This article examined the relationship between work dissatisfaction and sleep problems, in Canadian adults age 45 and older, as well as how gender and social contact moderate this association. Problems falling asleep and staying asleep can interfere with a person’s daily life, as these problems create drowsiness and lower energy levels resulting from a lack of quality sleep (Luyster, Strollo, Zee, & Walsh, Reference Luyster, Strollo, Zee and Walsh2012; Moore, Adler, Williams, & Jackson, Reference Moore, Adler, Williams and Jackson2002). In addition, research has shown that sleep problems have negative impacts on health outcomes (Hicken, Lee, Ailshire, Burgard, & Williams, Reference Hicken, Lee, Ailshire, Burgard and Williams2013; Moore et al., Reference Moore, Adler, Williams and Jackson2002), such as cardiovascular problems, diabetes, and obesity (Hicken et al., Reference Hicken, Lee, Ailshire, Burgard and Williams2013; Luyster et al., Reference Luyster, Strollo, Zee and Walsh2012). Sleep problems are especially important to consider in aging populations because research indicates that sleep problems are more prevalent in the latter half of life, and sleep problems tend to increase across the latter half of life (Baiden, Fallon, den Dunnen, & Boateng, Reference Baiden, Fallon, den Dunnen and Boateng2015; Birath & Martin, Reference Birath and Martin2007; Kaufmann et al., Reference Kaufmann, Mojtabai, Hock, Thorpe, Canham, Chen and Spira2016; Luyster et al., Reference Luyster, Strollo, Zee and Walsh2012). Given the increased risk for and prevalence of sleep problems in the latter half of life, it is, therefore, important to examine contributors to sleep problems at this stage of the life course, as well as the resources that can help weaken the deleterious influences on sleep problems.
Work Stress and Sleep Problems
A substantial contributor to sleep problems is the level of stress that an individual endures, because poor sleep can “act as a sensitive marker of the consequences of stressful experiences” (Burgard & Ailshire, Reference Burgard and Ailshire2009, p. 476). Studies show that a common source of stress that influences sleep problems is work (Åkerstedt, Nordin, Alfredsson, Westerholm, & Kecklund, Reference Åkerstedt, Nordin, Alfredsson, Westerholm and Kecklund2012; Burgard & Ailshire, Reference Burgard and Ailshire2009), as demonstrated by research showing that patients who were diagnosed with insomnia often mention their work stressors as the primary causal agent for developing insomnia (Henry, McClellen, Rosenthal, Dedrick, & Gosdin, Reference Henry, McClellen, Rosenthal, Dedrick and Gosdin2008). Work stress can influence sleep because changes in hormonal levels resulting from stress can cause individuals to become more alert and aware, in turn keeping them awake at night and impinging on their sleep quality (Burgard & Ailshire, Reference Burgard and Ailshire2009; Shivpuri, Gallo, Crouse, & Allison, Reference Shivpuri, Gallo, Crouse and Allison2012). A similar process can also occur cognitively, as high levels of work stress can create cognitive arousal that will interfere with sleep patterns (Hall et al., Reference Hall, Buysse, Nowell, Nofzinger, Houck, Reynolds and Kupfer2000; Harvey, Tang, & Browning, Reference Harvey, Tang and Browning2005; Henry et al., Reference Henry, McClellen, Rosenthal, Dedrick and Gosdin2008).
