Factors Buffering Against the Effects of Job Demands

Abstract

Given the increasing role that paid work is likely to play in older adulthood in the coming decades, the goal of this study was to understand the circumstances under which work is related to mental health for older adults and whether these circumstances differ by age. Using a multiworksite sample of 1,812 U.S. workers age 18 to 81, we use the life-span theory of control to hypothesize that older and younger workers may benefit differentially from job and personal control in the context of high job demands. Results suggest that for younger workers with high personal control, job control buffers the impact of job demands on mental health. For older workers, personal control alone buffers the impact of job demands on mental health. This study adds to previous research by addressing how the factors thought to buffer against the effects of job demands differ cross-sectionally by age.

Keywords

age job demands job control personal control mental health

The increased presence of older adults in the U.S. workplace—a byproduct of an aging population, increased longevity, and a shrinking prime age worker population (Toosi, 2007)—has amplified the need for research on older workers. Numerous studies have shown that work experiences differ across the life span (Birdi, Warr, & Oswald, 1995; Cennamo & Gardner, 2008; Kanfer & Ackerman, 2004; Rhodes, 1983; Shultz & Adams, 2007). Job satisfaction and employee engagement have been found to vary with age (Bernal, Snyder, & McDaniel, 1998; Clark, Oswald, & Warr, 1996; Hochwarter, Ferris, Perrewé, Witt, & Kiewitz, 2001; James, McKechnie, & Swanberg, 2011; Pitt-Catsouphes & Matz-Costa, 2008, 2009), as has the meaning of work (Loi & Shultz, 2007; Smyer & Pitt-Catsouphes, 2007). Few studies, however, have explored how workers of different ages experience job demands and workload.

One of many perceptions about the capabilities of workers of different ages is the notion that older workers are physically unable to deal with high levels of job demands (Hedge, Borman, & Lammlein, 2006; Lyon & Pollard, 1997; Rosen & Jerdee, 1976). While there is research focused on how declines in older adulthood might impact workers’ ability (e.g. Ilmarinen, 2001; Sluiter, 2006), studies have not shown a negative association between age and job performance (McEvoy & Cascio, 1989; Ng & Feldman, 2008), suggesting that older workers are capable of continued productive work despite age-related losses. Still, questions arise as to whether older workers are able to deal with high levels of job demands.

In the current work environment, employees’ job demands may have increased as organizations have “downsized” or “rightsized” in response to the economic downturn, making this an important time to examine how age affects employees’ ability to deal with job demands. High levels of job demands are often related to decreased well-being (Bakker, Demerouti, & Euwema, 2005; Schaufeli, Bakker, & Rhenen, 2009; Schaufeli, Bakker, van der Heijden, & Prins, 2009); however, not everyone experiences the effects of job demands in the same way. One person with high levels of demands might wither under the pressure while another might thrive. Furthermore, there may be differences related to age in the experience of job demands, such that the impact of job demands on the well-being of a 50 year old differs from that of a 25 year old.

Building on the Job Demands-Control Model (DCM), the current study examines the extent to which personal and job control serve to buffer the negative relationship between job demands and mental health (Johnson & Hall, 1988; Karasek, 1979; Karasek & Theorell, 1990) and whether this relationship differs for workers of different ages. Based on Heckhausen and Schulz’s (1995) life-span theory of control, which suggests that the types of control strategies employed by individuals change over the life-span, we propose that job and personal control interact in important ways to either buffer or exacerbate the negative relationship between job demands and mental health at younger ages. At older ages, however, we propose that the role of personal control becomes more prominent (and job control less prominent) as a protective buffer against the negative impact of job demands on mental health.

Job Demands-Control Model (DCM)

In the DCM, excessive job demands coupled with little job control are thought to create high job strain and, in turn, poor health outcomes. In contrast, low-strain jobs are those with low job demands and high job control (Karasek & Theorell, 1990). According to this model, high job control may buffer the impact of high job demands (van der Doef & Maes, 1999). Job demands are defined “as psychological demands such as high workpace, time pressures, and difficult work” and job control, often referred to as decision latitude, is “compris(ed) of the extent of authority to make decisions concerning the job [known as task authority] and the breadth of skills needed to perform the job [known as skill discretion]” (Daniels & Harris, 2005, p. 220).

Studies utilizing the DCM have produced mixed results; specifically there has been a failure to find interactive effects between job control and demands in predicting well-being (Melamed, Kushnir, & Meir, 1991; Rodríguez, Bravo, Peiró, & Schaufeli, 2001; van der Doef & Maes, 1999; Wall, Jackson, Mullarkey, & Parker, 1996). A possible explanation for this lack of finding is the omission of key variables. Research has proposed that the DCM may operate differently based on aspects of personal control, which can be defined as “the extent of freedom or choice one has over his or her own behavior” (Grandey, Fisk, & Steiner, 2005, p. 893), and is often measured as self-efficacy and/or locus of control (Skinner, 1996). One study by Daniels and Guppy (1994) found that the DCM only applies for individuals with an internal work locus of control. Schaubroeck and Merritt (1997) showed that the model applied for individuals with high self-efficacy. Salanova, Peiro, and Schaufeli (2002) found that increasing control for workers with low self-efficacy increased rather than decreased job stress. These findings suggest that workers who do not feel very confident in their ability to be effective in life (low self-efficacy) or who see the world as beyond their control (external locus of control)—both forms of personal control—may experience a high degree of job control as stressful, resulting in lower well-being.

Inconsistencies in the literature with regard to the DCM may also be explained by the lack of attention to the applicability of the model among workers of different ages. Indeed, one study found that only one of the job control mechanisms—having sufficient time to complete tasks—served as a buffer for younger workers, while all of the control mechanisms—having sufficient time to complete tasks, autonomy, and schedule flexibility—were significant buffers for older workers, suggesting that job control is especially important to the well-being of older workers (Shultz, Wang, Crimmins, & Fisher, 2010).

In a second study, Perrewe and Anthony (1990) examined age as a moderator of the job control–demands relationship and the job control–health strain relationship in a sample of steel pipe mill workers and found that age moderated both of these relationships. High levels of job control were related to lower work overload/health strain for older workers but higher work overload/health strain for younger workers. These findings are consistent with Shultz et al. (2010), and suggest that high job control may be more beneficial for older workers than for younger workers. Perrewe and Anthony’s study did not, however, explicitly test for a three-way interaction between age, job control, and job demands on well-being, so it is unclear from these findings how exactly the DCM varies with age. The life-span theory of control (Heckhausen & Schulz, 1995) may provide insight as to why the factors that buffer against job demands would not be the same for adults of all ages.

Life-Span Theory of Control

The life-span theory of control focuses on the use of primary and secondary control strategies (Heckhausen & Schulz, 1995). Primary control “involves attempts to change the world so that it fits the needs and desires of the individual” (Heckhausen & Schulz, 1995, p. 285). Secondary control involves changing oneself to fit with the environment. Primary control is exerted on the external environment and involves actual external action, while secondary control is exerted internally and involves internal cognitive action (Heckhausen & Schulz, 1995).

