarly, with middle-skill jobs and routine tasks being replaced by technology, the economy has become polarized into higher- and lower-skill groups (Acemoglu & Autor, 2011). In light of polarization, we also examine whether changes in job characteristics are moderated by occupational complexity.
Research Question 1: Will changes in job characteristics differ as a function of occupational complexity?
Sample Gender
In 2010, the workforce was composed of more women than men for the first time in U.S. history (50.2%; U.S. Census Bureau, 2012). In addition, women are increasingly occupying more important and prestigious roles in organizations. For instance, in 1972, only 16% of management positions were held by women; but by 2009, women held over half of all management positions (51.4%, U.S. Census Bureau, 1975, 2011). Given increasing number of women entering the workplace and increased female representation in professional and managerial roles, we expect that samples with more women to report more enriched work in recent years relative to earlier years.
Hypothesis 6: The relationship between year and perceptions of job characteristics will be stronger for samples with more women.
Study 2: Changes in Interdependence
Social context is an important aspect of work context that was not captured by the JDS, and interdependence is among the most studied aspects of the social context (Dierdorff & Morgeson, 2007; Grant, Gino, & Hofmann, 2011; Grant & Parker, 2009; Humphrey et al., 2007; Kelley et al., 2003). Jobs are considered interdependent when “the people performing them must rely on or collaborate with others to complete their work” (Dean & Snell, 1991: 784).
Shifts in organizational structures and the implementation of team-based work are commonly proposed to predicate increasing interdependence perceptions. As global competition has increased, organizations have cut levels of management resulting in leaner hierarchies (Cappelli, 1999; Howard, 1995), with greater flexibility to navigate changing competitive demands and market uncertainty (Cappelli, 1999). In these flatter hierarchies, work is often organized by project teams, rather than with a traditional hierarchical functional orientation (O’Toole & Lawler, 2006). Flattening and the shift into project-based work enable the management of more diverse, complex projects and greater flexibility. Consistent with these trends, the National Academy of Sciences (1999: 270) notes that the increased use of teams is “among the most visible changes in the structure of work.” The rise in teamwork is often cited as evidence that modern jobs have become highly interdependent (Grant & Parker, 2009). Team-based structures require employees to communicate with teammates, coordinate responsibilities, and work with others in order to successfully perform on the job (Stevens & Campion, 1994). We propose:
Hypothesis 7: Mean levels of interdependence perceptions have increased since 1985.
Study 3: Changes in Work vs. Changes in Workers
Although the primary purpose of this study is the examination of changes in job characteristics, a secondary purpose is to examine whether employees’ value for certain job characteristics have also changed. There are two approaches to examine changes in the value workers place on enriched jobs. First, research has directly compared self-reported work values across the years. Despite frequent predictions, the available empirical literature typically does not substantiate proposed increases in workforce value for enriched work. Past research, typically falling under the auspices of generational research, has examined changes in worker values for various job characteristics. For example, primary studies have compared the value for autonomy (Hansen & Leuty, 2012; Lester, Standifer, Schultz, & Windor, 2012), skill variety (Hansen & Leuty, 2012), and overall enrichment (Twenge et al., 2010) across generations of workers. These studies have uniformly indicated weak and nonsignificant evidence of changes in value for enrichment. Indeed, Twenge et al. (2010) actually documented the reverse trend: slight decreases in the value for enriched and socially interdependent work among millennials compared to earlier generations. Together, the extant literature does not seem to support increases in the value for enriched work.
The second approach, and the one adopted in this study, is examining whether the relationship between certain job characteristics and job satisfaction has changed in recent years. As noted above, changes in job characteristics would be consistent with a D-A perspective on fit; however, changes in worker value for a given characteristic would be consistent with an S-V perspective on fit. Specifically, the relationship between a given job characteristic and job satisfaction differs based on the extent to which an employee values the work characteristic (Locke, 1976). From this perspective, changes in the relationship between the job characteristic and job satisfaction are reflective of temporal changes in the value for the job characteristic.
Specifically, if employees value more enriched work in more recent years, then changes in this relationship would reflect employees’ increased desire for enriched work. This finding would be important because it would provide organizations with important information about the type of job characteristics that are desired by employees and prospective employees, thus providing valuable information in the recruitment, retention, and motivation of the workforce. On the other hand, a lack of significant changes in the association between job characteristics and job satisfaction would indicate that modern employees have a similar desire for enriched work as their predecessors. Despite numerous studies investigating mean changes in self-reported value for enriched work, past studies have not yet considered whether the relationship between job characteristics and job satisfaction has changed over time. Consequently, a secondary purpose of this study is to provide a moderated meta-analysis exploring the relationship between job characteristics and job satisfaction across time. Because past research has not examined this relationship and because generational researchers have proposed both increases and decreases in the value for enriched work, we do not offer a formal hypothesis for these analyses.
Research Question 2: Has the relationship between perceptions of job characteristics and job satisfaction changed since 1975?
Study 1 Method
Literature Search
Studies were located using Web of Science, Google Scholar, and Dissertation Abstracts International. In order to ensure that articles relevant to the JCM were identified, two citation searches (Hackman & Oldham, 1975, 1976) along with key word searches (i.e., “job characteristics model; “job characteristics theory”) were used, producing 2,249 nonredundant citations between 1975 and May 2011.