Work dissatisfaction is likely to be a central component of work stress that is associated with sleep problems (e.g., Murata, Yatsuya, Tamakoshi, Otsula, Wada, & Toyoshima, Reference Murata, Yatsuya, Tamakoshi, Otsuka, Wada and Toyoshima2007; Nakata et al., Reference Nakata, Haratani, Takahashi, Kawakami, Arito, Kobayashi and Araki2004). Work dissatisfaction can be a prominent and potent stressor because dissatisfaction is indicative of disappointment and frustration by indicating a disparity between an individual’s set goals and one’s evaluation of his or her present state (Ross & van Willigen, Reference Ross and van Willigen1997). Work dissatisfaction may be especially stressful in the latter half of life because individuals in this stage of life have often established a career trajectory or are actively considering retirement, with the result that dissatisfaction with one’s work conditions may have a sense of irrevocability that was absent earlier in the life course. Although younger adults may also be dissatisfied with their jobs, there is generally a sense of impermanence or eventual opportunity for advancement that comes with entry into the job market which may not be present in the latter half of life. Particularly for individuals in the early stages of life, job dissatisfaction will often be considered synchronous with this stage of the life course, as there is likely to be a sense of “paying one’s dues” that is absent or resented by individuals with a far more extensive work and life history. Therefore, in this research, we examined whether work dissatisfaction is positively associated with the likelihood of experiencing sleep problems among Canadian adults in mid- to late-life.
Despite research suggesting that work dissatisfaction will create sleep problems, a stress process perspective suggests that this relationship will not be unitary. A stress process perspective is a sociologically grounded model of the process through which stress influences individuals, and this perspective suggests that social resources can weaken the negative effects of stress (Pearlin, Reference Pearlin, Aneshensel and Phelan1999; Pearlin & Bierman, Reference Pearlin, Bierman, Aneshensel, Phelan and Bierman2013), with this weakening called “stress buffering” (Wheaton, Reference Wheaton1985). One prominent social resource is the frequency of contact within one’s social network (House, Umberson, & Landis, Reference House, Umberson and Landis1988), and social contact may buffer the relationship between work dissatisfaction and sleep problems for multiple reasons. First, individuals with greater social contact report greater social support (Rote, Hill, & Ellison, Reference Rote, Hill and Ellison2012), and greater social support can buffer the deleterious effects of stress through various mechanisms such as love, caring, or sympathy (Turner & Turner, Reference Turner, Turner, Aneshensel, Phelan and Bierman2013).
Second, social contact can serve a social control function that restricts engaging in poor health behaviours that can harm sleep, such as drinking and unhealthy eating (Procidano, Nausheen, & Gupta, Reference Procidano, Nausheen, Gupta, Buchwald, Ringeisen and Eysenck2008; Umberson, Crosnoe, & Reczek, Reference Umberson, Crosnoe and Reczek2010). Thus, although research suggests that individuals experiencing stress may respond with these behaviours (Grzywacz & Almeida, Reference Grzywacz and Almeida2008; Ng & Jeffery, Reference Ng and Jeffery2003), individuals with greater social contact are less likely to respond to work dissatisfaction with these behaviours. An additional way that social contact can buffer the effects of work dissatisfaction is through allowing an individual to feel devoted and invested in a non-work role (Wiley, Reference Wiley1991). The greater the non-work social connections become, the less salient the working role becomes (Greenhaus & Beutell, Reference Greenhaus and Beutell1985; Stryker & Vryan, Reference Stryker, Vryan, DeLamater and Ward2006; Wiley, Reference Wiley1991), and consequently the stress associated with work will be a less potent force in individuals’ lives. Therefore, in this research, we also examined whether social contact buffers the relationship between work dissatisfaction and sleep problems.
Research has also shown that the importance of social resources for well-being may vary depending on whether the resources are derived from friends or family (e.g., Pettit, Roberts, Lewinsohn, Seeley, & Yaroslavsky, Reference Pettit, Roberts, Lewinsohn, Seeley and Yaroslavsky2011; Shor, Roelfs, & Yogev, Reference Shor, Roelfs and Yogev2013). Specifically, evidence suggests that social resources derived from friends may be more beneficial in buffering the consequences of stress across multiple aspects of well-being than resources derived from one’s family (Nezlek & Allen, Reference Nezlek and Allen2006). Social ties with family may be less efficacious as a stress buffer because, even while serving as a conduit for emotional care and comfort (Ailshire & Burgard, Reference Ailshire and Burgard2012), familial social ties can also create a sense of responsibility for, and obligation to, one’s family (Umberson et al., Reference Umberson, Crosnoe and Reczek2010). Social demands implicit within familial social contacts may counter the benefits of these social contacts, particularly when work dissatisfaction signals that these demands are not being met because employment is not sufficiently providing for one’s family.