The salience of primary and secondary control is thought to change across the life-span. According to Heckhausen and Schulz, there is an inverted U-shaped trajectory of primary control capacity across the life-span; specifically it increases from early adulthood into midlife and decreases with the loss of social roles and physical fitness commonly experienced in older adulthood. In contrast, secondary control is thought to increase across the life-span with the highest levels occurring in older adulthood. It is argued that secondary control serves to compensate for losses in primary control capacity with age (Schulz & Heckhausen, 1996).

The life-span theory of control has not commonly been applied in workplace research. We propose, however, that job control may serve as primary control within the domain of work, as it represents control exerted on the external environment and involves the ability to adapt work demands to fit personal needs. When individuals have high job control, they are able to proactively modify their environment in various ways to compensate for the resource losses associated with aging. For example, individuals may strategically delegate tasks to coworkers or subordinates, make modifications to the means of approaching the task, or exert more effort toward a challenging task. Also, people bring several personal characteristics with them to the workplace, including personal control, which impacts the negotiation of job demands. Since personal control may be exerted internally and involves the ability to adapt oneself to meet work demands, we argue that it may serve as secondary control at the workplace. Secondary control strategies at work may include internal recalibrations of the importance or necessity of particular tasks within one’s job, positively reframing a task to highlight one’s own strengths, or comparing oneself positively to coworkers.

Well-being in later life is thought to depend on the effective use of secondary control strategies to compensate in domains in which primary control striving is deemed ineffective. For example, one can only reconfigure work tasks to a certain extent before such environmental modifications become ineffective and mental recalibrations, such as a more optimistic attitude, become more adaptive. The life-span theory of control highlights the declining importance of primary control and the increasing importance of secondary control to well-being in later life. In the next section, we use arguments derived from this theory and previous research to hypothesize that older and younger workers may benefit differentially by primary control, in this case job control, and secondary control, in this case personal control, in the context of high job demands, and thus, the processes of the DCM will change with age accordingly.

Hypotheses

The first set of hypotheses addresses the relationship between job demands and mental health. The DCM and previous research suggest that there is a negative relationship between job demands and mental health (de Lange, Taris, Kompier, Houtman, & Bongers, 2004). Shultz et al. (2010) found that this relationship did not vary based on age, with higher job demands corresponding to a greater likelihood of perceiving stress in both younger and older workers.

Hypothesis 1: Job demands are expected to be negatively associated with mental health, regardless of employee age.

In studies examining the relationship between general job control and mental health, positive relationships between the two have been found (van der Doef & Maes, 1999). Shultz et al. (2010), using facet-specific job control, found that the main effect of having the time to get the job done decreased the likelihood of perceiving stress for both older and younger workers. However, the main effects of control around flexibility and autonomy did not significantly increase or decrease the likelihood of perceived stress for either age group. Perrewe and Anthony (1990) found a positive relationship between general job control and health strain (measured using a somatic complaints checklist) for younger workers, whereas this relationship reversed for older workers. However, it is unclear whether the relationship between general job control and mental health differs for younger and older workers. Building on the life-span theory of control, job control—which we propose is a form of primary control at work—will have a greater impact on outcomes for younger workers compared to older workers, as job control becomes less effective in compensating for age-related losses in older adulthood (Schulz & Heckhausen, 1996).

Hypothesis 2a: Job control is expected to be positively associated with mental health.

Hypothesis 2b: The strength of the positive relationship between job control and mental health is expected to weaken with employee age.

Our third set of hypotheses addresses the relationship between personal control and mental health. The positive relationship between personal control and well-being is well documented (Kunzmann, Little, & Smith, 2002; Myers & Diener, 1995; Peterson, 1999). This relationship may be dependent on age, however. One study found that perceived control was a positive predictor of life satisfaction for young, middle-aged, and older adults but that the relationship was stronger as age increased (Lang & Heckhausen, 2001). Furthermore, according to the life-span theory of control, while secondary control—which we assess using personal control—is important at all ages, its use is seen as particularly adaptive when primary control striving is deemed ineffective in overcoming age-related losses. Thus, the impact of secondary control on mental health is also expected to increase with age.

Hypothesis 3a: Personal control is expected to be positively associated with mental health.

Hypothesis 3b: The strength of the positive relationship between personal control and mental health is expected to increase with employee age.

Next, we address the buffer hypothesis of the DCM—that higher job control will buffer against the negative effects of high job demands on individual outcomes (Johnson & Hall, 1988; Karasek & Theorell, 1990). Previous research has suggested that the role of job control as a moderator of the job demands–mental health relationship may be dependent on age (Perrewe & Anthony, 1990; Shultz et al., 2010). Per the life-span theory of control, at younger ages, both primary and secondary control are thought to be increasing (Schulz & Heckhausen, 1996), thus both would be expected to impact one’s ability to deal with job demands. However, previous research has also suggested that job control may only buffer against the impact of job demands when personal control is high (Daniels & Guppy, 1994; Salanova et al., 2002; Schaubroeck & Merritt, 1997).

Hypothesis 4a: At younger ages, both job control and personal control are expected to moderate the negative relationship between job demands and mental health, with job control buffering the negative relationship at high levels of personal control and exacerbating the negative relationship at low levels of personal control.

At older ages, Schulz and Heckhausen (1996) contend that primary control striving becomes less effective in overcoming age-related losses, while secondary control striving may become more effective and serves to compensate for the loss of primary control (Schulz & Heckhausen, 1996). Therefore, job control is expected to be less salient as a buffer against the negative impact of job demands and personal control is expected to be more salient as a buffer among older workers.

Hypothesis 4b: At older ages, only personal control is expected to buffer the negative relationship between job demands and mental health; job control is not expected to impact the relationship.

Method

Data and Sample

We used data from the Age & Generations Study, conducted in 2007/2008 by the Sloan Center on Aging and Work (SCAW; Pitt-Catsouphes, Matz-Costa, & Besen, 2009). The nine organizations participating in the study were affiliated with a range of industry sectors: Education (2 organizations); health care (2); retail (1); finance and insurance (2); professional, scientific, and technical services (1); and pharmaceuticals (1). All respondents worked in U.S. locations; all organizations had over 1,000 employees. Participating organizations selected one or two departments to take part in the study; employees were invited to complete a survey during company time. In total, we collected responses from 2,195 employees from 13 departments within the nine organizations. Within-department response rates ranged from a low of 28.5% to a high of 88.8%, with an average response rate of 55.3%.

Employees in the sample were 60% female and 41 years old on average. Seventy-three percent were married/cohabiting; 43% had children under age 19. Eighty-six percent worked full-time and 14% worked part-time. The average organizational tenure was 8.5 years. Fifty-one percent of the employees were hourly; these workers earned US$22 per hr on average. Forty-nine percent were salaried employees earning an average of US$80,000 per year.