Criteria for Inclusion
Studies were included if they (a) measured one or more job characteristic with the JDS (Hackman & Oldham, 1975) or the Revised Job Diagnostic Survey 1 (JDS-R; Idaszak & Drasgow, 1987) using the original 7-point scale; 2 (b) reported the mean and associated sample size for at least one job characteristic; and (c) were conducted using U.S.-based workplace samples. Consistent with past cross-temporal meta-analyses (Twenge & Campbell, 2001; Twenge & Im, 2007), we focused on U.S. samples to avoid confounding our results with cultural differences and to ensure as much homogeneity as possible in the studies included (Fried et al., 2008). Based on these criteria, 102 studies were included in the cross temporal meta-analysis, yielding 107 independent samples and a total sample size of between 12,033 and 15,510 depending on the specific job characteristic.
Coding of Studies
Main effects and gender
The mean, standard deviation, and sample size associated with available job characteristics were collected from each study. As suggested by Oliver and Hyde (1993) and based on common practice in prior cross temporal meta-analyses (Twenge, 2000, 2001; Twenge & Campbell, 2001), the year of data collection was coded as two years before the study’s publication unless stated otherwise. Additionally, consistent with past meta-analyses investigating gender differences (e.g., Eagly, Karau, & Makhijani, 1995; T. Ng, Eby, Sorensen, & Feldman, 2005), the percentage of males in the sample was also recorded. All studies were coded by at least two trained coders, and agreement among raters was high (93%). Disagreements were resolved by discussion.
Occupational complexity
Each primary study reporting occupational or job-type information (about 60%) was coded using Hollingshead’s Index (Hollingshead, 1957). This index is a commonly used occupational categorization (Adams & Weakliem, 2011) that groups occupations based on the complexity of skills necessary for performance (Table 1) ranging from 9 (most complex) to 1 (least complex). The occupation from each primary study was searched using Hollingshead’s list and coded as high (score of 9, 8, or 7), middle (score of 6, 5, or 4) and low (score of 3, 2, or 1). Studies with multiple occupations in one sample were only included when a single occupational code could be derived (e.g., a sample of physicians and surgeons would both be coded as high complexity). When occupations in one sample were not of the same complexity level, complexity was coded as unable to code. Agreement between coders was found to be high (87%). Two types of analyses were conducted using occupational complexity. The first was a control analysis to determine whether the relationship between job characteristics and year holds when complexity level is controlled. In the second, moderated multiple regression was used to examine whether occupational complexity moderated the job characteristics–year relationship.
Hollingshead Index of Occupational Categories
Cross-Temporal Meta-Analysis
Cross-temporal meta-analysis differs from typical meta-analysis in that cross-temporal meta-analysis focuses on mean levels of a scale (Twenge, 2001; Twenge & Campbell, 2001; Twenge, Konrath, Foster, Campbell, & Bushman, 2008; Twenge & Nolen-Hoeksema, 2002), rather than summarizing effect sizes. Specifically, the correlation between the year data were collected and relevant scale means from the study is computed to analyze changes in the mean-levels of a given construct over time. Consistent with prior cross-temporal meta-analyses (Gentile, Twenge, & Campbell, 2010; Twenge, 2001; Twenge & Campbell, 2001; Twenge et al., 2008; Twenge, Zhang, & Im, 2004), we weighted data points using the inverse variance, w, which weights means by the variance and sample size (Shadish & Haddock, 1994). In this method, the standard deviation (SD) from each individual study is squared and multiplied by 1/n of each study. This product is inverted to obtain the weighting term, w = (1/v). In total, 67% of data points included in this meta-analysis reported an SD, and when not reported, the average SD per job characteristic was used to obtain the weighting term. To calculate the magnitude of change in JDS scores, we conducted correlation and regression analyses. A significant relationship between year and the job characteristic is interpreted as evidence that the job characteristic has linearly changed over time. As typical in cross-temporal meta-analysis, we use the regression equation to derive predicted mean values of each job characteristic for the earliest (i.e., 1975) and latest year (i.e., 2011) and use this to calculate Cohen’s d. The weighted correlation estimates linear change, or how well the observed change approximates a linear trend. On the other hand, Cohen’s d estimates total change between the two time periods, regardless of whether the change conforms to a linear trend (Twenge, 2001; Twenge et al., 2008).
Study 1 Results
Main Effects
Hypotheses 1 through 5 predicted increases in the five core job characteristics (Table 2). Contrary to Hypotheses 1 and 2, neither task identity nor significance evidenced a significant increase since 1975. However, supporting Hypothesis 3, mean levels of skill variety increased significantly since 1975. These results indicate a large increase in the predicted level of skill variety for 1975 and 2011 (d = .94; Table 3). Hypothesis 4 was rejected; feedback has not significantly increased. Lastly, in support of Hypothesis 5, autonomy has significantly increased since 1975 and produced a moderate effect size (d = .42).
Meta-Analysis of Changes in Perceived Job Characteristics
Note: N = total sample size; k = number of independent samples; M = overall mean; SD = standard deviation of the mean; rw = inverse variance weighted correlation with year.
p < .01.
p < .001.
Predicted Means and Effect Size Associated With Changes in Job Characteristics
Note: Predicted means were estimated using results of the regression analysis. 1975 Mean = predicted mean for 1975; 2011 M = predicted mean for 2011; SD = average standard deviation of primary studies; d = Cohen’s d for the predicted value of 1975 and 2011.