Furthermore, family members may experience distress when they see individuals attempting to cope with a problem (e.g., Carpenter, Fowler, Maxwell, & Andersen, Reference Carpenter, Fowler, Maxwell and Andersen2010). The sense that the individual is burdening one’s family may also counter the benefits of support provided by family members. In addition, research suggests that social relationships are most beneficial when they are reciprocal in nature (Ailshire & Burgard, Reference Ailshire and Burgard2012; Takizawa et al., Reference Takizawa, Kondo, Sakihara, Ariisumi, Watanabe and Oyama2006; Thoits, Reference Thoits2011), and social contact with friends is more likely to be an unequivocal and reciprocal relationship, as friendships are often formed from a joint decision (Russell, Bergeman, & Scott, Reference Russell, Bergeman and Scott2012). Therefore, when examining whether social contact buffers the relationship between work dissatisfaction and sleep problems, we also considered whether social contact with friends provides greater buffering than social contact with family.
Gendered Contingencies
A stress process perspective also indicates that each component of the stress process is contingent on core social statuses (Pearlin, Reference Pearlin, Aneshensel and Phelan1999; Pearlin & Bierman, Reference Pearlin, Bierman, Aneshensel, Phelan and Bierman2013). One of the most pivotal of these statuses is gender (Rosenfield & Mouzon, Reference Rosenfield, Mouzon, Aneshensel, Phelan and Bierman2013), suggesting that the relationship between work dissatisfaction and sleep problems, as well as the extent to which social contact will buffer this relationship, are contingent on gender. Work stress exposure may impact sleep problems differently for men and women because work tends to be more salient for men. Historically, gender norms emphasized the role of men as providers for their families, leading men to place greater value on career goal attainment than women, and also to identify more strongly with work accomplishments (Shivpuri et al., Reference Shivpuri, Gallo, Crouse and Allison2012; Stryker & Vryan, Reference Stryker, Vryan, DeLamater and Ward2006; Wiley, Reference Wiley1991). Consequently, undergoing a negative work experience is often more of a concern for men than women (Wiley, Reference Wiley1991). The greater concern about, and investment in, the work role among men leads to a greater sensitivity to work-related stress (Davis, Burleson, & Kruszewski, Reference Davis, Burleson, Kruszewski, Contrada and Baum2011; Wethington, Reference Wethington, Thye and Lawler2000). In this research, we therefore also examined whether the relationship between work dissatisfaction and sleep problems was stronger for men than women.
The extent to which social contacts buffer the relationship between work dissatisfaction and sleep problems may also differ between men and women. Men are socialized to present the appearance of being a “sturdy oak” by conveying an impression of strength and self-reliance whereas women are socialized to value relational interdependence, with the result that women tend to be more relationally oriented than men (Kessler & McLeod, Reference Kessler and McLeod1984; Rosenfield & Mouzon, Reference Rosenfield, Mouzon, Aneshensel, Phelan and Bierman2013). Consequently, women are more likely to utilize social support resources in times of stress while men are more likely to hide their problems and avoid seeking guidance, which means that men will be less likely to gain support for these difficulties through social contact (Addis & Mahalik, Reference Addis and Mahalik2003; Brannon, Reference Brannon2010). Therefore, we also examined whether social contact more strongly buffers the relationship between work dissatisfaction and sleep problems among women than men.
Summary of Aims
In summary, this research had three main aims. The first was to examine whether work dissatisfaction was positively related to the probability of sleep problems among working Canadians in the latter half of life. The second was to examine whether social contact moderated this relationship, and if social contact with friends buffered the relationship more than social contact with family. The third was to examine whether the relationship between work dissatisfaction and sleep problems, as well as the extent to which social contact buffered this relationship, differed between men and women.