Measures

Mental Health: Mental health was measured using the Mental Component Summary (MCS) of the SF-8 (standard, 4-week recall version; Ware, Kosinski, Dewey, & Gandek, 2001). The SF-8 is an 8-item shortened version of the SF-36, which is designed to measure eight domains of health: Physical functioning, role limitations due to physical health, bodily pain, general health perceptions, vitality, social functioning, role limitations due to emotional problems, and mental health. The SF-8 yields two summary measures: The Physical Component Summary (PCS) and Mental Component Summary (MCS), the latter of which was used in the current analysis. The SF-8 has been found to have excellent psychometric properties and its scores correlate highly with those produced by the SF-36 (Ware et al., 2001). Items were scored using Ware et al.’s proprietary, norm-based scoring algorithm (Cronbach’s alpha = 0.86). In this sample, scores ranged from a low of 9 to a high of 70. Higher scores indicate better mental health.

Job Demands: To measure job demands, a 5-item work overload scale was used. Work overload has been used as a measure of job demands in previous research (Perrewe & Anthony, 1990; Sargent & Terry, 2000). Four items were adapted from Wallace (1997) that asked the extent to which respondents agreed with the following statements: “I do not have enough time to get everything done in my job,” “My workload is too heavy,” “I have to work very quickly to get everything done,” and “I do not have enough time to do my work to the best of my ability.” An original item, “I can work at a comfortable pace” (reverse scored) was included as well. The response scale ranged from (1) strongly disagree to (6) strongly agree. Items were averaged to create an overall score (Cronbach’s alpha = 0.90).

Job Control: Job control is measured using two items adapted from the Job Characteristics Inventory (Sims, Szilagyi, & Keller, 1976). Respondents were asked to indicate to what extent, on a 4-point scale ranging from (1) not at all to (4) to a great extent, their jobs give them opportunities “for independent thought or action” and “to do a number of different things.” In line with previous conceptions of job control, these items addressed decision authority and skill discretion, respectively. Items were averaged to create an overall mean scale score (Cronbach’s alpha = 0.73).

Personal Control: The constructs of locus of control and self-efficacy are frequently associated with personal control (e.g., Peterson & Stunkard, 1989, 1992; Ross & Broh, 2000; Skinner, 1996). Judge, Bono, Erez, & Locke (2005) found that items tapping locus of control and self-efficacy, together with items tapping self-esteem and neuroticism, tend to display a unitary factor structure that represents a higher order latent construct, termed “core self-evaluations.”

For our purposes, we adapted seven items from the 12-item Core Self-Evaluations Scale (CSE; Judge, Erez, Bono, & Thoresen, 2003) as a proxy measure of personal control. The three items reflecting neuroticism were omitted due to the strong relationship of those items with the outcome variable and two items regarding job-specific control were omitted as well. Respondents were asked to indicate how much they agreed, on a 6-point scale ranging from (1) strongly disagree to (6) strongly agree, with seven statements about themselves. Sample items include “When I try, I generally succeed” and “I determine what will happen in my life.” The seven items were averaged to create an overall mean scale score with negatively worded items reverse scored. Higher scores indicate an internal locus of control, high generalized self-efficacy, and high self-esteem—or greater personal control. The Cronbach’s alpha for the scale was 0.84. Extreme cases were bottom-coded to a value of 3 to help normalize the distribution of this variable. Since it could be argued that self-esteem is not part of the personal control construct, we also ran models excluding these items and results were substantively identical.

Age: Age was measured continuously as the respondent’s date of birth subtracted from the date of completion of the survey.

Control Variables: We controlled for several variables that have been thought to relate to mental health or to the variables of interest in this study. These included, gender (1 = female, 0 = male), marital status (1 = married or cohabitating, 0 = not married or cohabitating), caregiving status (1 = provided child or elder care on a regular basis, 0 = did not provide care), race/ethnicity (1 = White, 0 = non-White), education (1 = bachelors’ degree or higher, 0 = less than bachelors’ degree), total household income (before taxes, from all sources; 1 = under US$20,000, 2 = US$20,000 to US$39,999, 3 = US$40,000 to US$59,999, 4 = US$60,000 to US$79,999, 5 = US$80,000 to US$99,999, 6 = US$100,000 to US$119,999, 7 = US$120,000 to US$139,999, and 8 = US$140,000 and above), work hours (the number of hours worked in a typical week), and job type. Job type was categorized based on the 2010 Standard Occupational Classification (SOC). The 2010 SOC suggested six higher levels of aggregation for job type including (1) management, business, science and arts occupations, (2) service occupations, (3) sales and office occupations, (4) natural resources, construction, and maintenance occupations, (5) production, transportation, and material moving occupations, and (6) military specific occupations. Our sample did not include any military specific occupations. There were very few respondents in the natural resources, construction and maintenance occupations group and so that group was combined with production, transportation, and material moving occupations for analytic purposes. Job type was measured with a series of four dummies representing the above mentioned occupational classifications: (1) management, (2) service, (3) sales, and (4) construction/production (reference group).

Means and standard deviations for the entire sample and by age group can be found in Table 1.

Table 1.

Means and Standard Deviations for the Entire Sample and for Younger and Older Workers.

	Full sample (N = 1812)		Younger workers (N = 854; Under age 40)		Older workers (N = 958; Age 40 plus)		Age difference^a
	M	SD	M	SD	M	SD	Significance Level	η ²
Caregiver	.54	.51	.48	.51	.60	.51	*	.03
Married	.64	.48	.52	.50	.74	.44	*	.06
Full-time	.88	.33	.90	.30	.85	.35	*	.01
Female	.59	.49	.54	.50	.64	.48	*	.01
White	.85	.36	.80	.40	.90	.31	*	.03
Bachelor’s degree or higher	.66	.47	.70	.46	.63	.49	*	.01
Supervisor	.34	47	.29	.46	.39	.49	*	.01
Household income^b	4.79	2.24	4.13	2.13	5.38	2.22	*	.14
Tenure	8.56	8.63	3.80	3.92	12.80	9.51	*	.32
Hours	40.66	9.73	41.10	9.13	40.27	10.27		.00
Management	.69	.46	.63	.49	.75	.44	*	.03
Service	.12	.32	.18	.39	.06	.23	*	.05
Sales	.07	.26	.07	.27	.07	.25		.00
Construction/production	.12	.33	.11	.32	.13	.34		.00
Age	41.03	12.32	29.92	5.11	50.93	7.41	*	.75
Job demands	3.32	1.03	3.24	1.03	3.40	1.03	*	.01
Job control	3.19	.71	3.13	.70	3.23	.71	*	.01
Personal control	4.78	.60	4.79	.61	4.78	.60		.00
Mental health	50.93	8.41	49.68	9.02	52.05	7.73	*	.02

Independent samples t test comparing means for younger and older workers.

Coded as follows: (1) under US$20,000, (2) US$20,000 to US$39,999, (3) US$40,000 to US$59,999, (4) US$60,000 to US$79,999, (5) US$80,000 to US$99,999, (6) US$100,000 to US$119,999, (7) US$120,000 to US$139,999, and (8) US$140,000 and above.

p <.01.