Occupational Complexity Controls and Moderation
Controlling for occupational complexity levels, both skill variety and autonomy continued to evidence a significant relationship with year (Table 4). This provides evidence for an increase in levels of these job characteristics above and beyond what is explained by occupational complexity. However, for both of these variables, when the interaction term was entered in the final step, the relationships fell, indicating that high-, middle-, and low-complexity occupations have all experienced a similar increase in skill variety and autonomy in later years.
Effect Sizes Controlling for Occupational Complexity Levels
Note: k = number of independent samples, B = unstandardized coefficient; SE B = standard error of B; β = standardized coefficient; t = computed value of t-test.
p < .01.
p < .001.
Sample Gender and Changes in Job Characteristics
Moderated multiple regression analyses revealed a significant interaction between sample gender and year in predicting both mean-level task significance and feedback, providing partial support for Hypothesis 6 (Table 5). As depicted in Figures 1 and 2, for both task significance and feedback, samples with a higher proportion of women reported increased task significance and feedback in more recent years. Specifically, samples with a larger percentage of females reported significant increases in task significance (b = .03, t = 3.19, p < .01), whereas samples with more males reported significantly less task significance (b = −.02, t = −2.72, p < .01) in recent years. Similarly, samples with more females reported significant increases in feedback (b = .04, t = 4.98, p < .001), while samples with more men reported a modest but nonsignificant decrease in feedback (b = −.01, t = −1.955, p = .057). Interestingly, when occupational complexity was controlled, both interactions with sample gender dropped to nonsignificance.
Moderated Multiple Regression of Sample Gender Moderating the Relationship Between Year and Job Characteristics
Note: Male = 1; Female = 0; k = number of independent samples; B = unstandardized coefficient; SE B = standard error of B; β = standardized coefficient; t = computed value of t-test.
p < .05.
p < .01.
p < .001.
Interaction of Year and Sample Percent Male Predicting Mean Levels of Task Significance
Interaction of Year and Sample Percent Male Predicting Mean Levels of Feedback
Supplemental Nonlinear Analysis and Results
To examine whether quadratic or cubic terms explained significant variance beyond the linear term, we conducted curvilinear analyses. First, we created quadratic and cubic terms, using centered year to reduce multicolinearity. Then, using the linear regression function, centered year was entered in step one, followed by centered year squared in step two, and finally, centered year cubed in Step 3. Results are displayed in Table 6. Skill variety and autonomy evidenced only significant linear effects. Neither the linear nor the nonlinear models fit feedback. On the other hand, evidence of a nonlinear trend was supported for both task identity and task significance. Task identity showed evidence of a quadratic effect, and task significance showed a cubic effect. For task identity, an increase of 1.07 SD was observed between 1975 and 1991, at which point it began to decrease, showing a 1.05 SD decrease from 1991 to 2011. The cubic function of task significance showed that from 1975 to 1985 there was a sharp decrease of 2 SD, which mostly stabilized from 1985 to 2000, and then continued to decrease sharply thereafter. Both task identity and task significance had negative βs as their final terms, indicating decreases in more recent years (i.e., task identity displays an inverted U and task significance begins with an upward concave and ends with a downward concave).
Results of Linear, Quadratic, and Cubic Regression Models Predicting Job Characteristic Mean
Note: Cases are weighted by the inverse variance.
Study 1 Discussion
Study 1 provided the first analysis of changes in task context perceptions and supported mean-level increases in skill variety and autonomy since 1975, even when controlling for occupational complexity. Increases in perceptions of skill variety and autonomy are consistent with our expectations of increasingly enriched work based on the onset of the knowledge economy and the era of “doing more with less.” In other words, jobs, on average, demand greater levels of diverse skills from workers and demand that workers complete their tasks with less direct supervision.
Despite nonsignificant linear trends for task identity and significance, both evidenced significant nonlinear trends. Specifically, although task identity increased until 1991, as computers saw increasing usage, organizations were potentially better able to segment work both functionally and geographically. Likewise, in manufacturing settings, tasks traditionally completed by humans were increasingly automated. In both cases, it is possible that increasing technology resulted in a lower proportion of employees who were able to see the final, tangible results of their labor, triggering a drop in task identity. Task significance reduced initially then stabilized throughout the 1990s and again dropped significantly in the last decade of the study. It is possible that the falling levels of task significance are attributable to leaner organizations, resulting greater overemployment. Specifically, through the 1980s the U.S. economy was undergoing an era of downsizing, potentially resulting in overqualified employees working in lower level jobs. Then, in the final decade of the study, economic struggles and the associated reductions in force potentially triggered another increase in overemployment and decrease in task significance. We hasten to note that these interpretations are speculative and encourage research to more provide a more direct examination of these nonlinear trends.
Despite the insights gained in Study 1, the cross-temporal meta-analysis did not allow for an examination of social context due to the fact that the JDS does not measure social context. As noted by Oldham and Hackman (2010: 467) “we did not view social factors as essential for fostering internal work motivation. Indeed, we thought that we had identified the core motivational properties of jobs and that other properties were of relatively little motivational consequence.” We did, however, attempt a cross-temporal meta-analysis of interdependence, but the most popular measures of interdependence (i.e., Dean & Snell, 1991; Pearce & Gregersen, 1991) are relatively new and have not been used in sufficient studies to allow for the inclusion in a cross-temporal meta-analysis. Thus, the analysis of social context necessitated the identification of a primary sample that had measured social context over an extended time period.
Sudy 2 Method
Survey and Sample
The data for Study 2 are from the WorkTrendsTM employee opinion survey. This survey has been distributed annually since 1985 and is described elsewhere (Kaplan, Wiley, & Maertz, 2011; Kowske et al., 2010). In the present study, we used all available years of data on two interdependence items ranging from 1985 to 2011.