Methods
Data
Data for this study derived from Cycle 21 (Family, Social Support and Retirement) of the Canadian General Social Survey (GSS), an annual national survey distributed to collect information on family, social support, and retirement to inform public policy and understand trends occurring in Canada. Cycle 21 was distributed and collected in 2007 with a target population of all persons residing in Canada aged 45 and older. Through qualitative testing, the questionnaire was developed and distributed to the 10 provinces in Canada; an exclusion criterion was for residents of the Yukon, Northwest Territories, and Nunavut, as well as full-time residents of institutions. Cycle 21 gathered data for the 2007 survey through two models. The first model included participants in the previous year’s survey, Cycle 20, who met the age 45 and older requirement. The second model included participants in a new sample that was collected through random digit dialing, with surveys administered through computer-assisted telephone interviewing. Cycle 20 was collected in the same manner. Telephone ownership is 96 per cent across Canada, except for those with the lowest income ($10,000 or less), for which population telephone ownership is 88 per cent. The survey coverage ensures a generally representative sample, and survey estimates are weighted to be representative of all individuals in the target population, including those without telephones.
The overall response rate for Cycle 21 of the GSS was 58 per cent, providing 23,404 respondents. The current study was restricted to respondents who were currently employed part-time or full-time as their main activity because work dissatisfaction is likely to have a substantially different role in the lives of individuals for whom work is not a central activity. The number of currently employed respondents indicating that work was their main activity was 8,228. With listwise deletion removing cases that had at least one missing variable, the analytic sample size became 7,548. A large proportion of this Cycle 21 indicated that work was not their main activity, which can be attributed to the cycle aiming to collect information on retirement. Therefore, individuals heading towards retirement may no longer have classified their main activity as work. Cycle 21 remained an ideal sample as it captured individuals in the latter half of life for whom work was a central activity, and therefore an important part of life. All analyses utilized variance estimation through replicate weights that took the survey’s stratified design into account.
Focal Measures
Sleep Problems
Sleep problems were measured by asking respondents whether they “regularly have trouble going to sleep or staying asleep?” This question has been used repeatedly in the Canadian population as a single-item measure of sleep problems (Sutton, Moldofsky, & Badley, Reference Sutton, Moldofsky and Badley2001), and research supports the validity of a single-item question in measuring disordered sleep (Kupperman et al., Reference Kupperman, Lubeck, Mazonson, Patrick, Stewart, Buesching and Fifer1995). Furthermore, self-ratings of sleep have been shown to be better predictors of mental health outcomes than objective measures of sleep (McCrae et al., Reference McCrae, McNamara, Rowe, Dzierzewski, Dirk, Marsiske and Craggs2008), supporting a focus on influences on self-reported measures of sleep problems.
Work Dissatisfaction
Work Dissatisfaction is a scale, based on five questions, which asked: How satisfied are you with (1) the support received from your supervisor; (2) the types of tasks you perform in your job; (3) the amount of influence that you had in the decisions that affected your job; (4) the opportunities you had for promotion or advancement; and (5) overall, how satisfied are you with your job? Each response ranged from (1) very satisfied to (5) very dissatisfied. A principal components analysis of these items indicated one component with an eigenvalue above 1 that accounted for approximately 55 per cent of the variance in the items. Items had loadings at approximately 0.69 or above on this component, indicating that reports of work dissatisfaction cohered on one measure. Cronbach’s alpha also indicated sufficient reliability (.789). Responses were therefore coded so that higher values indicated greater work dissatisfaction, and the mean of responses to the set of items, as long as the respondent answered three or more items, was used as the indicator of work dissatisfaction.