Data Analysis

One of the key assumptions when using standard regression techniques (ordinary least squares regression) is that observations are independent. When nonindependence of observations is an issue due to groups (such as departments or organizations) but not controlled for appropriately in statistical models, the results can be biased (Hox, 2002; Kreft & de Leeuw, 1998; Raudenbush & Bryk, 2002). Before proceeding with our analyses, we assessed whether analytic adjustments were needed to address the potential bias introduced by the multilevel nature of the data (employees nested in departments and organizations). To do this we calculated the intraclass correlation coefficient (ICC) for employees nested within organizations as well as departments. The ICC in both cases was found to be approximately 0.5%. Hox (2002) suggests that an ICC of 5.0% or less is negligible and observations can be treated as statistically independent in analyses.

A common problem in survey research is missing data resulting from participants not completing all questions in the survey as a result of time constraints, lack of interest, or unwillingness to answer certain questions. Using listwise deletion for the analyses in this article would have resulted in a loss of 717 cases or 32.3% of the sample. In listwise deletion, all noncomplete cases are excluded from analyses, thus, not only is valid information lost, but also sample size and accordingly, power, is reduced and the resulting subsample may not be representative of the full study sample (Schafer & Olsen, 1998). In an effort to restore these missing values and prevent biasing results, we used Stata IC, 11.0 (the ICE package, Royston, 2005) to implement the multivariate imputation by chained equations (MICE) method (van Buuren, Boshuizen, & Knook, 1999). In this approach, a series of conditional distributions are generated using models appropriate to the distributional assumptions of each variable being imputed (e.g., linear, Poisson, logistic, etc.). The final estimates presented for multivariate analyses are averaged across 10 complete-datasets according to Rubin’s (1987) rules. Imputed values for the dependent variable were restored to missing before proceeding with analyses (von Hippel, 2007), resulting in a final N of 1,812. Estimates obtained through listwise deletion did not substantively vary from the estimates presented.

Moderated multiple regression analyses were conducted in which mental health was regressed on controls, job demands, age, job control, personal control, and two-, three-, and four-way interactions between job demands, age, job control, and personal control. We built our model in a series of sequential steps for two reasons (1) so that the relative contribution of each variable (or sets of variables) could be assessed through the change in the R² statistic and (2) so that the “terms of lower order are partialled from those of higher order and not vice versa” (Cohen, 1978, p. 866). Thus, interaction terms—represented by the cross-product of the constituent variables—were entered after main effects (Cohen & Cohen, 1983). All variables involved in interaction terms have been standardized (transformed to have a mean of zero and a standard deviation of one) for analyses in order to reduce issues of multicollinearity and improve interpretability of the interaction terms (Jaccard & Turrisi, 2003). The Variance Inflation Factor (VIF) ranges from a low of 1.1 to a high of 3.8, with an average of 1.5, suggesting that multicollinearity is not an issue among the variables in the analyses.

Results

Table 2 reports standardized (β) coefficients and R²-change from a series of sequentially built moderated regression analyses predicting mental health. As expected in hypothesis 1, higher job demands were found to be negatively associated mental health (β = −.16, p < .001). The introduction of job demands to the model accounted for an additional 2.5% of the variance in mental health beyond that explained by the control variables. Age was found to be positively associated with mental health (β = .17, p < .001), and it explained an additional 1.8% of the variance in mental health. In line with Shultz et al.’s (2010) findings, the interaction between age and job demands was not associated with mental health (β = −.01, p > .05). Higher job control (β = .05, p < .05) and personal control (β = .42, p < .001) were associated with greater mental health, lending support for hypotheses 2a and 3a, respectively. These predictors accounted for an additional 16.1% of the variance in mental health. Significant interactions were found between personal control and job demands (β = .06, p < .001) and age and job control (β = −.05, p < .01) in predicting mental health, supporting hypothesis 2b. Contrary to hypothesis 3b, however, the interaction between personal control and age in predicting mental health was not significant (β = −.04, p > .05), suggesting that the strength of the positive relationship between personal control and mental health did not increase with age in this sample. The two-way interaction terms accounted for a statistically significant 1.0% of the variance in mental health.

Table 2.

Moderated Regression Analyses of Factors Predicting Mental Health (N = 1,812).

Predictors	Step 1	Step 2	Step 3	Step 4	Step 5	Step 6	Step 7	Step 8
Predictors	β	β	β	β	β	β	β	β
Caregiver	−.07*	−.06*	−.06*	−.05*	−.07**	−.07**	−.07**	−.06**
Married	.05	.06*	.05	.05	.06*	.06*	.06*	.06*
Full-time	.02	.00	.00	.00	.02	.00	.00	.00
Female	−.01	−.01	−.01	−.01	−.03	−.04	−.04	−.04
White	−.04	−.04	−.04	−.04	−.04	−.04	−.04	−.04
Bachelor’s degree or higher	−.04	−.04	−.04	−.04	−.01	−.01	.00	.00
Supervisor	.00	.02	.02	.02	−.01	−.01	−.01	−.01
Household Income^a	.12***	.13***	.08*	.08*	.03	.02	.03	.02
Tenure	.05*	.06*	−.03	−.03	−.01	.00	.00	.00
Hours	−.06	−.02	−.01	−.01	−.03	−.02	−.01	−.01
Management^b	.04	.05	.08	.08	−.03	−.03	−.03	−.03
Service^b	.06	.05	.07*	.07*	−.02	−.01	−.01	−.01
Sales^b	.02	.02	.04	.03	−.03	−.02	−.02	−.02
Job demands		−.16***	−.17***	−.17***	−.07***	−.07***	−.08***	−.08***
Age			.17***	.17***	.16***	.16***	.16***	.16***
Age by job demands				−.01	−.02	−.02	−.02	−.01
Job control					.05*	.05*	.05*	.05*
Personal control					.42***	.42***	.41***	.41***
Job control by job demands						−.01	−.01	−.01
Personal control by job demands						.06**	.07**	.08***
Job control by personal control						.03	.03	.04
Age by job control						−.05*	−.05*	−.06**
Age by personal control						−.04	−.04	−.04
Job control by personal control by job demands							.04	.03
Age by job control by job demands							−.01	.00
Age by personal control by job demands							.04	.04
Age by job control by personal control							.03	.03
Age by job control by personal control by job demands								−.06*
R ²	.029***	.054**	.072***	.072***	.233***	.242***	.247***	.250***
ΔR²		.025***	.018***	.000	.161***	.010***	.005*	.003**

Notes: All continuous variables in the model are standardized.

Reference group is Construction/Production.

p < .05. **p < .01. ***p < .001.

Following Aiken and West’s (1991) procedure, the two-way interaction between job demands and personal control was plotted (see Figure 1) for one standard deviation above and below the mean of personal control. As can been seen, the negative relationship between job demands and mental health decreases as personal control increases, suggesting that the impact of job demands on mental health is less severe in workers with higher personal control. To explore hypothesis 2b, the relationship between job control and mental health was plotted for workers at age 25, age 45, and age 65 (Figure 2). As expected, the positive relationship between job control and mental health appears to decrease with age, with the relationship becoming slightly negative for the oldest workers, suggesting that job control may be especially important for mental health at younger ages.

Figure 1.

Personal control as a moderator of the job demands–mental health relationship.

Figure 2.

Age as a moderator of the job control–mental health relationship.