Measure and Analysis
Two items from the WorkTrendsTM survey were relevant to social context: “The people I work with cooperate to get the job done” (total N of 185,681, across 25 years, 1985-2011, omitting 2000) and “Other work groups give us the support we need to serve our customers” (total N of 138,881, across 17 years, 1993-2010). To determine if interdependence has increased over time, analyses of both items were conducted using analysis of variance (ANOVA). 3 Respondent gender and occupational complexity (high, middle, low; Hollingshead, 1957) were examined as covariates.
Study 2 Results
Hypothesis 7 proposed increases in work interdependence with later years. For the first interdependence item, “The people I work with cooperate to get the job done,” the ANOVA was significant, F(101, 103,408) = 16.82, p < .001, R2 = .015, supporting the predicted increase in interdependence. The main effect of gender was nonsignificant, F(1, 103,408) = 1.715, p = .190; whereas the main effects for year, F(16, 103,408) = 21.54, p < .001, η2 = .003; and skill level, F(2, 103,408) = 488.34, p < .001, η2 = .009 were significant. Significant two-way interactions were found for the interaction between year and skill level, F(32, 103,408) = 1.93, p = .001, η2 = .001 (higher interdependence found with higher complexity jobs; see Figure 3) and gender and year, F(16, 103,408) = 3.14, p < .001, η2 < .001 (men and women reported similar levels of interdependence until the mid-2000s; Figure 4). The interaction between gender and skill level was nonsignificant, as was the three-way interaction of all factors.
Changes in Interdependence (Item 1) by Occupational Complexity
Differences in Interdependence (Item 1) by Gender
The ANOVA for the second interdependence item, “Other work groups give us the support we need to serve our customers,” was also significant, F(71, 81,430) = 19.34, p < .001, R2 = .016, again supporting an increase in interdependence. The main effects of gender, F(1, 81,430) = 27.37, p < .001, η2 < .001; year, F(11, 81,430) = 54.95, p < .001, η2 = .007; and skill level, F(2, 81,430) = 221.79, p < .001, η2 = .005, were all significant. Significant two-way interactions were found for the interaction between year and skill level, F(22, 81,430) = 2.94, p = .001, η2 = .001 (in general, higher skill levels also had higher interdependence; Figure 5); gender and year, F(11, 81,430) = 2.75, p < .001, η2 < .001 (compared to women, men reported slightly lower interdependence until the mid-2000s; Figure 6); and gender and skill level F(2, 81,430) = 12.59, p < .001, η2 < .001 (men and women in higher skilled jobs reported similar high levels of interdependence; in low and medium complexity jobs, women reported higher interdependence; Figure 7). The three-way interaction was nonsignificant.
Changes in Interdependence (Item 2) by Occupational Complexity
Changes in Interdependence (Item 2) by Gender
Differences in Interdependence (Item 2) by Occupational Complexity and Gender
Study 2 Discussion
The analysis of the two WorkTrendsTM items supports predicted increases in interdependence and provides some evidence that interdependent work varies by occupational complexity and gender. These findings support suggestions that work requires greater cooperation with others to effectively complete job task and, thus, suggest that management systems and processes that reinforce cooperative behaviors will be increasingly important.
Although this study was valuable in allowing for an examination of changes in social context not possible with cross-temporal meta-analysis, a few limitations warrant discussion. The use of single item measures of interdependence was necessary in order to examine social context, as there was no other available data to allow for this examination. Other researchers have also used single-item measures to examine change over time for this reason (e.g., Highhouse, Zickar, & Yankelevich, 2010; Kowske et al., 2010). In addition, the content of these items are consistent with items on multi-item interdependence scales (e.g., “How much does success in this unit depend on cooperation with other people?” [Dean and Snell, 1991]; and “I frequently must coordinate my efforts with others” [Pearce & Gregersen, 1991]). Having provided empirical evidence for increases in mean levels of skill variety, autonomy, and interdependence, we next turned our attention to examining whether the relationships between the core job characteristics and job satisfaction has changed over time.
Study 3 Method
Study 3 considered changes in the relationship between the core job characteristics and job satisfaction. The literature search used the same method as Study 1. To be included in this meta-analysis, the study had to (a) be conducted in an organization, (b) measure one or more of the five core job characteristics using the JDS, and (c) report a correlation of the relationship between job satisfaction and one or more of the five core job characteristics. We also coded sample size along with predictor and criterion reliability. On the basis of the criteria for inclusion, 37 studies, published between 1975 and May 2011, met the inclusion criteria, yielding ks between 31 and 45 for each job characteristic.
Meta-Analysis Methods
We used an adapted version of Arthur, Bennett, and Huffcutt’s (2001) SAS PROC MEANS program to conduct the meta-analysis based on Raju, Burke, Normand, and Langlois’s (1991) random effects meta-analysis method. Individual correlations were weighted by N as well as corrected for predictor and criterion unreliability. For studies that did not report reliability information, the average reliability was used.
To determine if the relationship between each job characteristic and job satisfaction has changed, separate continuous moderator analyses were conducted with year as the moderator. Hedges and Olkin’s (1985) meta-regression method was used for this analysis. This method uses weighted least squares multiple regression, where the correlations are weighted by the inverse sample variance (w). Specifically, the correlation between job satisfaction and a job characteristic is regressed onto the continuous moderator, year. Significant results indicate that the job satisfaction–job characteristics relationship varies by year.