Social Contact
Frequency of social contact with friends was based on the mean of two questions: The first asked how often the respondent saw friends in person and the second, how often the respondent communicated with friends when not in person, whether by telephone, internet/e-mail, fax, or letter. Responses to both questions were based on response categories of (1) every day; (2) a few times a week; (3) a few times a month; (4) once a month; (5) not in the past month. The measures of social contact with family, for which the General Social Survey uses the term “relatives”, were phrased such that study participants would exclude relatives with whom they currently lived. A principal components analysis of the four items indicated two components with eigenvalues above 1 that accounted for more than 75 per cent of the variance in the items. When these two components were extracted and submitted to a varimax rotation, the friend contact items loaded on the first component above .85 and less than .15 on the second component; conversely, the relatives contact items showed a strong pattern of loadings above .85 on the second component and weak loadings on the first component. Cronbach’s alpha also indicated sufficient reliability for the separate measures of friend contact (.732) and family contact (.675). Responses were therefore coded so that higher values indicated greater frequency of social contact, with separate means based on valid responses to both items used to indicate frequency of contact with friends and with family.
Gender
Gender was measured as a dichotomous variable, with women coded as the higher value.
Control Measures
Work Characteristics
Work characteristics controlled included work hours, type of work, and work schedule. Work hours was controlled using a dichotomous indicator of whether the respondent was working full- or part-time, with part-time being less than 37.5 hours a week worked and full-time 37.5 hours a week and up. Because additional components of job stress, such as job demands, vary in Canada according to occupation (Christie & Barling, Reference Christie and Barling2009), type of work was controlled as a proxy for these additional aspects of work stress, with job classification based on the National Occupational Classification framework. In this framework, management occupations is used as a reference in comparison to a set of nine dichotomous indicators – (a) business, finance, and administrative; (b) natural and applied sciences and related occupations; (c) health; (d) social science, education; (e) art, culture, recreation, and sport; (f) sales and services; (g) trades, transport, and equipment operators and related occupations; (h) occupations unique to primary industry; and (i) occupations unique to processing, manufacturing, and utilities.
In addition, to help rule out reverse causation, in which sleep interference resulting from work led to work dissatisfaction, aspects of work schedule that could lead to or ameliorate sleep inference were controlled using two measures. First, regular daytime schedule was contrasted with (a) a regular evening or night shift and (b) an irregular shift, including a split shift, a compressed work week, and an irregular schedule. Second, schedule flexibility was controlled using a dichotomous variable in which the higher value indicated that the respondent had a choice of the start and end of the workday.
Regional and Cultural Variances
Regional and cultural variances were controlled using measures of geographic location, language of origin, nativity, and race. Region was controlled based on a set of dichotomous indicators – Atlantic region; prairie region; Ontario and British Columbia, with Quebec as a reference group; and an additional dichotomous measure indicating urban residence. In order to account for differences between English- and French-speaking Canadians, respondent’s first childhood language was controlled with a set of measures indicating French or a different non-English language, with English-only as a reference group. Nativity was taken into account with one dichotomous variable indicating if the respondent was born outside of Canada in Europe or North America, and a second identifying if the respondent was born outside of Canada in another region. Race was controlled using a dichotomous measure of Aboriginal person and a second indicating non-Aboriginal visible minority.
Socioeconomic Status
Socioeconomic status was controlled with measures of education, home ownership, and income. Education was based on the categories of obtaining a high school diploma, diploma/certificate from community college or trade/technical school, some university or community college but no degree, and university degree or higher, with less than high school as a reference group. Home ownership was controlled by a dichotomous measure in which “1” indicated that the respondent’s home was not owned by a member of the household. Household income was measured using a set of nine categories, with $100,000 and more as reference; this measure also included a “non-response” category to take missing data into account.
Age
Age of the respondent in the publicly available data set was broken into eight categories, with each category ranging five years starting from age 45, and a top category of age 80 and older; we used this variable as an interval measure.
Family Statuses
Family statuses were controlled using measures of marital status and child in the home. A respondent’s marital status was also controlled based on a series of dichotomous indicators – divorced/separated, widowed, never-married, and common-law – with married as the reference group. The presence of a minor child in the household was controlled with a dichotomous measure in which the higher value indicated a minor was present.