As can be seen in Table 2, none of the three-way interactions were significant, although combined they accounted for 0.5% of the variance in mental health, which was statistically significant (p<.05). In the final step of the model, as predicted in hypothesis 4, the four-way interaction between age, job control, personal control, and job demands was significant (β = −.06, p < .05) and it accounted for a statistically significant 0.3% of the variance in mental health beyond that explained in all previous steps. In total, 25.0% of the variance in mental health was accounted for by the predictors in the model.

Four-way interactions can be very complex to interpret. Therefore, to aid in further elucidating the effects found here, we have plotted the three-way interactions between job control, personal control, and job demands for adults age 30 and age 50. In line with hypothesis 4a, Figure 3 reveals that among younger workers, job control (i.e., having the opportunity for independent thought at work and skill discretion) appears to buffer the negative relationship between job demands and mental health for workers with high personal control (i.e., greater self-efficacy), but not for those with low personal control. Among younger workers with low personal control, higher job control actually exacerbates the negative relationship between job demands and mental health. As proposed in hypothesis 4b, among older workers, job control has little impact on the relationship between job demands and mental health (see Figure 3). Instead, personal control appears to buffer the relationship between job demands and mental health, such that the job demands–mental health relationship becomes less negative as personal control increases, regardless of the level of job control. As hypothesized, these findings suggest that, at younger ages, both job control and personal control moderate the negative relationship between job demands and mental health, with job control buffering the negative relationship at high levels of personal control and exacerbating the negative relationship at low levels of personal control. While at older ages, only personal control moderated the negative relationship between job demands and mental health.¹

Figure 3.

Personal control and job control as a moderator of the job demands–mental health relationship by age.

Discussion

The current study uses the life-span theory of control, to develop hypotheses regarding the extent to which older and younger workers may benefit differentially by job control (which we argue represents primary control at work) and personal control (which we argue represents secondary control) in the context of high job demands, and whether the processes of the DCM change with age. There were several important findings. First, as expected, job demands were negatively related to mental health, regardless of age, suggesting that the negative impact of job demands on mental health is the same for younger workers and older workers.

Second, as predicted, both job control (i.e., having the opportunity for independent thought and to do different things at work) and personal control (i.e., feeling confident in one’s ability to accomplish goals) were positively related to mental health, but only the relationship between job control and mental health varied by age. Specifically, the job control–mental health relationship weakened with age, suggesting that job control plays a greater role in employee well-being at younger ages. An alternative interpretation is possible, however. Since job control was measured as the extent to which employees had the opportunity for job control which is closely related to job type, it is possible that older workers, because they have been in the workforce longer, occupy more desirable jobs that provide greater opportunity for job control. This possibility was supported by our findings in that job control levels were significantly greater for older compared to younger workers. Along these lines, the decrease in the strength of the job control–mental health relationship could be due to older workers’ tendency to have more job control, therefore weakening the impact of different levels of job control on mental health. However, the variability in job control scores was similar for older and younger workers in our sample, suggesting that older workers do not necessarily have a greater opportunity for job control nor does job design/type alone explain the job control-by-age interaction.

Regarding the DCM (Karasek & Theorell, 1990), job control alone did not buffer the negative relationship between job demands and mental health, but personal control did. The negative impact of job demands on mental health was almost completely mitigated at higher levels of personal control, implying that employees with higher personal control may suffer less from high job demands than employees with lower personal control. Although job control did not independently buffer the relationship, it is likely that employees high in personal control who are confident in their ability to control what happens to them select into jobs with the opportunity for control, and so the influence of job control on the job demands–mental health relationship may operate through personal control.

On the basis of our final hypotheses, we tested the extent to which older and younger workers benefit differentially from job control and personal control in the context of high job demands. Indeed, among younger workers, we found that job control buffered the impact of job demands on mental health at high levels of personal control, but job control actually worsened the impact of job demands at low levels of personal control. Our results are consistent with previous findings that the DCM only holds for people who have a strong internal locus of control (Daniels & Guppy, 1994) and that increasing job control may actually be detrimental in employees with low self-efficacy (Salanova et al., 2002). Among older workers, however, a different pattern emerged. While job control played a negligible role in mental health at older ages, personal control served to buffer the job demands–mental health relationship for older workers. Consistent with the life-span theory of control, primary control (conceptualized as job control in this study) may be more important to well-being at younger ages, while secondary control (conceptualized as personal control in this study) becomes more prominent at older ages (Schulz & Heckhausen, 1996). Again, an alternative explanation of this finding could be that, for older workers with high personal control, it is likely that these workers have found their way into jobs with more desirable characteristics, such as greater job control, whereas older workers with low personal control may seek out employment that does involve high job control. Accordingly, the buffering role of personal control on the demands–mental health relationship may be due to older workers with high personal control seeking jobs with more control to negotiate high job demands.

It is unclear whether the findings of the current study were in alignment with the findings of the previous two studies that examined how the DCM varies with age (Perrewe & Anthony, 1990; Shultz et al., 2010). While these studies did not take personal control (as measured here by internal locus of control, generalized self-efficacy, and self-esteem) into account, it could be that some of the previous studies’ measures tap domain-specific aspects of the personal control construct assessed here that are not captured by our measure of job control. We assessed job-specific decision authority and skill discretion in our job control measure, but not job-specific locus of control, self-efficacy or self-esteem (core self-evaluations). The measure of control used by Perrewe and Anthony included items that may be aligned with core self-evaluations, such as “in general, how much influence do you have over work and work-related factors” (p. 90), however their study did not explicitly test for a three-way interaction between demands, control, and age and they had a different outcome variable than mental health, so their findings are not directly comparable to the findings of the current study. Shultz et al.’s schedule flexibility and autonomy measures may tap into these other aspects of job-specific personal control as well. For example, the item “I determine what will happen in my life” from the personal control scale in this study is likely to be highly related to the item “You can take your break when you wish” from the schedule flexibility scale used in Shultz et al. If indeed that is the case, the findings are in line, in that measures of control focused around self-efficacy and/or locus of control were found to be more beneficial to older workers under high demand conditions than younger workers. However, as we found in the present study, the lack of findings among younger workers in the Shultz et al. study could be attributed to a more complex relationship between these various types of control and job demands within this group.

Our study makes important contributions to the literature. First, we inform both workplace theory and aging theory by testing whether the commonly applied DCM operates differently based on age in a relatively large, heterogeneous sample of workers in the United States representing six different industry groups and a broad range of job types. Second, we build on previous work demonstrating the importance of taking personal control into account, in addition to job control, as a factor in the job demands–well-being relationship (Daniels & Guppy, 1994; Rodríguez et al., 2001; Salanova et al., 2002; Schaubroeck & Merritt, 1997). Finally, findings suggest changes to the theoretical underpinnings of the DCM model, and practical suggestions for managing today’s age-diverse workforce.

Implications

Our study has several implications for workplace decision makers and practitioners. First, the DCM model of occupational stress has had a large influence on the job design and occupational health literature, in part, due to the ease at which it translates to job design interventions. In this model, well-being is a function of how demanding a person’s job is and how much control the person has over their own responsibilities. It is thought then, that jobs can be designed to optimize employee well-being by managing the control–demands balance. However, we found evidence that this model operates differently across the life-span for the outcome of mental health, suggesting that the “universal” design approach to human resource management may need to be reconsidered in order to meet the needs of different workers.