Study 3 Results
Table 7 presents the meta-analytic results for the associations between each job characteristic and job satisfaction. The meta-analytic estimates were consistent with past reviews of these relationships (e.g., Humphrey et al., 2007). We next regressed the relationship between job satisfaction and each of the job characteristics onto year using weighted least squares regression to examine whether this relationship is moderated by year. The results indicated that the relationship between job satisfaction and each of the job characteristics was not significantly associated with year (Table 8).
Meta-Analysis of the Relationships Between JDS Job Characteristics and Job Satisfaction
Note: N = total sample size; k = number of independent samples; r = uncorrected mean weighted correlation; ρ = rho, population effect size; SD ρ = standard deviation of rho; CR = credibility interval; CI = confidence interval; Q = Q-statistic.
Results of the Job Characteristic–Job Satisfaction Relationship Regressed on Year
Note: Adjusted standard errors are in parentheses following effect sizes.
Study 3 Discussion
Despite frequent speculation of changes in the value of enriched work to modern workers, the results of Study 3 do not support the idea that the core job characteristics are significantly more satisfying to modern workers. Specifically, the association between the core job characteristics and job satisfaction did not differ significantly over time. These findings correspond with other findings that examined changes in the value for enriched work using self-reports of employee values (e.g., Lester et al., 2012; Real, Mitnick, & Maloney, 2010; Twenge et al., 2010). When taken with the findings of Study 1 and 2, these findings suggest that although there is evidence that job characteristics have changed, these changes do not appear to be a function of organizations changing policies to meet changing preferences of employees. To the contrary, it seems as though the nature of work has changed, regardless of whether employees desire the changes or not.
However, a few limitations should be noted. First, there were a relatively small number of studies for the analyses, meaning there may not have been sufficient power to detect significant effects. Further, the small number of studies precluded the investigation of potential nonlinear relationships that might have emerged. However, the number of studies available here are on par with other meta-analyses that have examined continuous moderators. In addition, although we only included studies that used the JDS to measure job characteristics, in order to bolster the number of independent samples, we included all measures of job satisfaction (e.g., non-JDS measures of job satisfaction were included). Finally, because the analysis of social context is a relatively recent advancement in the management literature, we were unable to examine whether the relationship between interdependence and satisfaction has significantly changed over time.
General Discussion
Changes in the nature of work are commonly referenced but rarely empirically scrutinized. The present findings substantiate proposed increases in skill variety, autonomy, and interdependence, yet failed to support proposed increases in task significance, task identity, and feedback. Despite apparent increases in skill variety, autonomy, and interdependence, the results of the third study do not support the idea that enriched jobs are more satisfying to employees relative to earlier years. These findings provide a needed empirical foundation to evaluate and refine past conceptual propositions of changes in the nature of work. In light of the reemergence of context as a focal work-related variable (Dierdorff & Morgeson, 2007; Dierdorff et al., 2009; Fried et al., 2007; Grant et al., 2010; Johns, 2006), empirically isolating salient contextual features has implications for a variety of management functions. We highlight three overarching implications, including (a) the empirical description of modern work context, (b) a contextual interpretation of proposed changes in worker values, and (c) the implications of observed changes in context for research and practice across areas of inquiry.
Main Findings: Describing the Context of Modern Work
By operationalizing omnibus context using time (Johns, 2006), this study describes changes in work context since 1975 and, in doing so, provides an empirically based description of the modern work context in the U.S. Given the critical role of work context in shaping employee attitudes and behaviors (Johns, 2006), the present findings are especially useful in directing attention to contextual features of increasing relevance to today’s organizations and researchers. From the outset we should note that the present findings concern average job-level changes and, thus, would not be expected to apply to all individuals in the U.S. workplace.
Consistent with the proposition that omnibus changes in global competition and a trend toward leaner organizations have yielded increasing skill variety, moderate increases in skill variety were supported. Omnibus changes have yielded a modern work context that demands employees who possess the ability to gain the skills needed to fill multiple functional roles (Cappelli et al., 1997) and is increasingly complex in terms of cognitive demands (National Academy of Sciences, 1999). Given changing demands, these findings support the idea that intelligence, learning orientation, and the openness needed to learn and apply new information will potentially become as important to job success as cultivated technical skills (e.g., “learn how” vs. “know how”; Fried et al., 2008).
Next, our findings indicate that modern jobs offer greater control in determining when and how work is completed relative to the 1970s. Autonomy is typically conceptualized as a key work resource (Karasek, 1979), one that is beneficial to employees in navigating an increasingly demanding work schedule (Rau & Hyland, 2002). Given increases in hours worked and dual-income families, it is possible that autonomy has become a necessary requirement to balance family and work. Nevertheless, it is noteworthy that autonomy was characterized by the lowest predicted mean of the five core job characteristics in both 1975 and 2011. With the documented evidence of increases in autonomy in scheduling work (Matos & Galinsky, 2012), the relatively low level of autonomy was surprising. Given that autonomy is sometimes conceptualized as a multifaceted construct (cf. Breaugh, 1985; Breaugh & Becker, 1987; Latham & Pinder, 2005; Morgeson & Humphrey, 2006), it is possible that differential effects across facets attenuated observed findings. With the JDS assessing global autonomy and few popular facet-level measures of autonomy until recent years, we could not examine facet-level changes, but future research on this topic should be conducted. For instance, although autonomy in scheduling policies may be on the rise, autonomy in decision making may not have increased as rapidly, attenuating the overall effects. Finally, given the nature of most employment contracts and organizational structures, some degree of organizational control (e.g., supervision) and performance monitoring has been and will likely continue to be necessary for most jobs.