Plan of Analysis
Because of the dichotomous nature of the outcome variable, logistic regression was used with a set of models to examine relationships with trouble falling asleep. Analyses were carried out in a set of stages. In the first stage, we examined the bivariate relationship between work dissatisfaction and the outcome; in the second we examined the adjusted relationship between work dissatisfaction and the measure of sleep problems. The third stage was an interaction analysis. The interactions examined the relationship between work dissatisfaction and sleep problems (1) when gender is a moderating variable, (2) when social contact with friends is a moderating variable, (3) when social contact with family is a moderating variable, and (4) how gender and each measure of social contact intersect to moderate the focal relationship. Steps 1–3 of the interaction analyses were examined using two-term interactions between the measure of work dissatisfaction and each moderator, while the fourth step was tested using three-term interactions between work dissatisfaction, gender, and a measure of social contact. Figures of predicted probabilities illustrate any significant interactions. All analyses were performed using Stata version 13.1 (StataCorp, College Station, TX).
Results
Table 1 displays the weighted descriptive for the measures in the study for the analytic sample. As would be expected from the focus on a working population, the majority of respondents were in midlife, although some were well into late life. This generally supported the aims of our study to examine a representative sample of adults across the latter half of life with work as their main activity. Further, the sample was split approximately evenly between men and women. Individuals at higher levels of income and education were a predominant part of the sample, generally due to the focus on individuals who continued to work in the latter half of life rather than individuals who may have worked at jobs where retirement and a pension after a set time period were more common. Further, over a quarter of the sample reported experiencing sleep problems, indicating the importance and prevalence of sleep problems among individuals who continued to work in the latter half of life.
Percentages for all measures used in analyses
a Percentage of those who answered yes.
b Represents mean of measure.
Table 2 displays the logistic regression models predicting sleep problems. Model 1 examines the bivariate relationship between work dissatisfaction and trouble sleeping. This model shows that the relationship between work dissatisfaction and trouble sleeping is positive and significant (p < .001), indicating that higher levels of work dissatisfaction are associated with greater risk for trouble sleeping.
Effects of work dissatisfaction on trouble falling asleep and staying asleep a
* p < .05. ** p < .01. *** p < .001 (two-tailed tests).
a Unstandardized coefficients are presented (with odds ratio in parentheses). Model 1 includes no control variables, and all subsequent models include the full set of control variables. To highlight the focal analyses, coefficients for control variables are presented in Appendix A.
Model 2 examines the relationship between work dissatisfaction and trouble sleeping while controlling for background characteristics, gender, and social contact with friends and family. An important finding is that Model 2 indicates that the relationship between work dissatisfaction and sleep problems is similar in strength compared to the bivariate model, even when a wide array of control variables is included. The odds ratio for this relationship is essentially the same both with and without these controls included in the model. The similarity in the relationship between work dissatisfaction and sleep problems even with an extensive set of controls demonstrates a relatively robust relationship between work dissatisfaction and sleep problems. Another important finding in Model 2 shows that women have a higher odds of developing sleep problems than men, while friend support is inversely associated with risk of sleep problems. Family support, however, has no associated with sleep problems.
Model 3 tests gender differences in the relationship between work dissatisfaction and trouble falling asleep by including an interaction between work dissatisfaction and gender. The interaction variable is not significant, indicating that the relationship between work dissatisfaction and trouble falling asleep does not vary significantly between men and women. Model 4 includes an interaction between work dissatisfaction and social contact with friends. The interaction is significant, indicating that the relationship between work dissatisfaction and trouble falling asleep or staying asleep differs significantly between the levels of social contact with friends. Furthermore, the coefficient for the interaction in log odds form is negative, indicating that the positive relationship between work dissatisfaction and trouble falling asleep weakens as social contact with friend’s increase
Figure 1 illustrates this interaction by depicting the relationship between work dissatisfaction and trouble falling asleep or staying asleep at high and low levels of contact with friends. High and low levels of both work dissatisfaction and friend contact were determined to be one standard deviation below and above the mean, at approximately the 15th and 85th percentiles, while all control variables were held constant at their respective means. This figure shows that, for individuals with both a low and high level of friend contact, a higher level of work dissatisfaction is associated with a greater probability of trouble falling asleep or staying asleep. However, the difference in the probability of trouble falling asleep or staying asleep between those with low and high levels of work dissatisfaction is greater when individuals have low levels of contact with friends. Social contact with friends, therefore, appears to weaken the relationship between work dissatisfaction and trouble getting to sleep or staying asleep.