Specifically, findings indicate that policies, programs, or practices that promote job control in general can support mental health, however, they also point to circumstances under which high levels of job control may be particularly valuable or harmful for the mental health of workers. While high levels of job control can promote mental health for younger workers with high levels of personal control, it may undermine mental health for younger workers with low levels of personal control. Thus, policies and practices that promote job control by providing opportunities for independent thought and skill discretion within work teams and across the organization can have powerful positive influences on worker well-being, but such policies should be careful to take into account whether the worker is comfortable with such responsibility, particularly among younger workers.

Second, the findings speak to the importance of personal control in mitigating the negative effects of job demands on mental health, regardless of age. This suggests that efforts on the part of managers and business leaders to improve perceptions of personal control through interventions aimed at increasing self-efficacy, self-esteem, and locus of control might be beneficial to employees. Strategies such as developing signature strengths, expressing gratitude, and nurturing social relationships have been found to be effective for enhancing personal control (Bakker, Albrecht, & Leiter, 2011). Other strategies may include identifying and working to mitigate those aspects of the work environment that contribute to workers’ negative beliefs and working to enhance those that contribute to positive beliefs. For example, if an employee is feeling ineffective due to a specific skill deficiency, mangers can help them get the training they need to feel more confident in their abilities. Furthermore, managers could work to identify areas in which employees are strong at work, help to illuminate these strengths, and work with them to build on the possibilities afforded by these strengths. Such interventions may also help younger workers to improve their ability to leverage job control to reduce the impact of job demands.

Finally, this study has broad implications in terms of dispelling myths about older workers. A common misconception among managers, supervisors, coworkers, and perhaps even older adults themselves is that older workers are physically unable to deal with high levels of job demands or are uninterested in taking on a high level of job demands, and thus that they may become more stressed under such conditions than their younger counterparts. Findings from this study indicate that older adults can successfully use control strategies—particularly personal control—to maintain positive mental health in the context of high job demands. Thus, older workers are capable of continued high-demand work, despite age-related losses. Accordingly, age alone should not be a determinant in decisions to recruit or retain older workers. Furthermore, training programs may be effective in helping managers or supervisors to identify practices within their work unit or department that support or engender myths about workers of different ages and to brainstorm strategies to address them.

In sum, policies, programs, and practices aimed at helping employees better manage their work demands should consider how appropriate different interventions are based on individual employees’ needs and preferences. Managers should be careful not to make assumptions based solely on age, however. While age is a proxy for a large variety of factors, including physical health, tenure, life-stage related factors, occupational status, and change in perspective, individuals at any age may rely differently on job or personal control strategies to manage the demands of their work.

Limitations

The Age and Generations Study data includes a large number of employees across several industries, which allows for the examination of a range of experiences at the workplace; however, there are also limitations. First, the data are cross-sectional. For organizational research, survey designs are appealing due to their cost and time efficiency, but these surveys are often conducted at a single time point and do not allow for causal inferences to be drawn. Thus, any age differences in processes discussed here refer strictly to group-level differences between individuals at one age and individuals at another age at one point in time and, as such, it is not possible to disentangle age effects from cohort and period effects. Future research should aim to address the impact of age on the DCM longitudinally to examine how the relationship may vary across time as an employee ages.

Second, generalizability within this sample may be limited. Because organizations were chosen from a convenience sample of organizations, these organizations are likely to be among the more progressive in terms of their talent management strategies and their awareness of issues of the aging workforce. Also, because departments within each organization were not randomly sampled, respondents from each organization are not necessarily representative of their organizations. Given the relatively large number of employees in the study, as well as the variety of organizations and industries represented, however, the analytic findings are likely to be suggestive of trends in similar large organizations.

Third, the operationalization of some of the constructs used in this study may be problematic. For job demands, a measure of work overload was used. While previous research has used overload to assess job demands, our measure of work overload assumes that high scores represent the experience of job stress as opposed to high demands which have the possibility of leading to job stress. Moreover, our work overload measure is likely representative of not only individual differences in the ability to deal with job demands but it is also dependent on the design of work tasks. Our measure of job control, which was used as a proxy for primary control, was also not ideal. This measure asked the extent to which respondents had the opportunity for job control as opposed to actually using job control, meaning that job control is likely related to job type, although multicollinearity was not an issue for our measures of job control and job type. Standard measures of primary control focus on its use as opposed to just access to it. Future research should examine our research questions with measures of job demands that do not assume that stress is experienced and measures of job control that assess the extent of its use.

Another limitation of the current study is that it did not explore the role of support in the DCM, a factor that has been found in some studies to help workers deal with job demands (Johnson & Hall, 1988). Future studies should examine the interaction between personal control and support in buffering against job demands and how it varies with age.

Finally, in the current study, we focused only on mental health, and it is unclear what role personal and job control may play as buffers of the impact of job demands on other outcomes such as job performance and productivity. It is possible that while older workers primarily used personal control to buffer the impact of job demands on mental health, job control may be a more important buffer when examining different outcomes. Future research should address this possibility.

In conclusion, as employers face an increasingly older workforce they must consider how work experiences may differ for employees across the life-span in order to provide a high quality of employment to all workers. Understanding how employees of different ages best deal with their job demands may be one important step toward this end.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The work presented here is funded by the Alfred P. Sloan Foundation (Grant no. 2005-6-18).

Notes

Author Biographies

Elyssa Besen, PhD, is a research scientist at the Liberty Mutual Research Institute for Safety. She earned her doctoral degree in Applied Developmental and Educational Psychology from Boston College. Her research interests focus on the return to work process for workers across the lifespan.

Christina Matz-Costa, MSW, PhD, is an Assistant Professor in the older adults and families concentration of the Graduate School of Social Work at Boston College and a Senior Research Associate at the Sloan Center on Aging & Work. She earned her doctorate at Boston College. Her research interests include the productive engagement of older adults, aging of the workforce, quality of employment, and role transitions in later life.

Jacquelyn B. James, PhD, is the director of research at the Sloan Center on Aging & Work and a research professor at the Lynch School of Education at Boston College. She earned her doctorate at Boston University. Her research has focused on the meaning and experience of work, and adult development and aging.

Marcie Pitt-Catsouphes, PhD, is an Associate Professor in the Graduate School of Social Work, and has appointments at the Boston College Carroll School of Management and the Business School at Middlesex University in London. She is also the director of the Sloan Center on Aging & Work and the co-director of the Center for Social Innovation at Boston College. She earned her doctorate at Boston University. Her research interests include quality of employment, the multi-generational workforce, and work-family issues.

References

Aiken

L. S.

West

S. G.

(1991). Multiple regression: Testing and interpreting interactions. Thousand Oaks, CA: SAGE.

Bakker

A. B.

Albrecht

S. L.

Leiter

M. P.

(2011). Key questions regarding work engagement. European Journal of Work and Organizational Psychology, 20, 4-28.