It is also important to acknowledge potential challenges associated with increasing autonomy. For instance, a recent high-profile media story reported on a programmer who outsourced his job to a worker in China at a fraction of the cost. Not only did this employee remain undetected while outsourcing his job, but the employee’s supervisor rated him as one of the highest-performing programmers in the business unit (Inocencio, 2013). This case is illustrative of the difficulty inherent in managing autonomous employees (e.g., Hertel, Geister, & Konradt, 2005; Hertel, Konradt, & Orlikowski, 2004). Given increasing levels of autonomy, increased attention is needed to identify strategies that provide employees autonomy, while simultaneously facilitating effective performance management processes.
Next, our findings implicate increasing perceptions of interdependence as a key change characterizing the modern workplace. These findings suggest that, along with the increasing cognitive skills and learning orientation needed to navigate the modern workplace, “soft skills” like interpersonal skills and the ability to get along with others are also increasingly important. In spite of, or perhaps because of, the trend toward a more interdependent workplace, the Society for Human Resource Management (SHRM; 2008a, 2008b; SHRM & WSJ.com/Careers, 2008) as well as the National Association of Colleges and Employers (NACE; 2012) list interpersonal skills as among the top skills most needed in the workplace; yet interpersonal skills are also among the top-rated deficiencies of employees entering the workforce (SHRM, 2008a, 2008b). Clearly, attention must be given to the identification and development of employees with the interpersonal skills required in the modern work context.
The observed increases in autonomy and interdependence may seem paradoxical. However, as layers of management have been reduced and replaced with team-based structures, workers have increased autonomy in that there is less direct supervision. Employees are more interdependent with peers in terms of making decisions, potentially yielding increases in both autonomy and interdependence. In addition, any increase in the scheduling aspect of autonomy would not be expected to impact interdependence. Thus, the findings depict a workforce in which employees work with less supervision but are more reliant on lateral relationships to effectively complete their work.
The nonsignificant linear trends for task identity, task significance, and feedback from the job contribute empirical evidence to our understanding of what has not changed, despite popular opinion (e.g., Cappelli et al., 1997; Grant, 2008; Morgeson & Campion, 2003; Turban & Greening, 1997). There are a few caveats to these nonsignificant findings. Notably, it is possible that the time period of the study was insufficient to detect trends in these variables. However, given vast economic changes over the past few decades, the lack of evidence of a linear trend is important as they suggest that in recent history, these job characteristics have not increased significantly relatively to the 1970s. For example, some have cautioned that the availability of feedback enabling technologies has created the risk of feedback overload (Lurie & Swaminathan, 2008; Van der Spiegel, 1995). Yet the present analysis did not support this proposed increase. Given the increasing complexity of jobs, especially with regards to interpersonal demands, feedback seems especially important to effective performance management in the modern workforce. A second caveat to these findings is that the nonlinear trends for task significance and identity provide some evidence for changes in task identity and task significance; however, both suggested a downward trend in recent years.
A third caveat is that nonsignificant main effects for feedback and task significance were qualified by a significant interaction with the percentage of women in the sample, with samples with more women reporting higher levels of both since 1975. To the extent that women’s occupations have become more professionally oriented over the past 30 years, the increase in task significance is not surprising. Similarly, to the extent that more important assignments require more feedback, it makes sense that the observed increase in task significance for samples with more women was accompanied by an increase in feedback. Given the lack of a significant interaction between gender and skill variety and autonomy, one potential roadblock to women’s continued advancement to the highest organizational levels may be reduced access to developmental experiences that actually facilitate the growth in skills needed to advance. Consistent with this interpretation, recent research found that although both male and female managers reported an equal number of developmental work experiences, these experiences were qualitatively different in that men’s experiences were significantly more challenging than those of women (King et al., 2012). It is however important to acknowledge that both moderation effects dropped to nonsignificance when occupational complexity was controlled. From this perspective, it seems that changes in the job characteristics of samples with more women is potentially attributable to changes in occupational complexity rather than changes to task significance and feedback within occupational groups.
Changes in Work, Not Workers
The results of Study 3 did not support the suggestion that modern workers find enriched work more satisfying than earlier workers. These findings correspond with past studies that have generally failed to support differences in values for enriched work among more recent workers (e.g., Twenge et al., 2010). On the basis of the collective pattern of results, these findings point to a D-A fit perspective rather than an S-V perspective on the changing nature of work (e.g., Lester et al., 2012). That is, although job characteristics have changed, employees do not find enriched work more satisfying than in previous years.
It is noteworthy that the three job characteristics that did evidence increases (autonomy, skill variety, and interdependence) are among the most commonly proposed changes to worker values (e.g., Hansen & Leuty, 2012; Lester et al., 2012; Myers & Sadaghiani, 2010; Twenge et al., 2010). Given that our review shows that these three aspects of context have increased, it seems possible that generational commentators have made a fundamental attribution error in observing changes in work.
Specifically, generational commentators have potentially misattributed anecdotal evidence of changes in patterns of employee behavior (e.g., an analysis of employee behavior would show that modern workers work with others more closely in more responsible and skilled jobs) to an internal cause: changes in employees’ values for these characteristics. Based on our findings, these patterns of behavior have an external cause: changing context in the form of increased prevalence of interdependent, skilled, and autonomous work. The source of changes (individual versus contextual) has important implications for the appropriate organizational response to the changing nature of work. In this way, recommendations to change work context to recruit, retain, and effectively manage employees on the basis of supposed changes in value for enrichment are potentially misguided. Instead, theorists and practitioners should focus on understanding the selection and development of employees with KSAs needed to succeed in the modern environment.
Given the increases in enriched work supported here, it is somewhat surprising that other studies have generally failed to support significant increases in job satisfaction over time (Bowling, Hoepf, & LaHuis, 2011; Costanza, Badger, Fraser, Severt, & Gade, 2012). In other words, to the extent that jobs are more enriched, we would expect workers to be more satisfied. There are a few potential explanations for this disconnect. One reason could be that there is a curvilinear relationship between job characteristics and satisfaction. According to Warr’s (1987) Vitamin Model, job characteristics yield increases in employee well-being until a saturation point is reached, at which point increased enrichment is associated with reductions in employee well-being. Indeed, a small body of research has evinced curvilinear relationships between job characteristics and outcomes (Champoux, 1980, 1981, 1992; Fletcher & Jones, 1993; Warr, 1990; Xie & Johns, 1995). As levels of enrichment have increased over time, it is possible that modern jobs are overenriched (Johns, 2010). Consistent with this logic, Twenge et al. (2010) supported a slight decrease in the value for enrichment across generations, whereas others have reported an increase in the value for work-life balance (Smola & Sutton, 2002; Twenge et al., 2010). Although never examined in the context of temporal changes, some research is supportive of the potential for negative effects of overenrichment. For instance, skill variety (Chen & Chiu, 2009), autonomy (De Jonge & Schaufeli, 1998; Golden & Veiga, 2005), interdependence (Harrison, Johns, & Martocchio, 2000; Wong, DeSanctis, & Staudenmayer, 2007), and job scope (i.e., combination of job characteristics; Xie & Johns, 1995; Champoux, 1980) have all been empirically linked with deleterious outcomes for workers, such as conflict, stress, and burnout. To the extent that autonomy, skill variety, and interdependence have increased as a set, it seems particularly relevant for researchers and practitioners to be aware of the potential for adverse side effects and curvilinear effects associated with these changes. Although only suggestive, this pattern of results warrants further attention.
An anonymous reviewer proposed an alternative interpretation of the finding that jobs are more enriched despite minimal prior evidence for increases in job satisfaction. Specifically, the failure of past studies to support increases in job satisfaction could also reflect adaptation by modern workers who have become accustomed to more enriched jobs. According to adaption level theory (Helson, 1947), prior exposure to stimuli impacts an individual’s reference point. From this perspective, it is possible that increasingly enriched work may not lead to greater satisfaction because employees have adapted to the current level of enrichment and subsequently have increased their work enrichment expectations. In other words, employees expect enrichment and increasing levels of job characteristics are needed to keep employees satisfied at baseline levels. Unfortunately, the present data do not allow us to tease apart whether workers have simply adapted to and come to expect high levels of enrichment or whether there is a disconnect between increasing levels of enrichment but not satisfaction.
Implications for Research
Understanding of context is essential to effective research and practice across human resource systems (Dierdorff et al., 2009). By describing modern work context and changes in work context, our findings have implications for numerous lines of inquiry and human resource management processes. Below we highlight the implications of changes in work context for three exemplary areas of management research, including: work context, performance appraisal, and leadership.
Most directly, our findings have theoretical implications for the study of work context. The present study tests a central aspect of Johns’s (2006) model of work context: that the temporal component of omnibus context influences discrete work context. The results support the proposed model of context change and codify the value of year as a valuable and explanatory operationalization of omnibus context. Future studies and in particular meta-analyses can benefit from incorporating time into analyses. In addition, although numerous work context changes have been proposed (e.g., Dierdorff et al., 2009; Grant, 2008; Grant et al., 2011; Morgeson & Campion, 2003; Parker et al., 2001), the present study helps to refine these proposals by identifying those features that actually have changed. Thus, these results can provide a useful guide to researchers when discussing the changing nature of work and when interpreting temporal differences that emerge in future studies.
Next, the present findings have implications for criterion development and performance appraisal research. From a D-A fit perspective, increases in interdependence will demand employees who have the ability to work effectively with others. Consistent with this, organizational citizenship behaviors (OCB) have been shown to be more redundant with ratings of task performance in interdependent contexts (Bachrach, Powell, Bendoly, & Richey, 2006). Given increases in interdependence, it is possible that OCB and task performance will be increasingly difficult to distinguish in modern contexts. In this way, this study has the potential to inform criterion theory by pointing to a change in discrete context that has the potential to change the dimensionality of the criterion space. Similarly, given increases in interdependence, it is possible that OCB will be increasingly rewarded by organizational decision-makers, implying that OCB will be potentially more instrumental to employee effectiveness in coming years.
Next, these documented changes in work context also have potential implications for leadership research. Based on our results, leaders increasingly manage employees who complete complex assignments, work closely with others, and do so with considerable autonomy. Grant, Gino, and Hofmann (2011) presented a series of studies that found that extraverted leaders may undermine employee proactivity in enriched jobs and proposed that communal leadership might be more valuable in such settings. To the extent that modern work is increasingly enriched, communal leader behaviors might be more desirable across settings. Consistent with this perspective, DeRue, Nahrgang, Wellman, and Humphrey’s (2011) meta-analytic path model indicated that consideration leader behaviors were the strongest predictors of leadership outcomes, and Dierdorff et al. (2009) found interpersonal requirements to be a universally important characteristic of managerial work. Similarly, Pilcher (2012) found that supervisor-subordinate relationship was the most critical aspect of effective performance management. However, our results suggest that it is possible that this broader pattern of results is bound by time.