Bakker

A. B.

Demerouti

Euwema

M. C.

(2005). Job resources buffer the impact of job demands on burnout. Journal of Occupational Health Psychology, 10(2), 170-180. doi:10.1037/1076-8998.10.2.170

Bernal

Snyder

McDaniel

(1998). The age and job satisfaction relationship: Does its shape and strength still evade us? Journals of Gerontology: Series B: Psychological Sciences and Social Sciences, 53, 287-293. doi:10.1093/geronb/53B.5.P287

Birdi

Warr

Oswald

(1995). Age differences in three components of employee well-being. Applied Psychology: An International Review, 44, 345-373. doi:10.1111/j.1464-0597.1995.tb01085.x

Cennamo

Gardner

(2008). Generational differences in work values, outcomes and person-organisation values fit. Journal of Managerial Psychology, 23, 891-906. doi:10.1108/02683940810904385

Clark

Oswald

Warr

(1996). Is job satisfaction U-shaped in age? Journal of Occupational and Organizational Psychology, 69(1), 57-81. doi:10.1111/j.2044-8325.1996.tb00600.x

Cohen

(1978). Partialed products are interactions; partialed powers are curve components. Psychological Bulletin, 85, 858-866. doi:10.1037/0033-2909.85.4.858

Cohen

(1983). Applied multiple regression/correlation analysis for the behavioral sciences. Mahwah, NJ: Lawrence Erlbaum.

10.

Daniels

Guppy

(1994). Occupational stress, social support, job control, and psychological well-being. Human Relations, 47, 1523-1544. doi:10.1177/001872679404701205

11.

Daniels

Harris

(2005). A daily diary study of coping in the context of the job demands-control-support model. Journal of Vocational Behavior, 66, 219-237. doi:10.1016/j.jvb.2004.10.004

12.

de Lange

A. H.

Taris

T. W.

Kompier

M. A. J.

Houtman

I. L. D.

Bongers

P. M.

(2004). The relationships between work characteristics and mental health: Examining normal, reversed and reciprocal relationships in a 4-wave study. Work & Stress, 18(2), 149-166. doi:10.1080/02678370412331270860

13.

Grandey

A. A.

Fisk

G. M.

Steiner

D. D.

(2005). Must “Service with a smile” be stressful? The moderating role of personal control for American and French employees. Journal of Applied Psychology, 90, 893-904. doi:10.1037/0021-9010.90.5.893

14.

Heckhausen

Schulz

(1995). A life-span theory of control. Psychological Review, 102, 284-304. doi:10.1037/0033-295X.102.2.284

15.

Hedge

J. W.

Borman

W. C.

Lammlein

S. E.

(2006). The aging workforce: Realities, myths, and implications for organizations. Washington, DC: American Psychological Association. doi:10.1037/11325-000

16.

Hochwarter

W. A.

Ferris

G. R.

Perrewé

P. L.

Witt

L. A.

Kiewitz

(2001). A note on the nonlinearity of the age–job-satisfaction relationship. Journal of Applied Social Psychology, 31, 1223-1237. doi:10.1111/j.1559-1816.2001.tb02671.x

17.

Hox

J. J.

(2002). Multilevel analysis: Techniques and applications. Mahwah, NJ: Lawrence Erlbaum.

18.

Ilmarinen

J. E.

(2001). Aging workers. Occupational and Environmental Medicine, 58, 546-546. doi:10.1136/oem.58.8.546

19.

Jaccard

Turrisi

(2003). Interaction effects in multiple regression (2nd ed.). Thousand Oaks, CA: SAGE.

20.

James

J. B.

McKechnie

Swanberg

(2011). Predicting employee engagement in an age-diverse retail workforce. Journal of Organizational Behavior, 32(2), 173-196. doi:10.1002/job.681

21.

Johnson

J. V.

Hall

E. M.

(1988). Job strain, work place social support, and cardiovascular disease: A cross-sectional study of a random sample of the Swedish working population. American Journal of Public Health, 78, 1336-1342. doi:10.2105/AJPH.78.10.1336

22.

Judge

T. A.

Bono

J. E.

Erez

Locke

E. A.

(2005). Core self-evaluations and job and life satisfaction: The role of self-concordance and goal attainment. Journal of Applied Psychology, 90, 257-268. doi:10.1037/0021-9010.90.2.257

23.

Judge

T. A.

Erez

Bono

J. E.

Thoresen

C. J.

(2003). The core self-evaluations scale: Development of a measure. Personnel Psychology, 56, 303-331. doi:10.1111/j.1744-6570.2003.tb00152.x

24.

Kanfer

Ackerman

P. L.

(2004). Aging, adult development, and work motivation. Academy of Management Review, 29, 440-458. doi:10.5465/AMR.2004.13670969

25.

Karasek

R. A.

(1979). Job demands, job decision latitude, and mental strain: Implications for job redesign. Administrative Science Quarterly, 24, 285-308. doi:10.2307/2392498

26.

Karasek

R. A.

Theorell

(1990). Healthy work: Stress, productivity, and the reconstruction of working life. New York, NY: Basic Books.

27.

Kreft

de Leeuw

(1998). Introducing multilevel modeling. Thousand Oaks, CA: SAGE.

28.

Kunzmann

Little

Smith

(2002). Perceiving control: A double-edged sword in old age. Journals of Gerontology: Series B: Psychological Sciences and Social Sciences, 57, 484-491. doi:10.1093/geronb/57.6.P484

29.

Lang

F. R.

Heckhausen

(2001). Perceived control over development and subjective well-being: Differential benefits across adulthood. Journal of Personality and Social Psychology, 81, 509-523. doi:10.1037/0022-3514.81.3.509

30.

Loi

J. L. P.

Shultz

K. S.

(2007). Why older adults seek employment: Differing motivations among subgroups. Journal of Applied Gerontology, 26, 274-289. doi:10.1177/0733464807301087

31.

Lyon

Pollard

(1997). Perceptions of the older employee: Is anything really changing. Personnel Review, 26, 245-257. doi:10.1108/00483489710172051

32.

McEvoy

G. M.

Cascio

W. F.

(1989). Cumulative evidence of the relationship between employee age and job performance. Journal of Applied Psychology, 74(1), 11-17. doi:10.1037/0021-9010.74.1.11

33.

Melamed

Kushnir

Meir

E. I.

(1991). Attenuating the impact of job demands: Additive and interactive effects of perceived control and social support. Journal of Vocational Behavior, 39(1), 40-53. doi:10.1016/0001-8791(91)90003-5

34.

Myers

D. G.

Diener

(1995). Who is happy? Psychological Science, 6(1), 10-19. doi:10.1111/j.1467-9280.1995.tb00298.x

35.

T. W. H.

Feldman

D. C.

(2008). The relationship of age to ten dimensions of job performance. Journal of Applied Psychology, 93, 392-423. doi:10.1037/0021-9010.93.2.392

36.

Perrewe

P. L.

Anthony

W. P.

(1990). Stress in a steel pipe mill: The impact of job demands, personal control, and employee age on somatic complaints. Journal of Social Behavior & Personality, 5(3), 77-90.

37.

Peterson

(1999). Personal control and well-being. In Kahneman

Diener

Schwarz

(Eds.), Well being: The foundations of hedonic psychology (pp. 288-301). New York, NY: Russell SAGE.

38.

Peterson

Stunkard

A. J.

(1989). Personal control and health promotion. Social Science & Medicine, 28, 819-828. doi:10.1016/0277-9536(89)90111-1

39.

Peterson

Stunkard

A. J.

(1992). Cognates of personal control: Locus of control, self-efficacy, and explanatory style. Applied and Preventive Psychology, 1(2), 111-117. doi:10.1016/S0962-1849(05)80151-9

40.

Pitt-Catsouphes

Matz-Costa

(2008). The multi-generational workforce: Workplace flexibility and engagement. Community, Work & Family, 11(2), 215-229. doi:10.1080/13668800802021906

41.

Pitt-Catsouphes

Matz-Costa

(2009). Engaging the 21st century multi-generational workforce: Findings from the age & generations study. (Issue Brief No. 20). Chestnut Hill, MA: Sloan Center on Aging and Work at Boston College. Retrieved from http://agingandwork.bc.edu/documents/IB20_Engagement_2009-02-10.pdf

42.

Pitt-Catsouphes

Matz-Costa

Besen

(2009). Age & generations: Understanding experiences at the workplace. (Research Highlight No. 6). Chestnut Hill, MA: Sloan Center on Aging & Work at Boston College. Retrieved from http://agingandwork.bc.edu/documents/RH06_Age&Generations_2009-03-20.pdf

43.

Raudenbush

S. W.

Bryk

A. S.

(2002). Hierarchical linear models: Applications and data analysis methods. London: SAGE.

44.

Rhodes

S. R.

(1983). Age-related differences in work attitudes and behavior: A review and conceptual analysis. Psychological Bulletin, 93, 328-367. doi:10.1037/0033-2909.93.2.328

45.

Rodríguez

Bravo

M. J.

Peiró

J. M.

Schaufeli

W. B.

(2001). The demands-control-support model, locus of control and job dissatisfaction: A longitudinal study. Work & Stress, 15(2), 97-114. doi:10.1080/02678370110066968

46.

Rosen

Jerdee

T. H.

(1976). The nature of job-related age stereotypes. Journal of Applied Psychology, 61(2), 180-183. doi:10.1037/0021-9010.61.2.180

47.

Ross

C. E.

Broh

B. A.

(2000). The roles of self-esteem and the sense of personal control in the academic achievement process. Sociology of Education, 73, 270-284. doi:10.2307/2673234

48.

Royston

(2005). Multiple imputation of missing values: Update of ice. Stata Journal, 5, 527-536.

49.

Rubin

D. B.

(1987). Multiple imputation for nonresponse in surveys. New York, NY: Wiley.

50.

Salanova

Peiró

J. M.

Schaufeli

W. B.

(2002). Self-efficacy specificity and burnout among information technology workers: An extension of the job demand-control model. European Journal of Work and Organizational Psychology, 11(1), 1-25. doi:10.1080/13594320143000735

51.

Sargent

L. D.

Terry

D. J.

(2000). The moderating role of social support in karasek’s job strain model. Work & Stress, 14, 245-261. doi:10.1080/02678370010025568

52.

Schafer

Olsen

(1998). Multiple imputation for multivariate missing-data problems: A data analyst’s perspective. Multivariate Behavioral Research, 33, 545-571. doi:10.1207/s15327906mbr3304_5

53.

Schaubroeck

Merritt

D. E.

(1997). Divergent effects of job control on coping with work stressors: The key role of self-efficacy. Academy of Management Journal, 40, 738-754. doi:10.2307/257061

54.

Schaufeli

W. B.

Bakker

A. B.

Rhenen

W. V.

(2009). How changes in job demands and resources predict burnout, work engagement, and sickness absenteeism. Journal of Organizational Behavior, 30, 893-917. doi:10.1002/job.595

55.

Schaufeli

W. B.

Bakker

A. B.

van der Heijden

F. M. M. A.

Prins

J. T.

(2009). Workaholism among medical residents: It is the combination of working excessively and compulsively that counts. International Journal of Stress Management, 16, 249-272. doi:10.1037/a0017537

56.

Schulz

Heckhausen

(1996). A life span model of successful aging. American Psychologist, 51, 702-714. doi:10.1037/0003-066X.51.7.702

57.

Shultz

K. S.

Adams

G. A.

(2007). Aging and work in the 21st century. Mahwah, NJ: Lawrence Earlbaum.

58.

Shultz

K. S.

Wang

Crimmins

E. M.

Fisher

G. G.

(2010). Age differences in the demand-control model of work stress: An examination of data from 15 European countries. Journal of Applied Gerontology, 29(1), 21-47. doi:10.1177/0733464809334286

59.

Sims

H. P.

Jr. Szilagyi

A. D.

Keller

R. T.

(1976). The measurement of job characteristics. Academy of Management Journal, 19(2), 195-212. doi:10.2307/255772

60.

Skinner

E. A.

(1996). A guide to constructs of control. Journal of Personality and Social Psychology, 71, 549-570. doi:10.1037/0022-3514.71.3.549

61.

Sluiter

J. K.

(2006). High-demand jobs: Age-related diversity in work ability? Applied Ergonomics, 37, 429-440. doi:10.1016/j.apergo.2006.04.007

62.

Smyer

M. A.

Pitt-Catsouphes

(2007). The meanings of work for older workers. Generations, 31(1), 23-30.

63.

Toossi

(2007). Labor force projections to 2016: More workers in their golden years. Monthly Labor Review, 130(11), 33-52.

64.

van Buuren

Boshuizen

H. C.

Knook

D. L.

(1999). Multiple imputation of missing blood pressure covariates in survival analysis. Statistics in Medicine, 18, 681-694. doi:10.1002/(SICI)1097-0258(19990330)18:6<681::AID-SIM71>3.0.CO;2-R

65.

van der Doef

Maes

(1999). The job demand-control (-support) model and psychological well-being: A review of 20 years of empirical research. Work & Stress, 13(2), 87-114. doi:10.1080/026783799296084

66.

von Hippel

P. T.

(2007). Regression with missing Ys: An improved strategy for analyzing multiply imputed data. Sociological Methodology, 37(1), 83-117. doi:10.1111/j.1467-9531.2007.00180.x

67.

Wall

T. D.

Jackson

P. J.

Mullarkey

Parker

S. K.

(1996). The demands-control model of job strain: A more specific test. Journal of Occupational and Organizational Psychology, 69(2), 153-166. doi:10.1111/j.2044-8325.1996.tb00607.x

68.

Wallace

J. E.

(1997). It’s about time: A study of hours worked and work spillover among law firm lawyers. Journal of Vocational Behavior, 50(2), 227-248. doi:10.1006/jvbe.1996.1573

69.

Ware

J. E.

Kosinski

Dewey

J. E.

Gandek

(2001). How to score and interpret single-item health status measures: A manual for users of the SF-8 health survey. Lincoln, RI: Quality Metric.