Indeed, Gentry, Harris, Baker, and Leslie (2008) found increases in the perceived importance of relationship-building for effective leadership in more recent years. In this way, it might be possible that the prototype of an effective leader has changed. It is possible that the traditional top-down style of leadership is less effective, on average, than in years past (e.g., Drath et al., 2008; Grant et al., 2011). To the extent that jobs are increasingly enriched, prescribed leader behavior may have shifted to a more communal form of leadership. Yet because these characteristics were not relevant to earlier leadership theories, they have rarely been explicitly considered. In this way, building these contextual features into the analysis of leadership is a critical need. For instance, we are unaware of leader contingency theories or even many individual studies that have explicitly examined the relative efficacy of various leader traits and behaviors in highly complex, autonomous, and/or interdependent settings. This is a timely area for research.
Practical Implications
The primary practical implication is that this study points to areas of context that warrant attention from organizations. First, organizations should be wary of changing organizational practices to meet some presumed change in employee values. Instead, work context itself has changed and, thus, organizations should focus on providing all employees with the resources needed for success in the modern work context. Next, given the trends documented here, it will be increasingly important to select employees with the KSAs needed to succeed in interdependent, complex, and autonomous settings. Thus, the average organization should consider including tools such as cognitive ability, learning orientation, and traits needed to get along with others when selecting employees. Similarly, training programs devoted toward working effectively through interpersonal conflict will likely be of increasing value to organizations. Finally, leader development programs should explicitly build in content devoted toward managing employees in the modern context. Similarly, should research confirm the proposed increase in importance of consideration leader behaviors, older leader development programs should be revised to include an emphasis on this form of leadership.
Limitations
Although the present set of studies contributes to the literature in multiple ways, it is important to acknowledge what these results do not tell us. First, our cross-temporal meta-analysis results center on trends in mean levels of job characteristics over time in the U.S. and should not be interpreted as relevant to every job in every country. Clearly, there are still many modern jobs characterized by low levels of autonomy, skill variety, and interdependence. Additionally, our results and implications are only applicable to the time period assessed and thus cannot aid in the understanding of job characteristics pre-1975. In addition, the analyses may have lacked power to detect significant effects. On the other hand, the significant effects were moderate to strong in magnitude, and our results were based on a larger amount of data (N = 12,033 to 15,510) than has typically been used to study time-based changes (cf. Highhouse et al., 2010; Twenge et al., 2010). Next, it is possible that expectations for job enrichment has changed over time and so are not necessarily comparable between time points (i.e., gamma change). Unfortunately, because our analyses were conducted at the scale (Study 1) or item (Study 2) level of analysis, we were unable to examine measurement equivalence. Research assessing measurement equivalence over time is needed.
Next, the literature on work characteristics has seen something of a renaissance in recent years (cf. Campion, Mumford, Morgeson, & Nahrgang, 2005; Fried et al., 2007, 2008; Grant, 2007; Grant & Parker, 2009; Grant et al., 2010, 2011; Humphrey et al., 2007; Morgeson et al., 2010; Parker et al., 2001). Recent work has focused on additional work characteristics not included in the original JCM (e.g., knowledge, social, relational, proactive, physical characteristics) as well as multifaceted conceptualizations of characteristics such as autonomy. As the literature base examining this expanded set of contextual characteristics grows, we urge future research to more directly examine changes in other aspects of work context.
As is often the case with meta-analysis, locating sample characteristics that are reported consistently across studies is challenging. Although it would have been informative to examine factors such as flextime plans and the nature of the organizational structure, this information is rarely, if ever, presented in primary studies. Additionally, our results might not be applicable to the description of contextual changes in other countries and cultures, and future cross-cultural studies are needed in this regard. Finally, to investigate the moderating role of gender, we used the percentage of males in the samples. Although this approach is not uncommon in meta-analysis (Eagly et al., 1995; Ng et al., 2005), it is not ideal. Future research replicating the gender findings at the individual level is needed.
Conclusion
Despite frequent reference to the changing nature of work, research has rarely documented proposed changes. This study provides three large-scale analyses to provide the first empirical foundation for changes in work context. In light of the key role that context plays in shaping employee attitudes and behaviors, the contextual features identified here will play an increasing critical role in management research and practice. Accordingly, it is critical that management researchers redouble their efforts to understand the implications of a work context characterized by high levels of employee autonomy, skill variety, and interdependence.
Footnotes
Acknowledgements
This article was accepted under the editorship of Patrick M. Wright. The authors would like to acknowledge Lillian Eby and Ron Piccolo for their helpful comments on earlier versions of this article and Amber Davidson, Danielle Eaton, Matt Griffith, Lindsey Griffin, Thieu Ha, Shelby Hughes, Lauren Locklear, Anna Lorys, and Jorge Lumbreras for their assistance with data collection. Lauren Wegman is currently employed by Yahoo, Inc. An earlier version of Study 1 was presented as Lauren Wegman’s Master’s Thesis at the University of Georgia under the supervision of Brian Hoffman.
Supplemental material for this article is available with the manuscript on the JOM website.
Notes
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