Sleep problems associated with work dissatisfaction at different levels of friend contact
This interpretation is supported by an examination of the coefficient for the relationship between work dissatisfaction and trouble falling asleep or staying asleep in ancillary analyses. In these ancillary analyses, this coefficient was tested at low and high levels of social contact. These ancillary analyses showed that, at high levels of friend contact, this coefficient was 0.299 (OR = 1.348; p < .001), and at low levels of friend contact it was 0.567. (OR = 1.763; p < .001). The coefficient for work dissatisfaction is therefore significant at both high and low levels of contact with friends, but, at high levels of friend contact the coefficient is almost halved, indicating that the greater social contact with friends substantially weakens the deleterious relationship between work dissatisfaction and sleep problems.
Model 5 examines the relationship between work dissatisfaction and sleep problems, while including the control variables and an interaction between work dissatisfaction and social contact with family. The inclusion of this interaction variable is not significant, indicating that social contact with family does not significantly moderate the relationship between work dissatisfaction and sleep problems. Together, then, these analyses indicated that social contact with friends but not social contact with family, weakens the adverse relationship between work dissatisfaction and trouble falling asleep or staying asleep.
However, these analyses do not examine whether the buffering effects of social contact differ between men and women. Models 6 and 7 test three-term interactions between work dissatisfaction, gender, and each measure of social contact, which tests gender differences in the extent to which social contact buffers work dissatisfaction. Model 6 includes a three-term interaction between work dissatisfaction, gender, and friend support, while Model 7 includes a three term interaction between work dissatisfaction, gender, and family support. Neither interaction is significant, indicating that the extent to which social contact buffers the relationship between work dissatisfaction and trouble falling asleep or staying asleep does not vary by gender. Thus, social contact from friends buffers the relationship between work dissatisfaction and sleep problems similarly for men and women.
Discussion
The results of this study align with existing literature indicating that stress can lead to sleep problems by showing that, as work dissatisfaction increases, the likelihood of developing sleep problems also increases. The current study also indicates that work dissatisfaction is a particularly powerful predictor of sleep problems, as there is a 10 per cent difference in the risk of having problems falling asleep among individuals with high and low levels of work dissatisfaction. This study is nationally representative of Canadian adults with employment as their main activity in the latter half of life, suggesting that older Canadians who experience work dissatisfaction often have an increased probability of having sleep problems.
Research has also frequently shown that men and women process stress differently (Pearlin & Bierman, Reference Pearlin, Bierman, Aneshensel, Phelan and Bierman2013), and that men tend to place a stronger emphasis on work accomplishments than women (Shivpuri et al., Reference Shivpuri, Gallo, Crouse and Allison2012; Stryker & Vryan, Reference Stryker, Vryan, DeLamater and Ward2006; Wiley, Reference Wiley1991), suggesting that men will be more likely to develop sleep problems stemming from work dissatisfaction. However, this study indicates that the relationship between work dissatisfaction and sleep problems is not significantly weaker for women than men. One explanation for the lack of gender differences in this relationship is that work dissatisfaction is so formidable that it overwhelms gendered sensitivities to work stress. This potency reinforces the idea that educating individuals to manage stress may be vital for preventing sleep problems among individuals in the latter half of life. Another explanation for why gender does not impact the relationship between work dissatisfaction and sleep problems may be seen in the growing gender equality in the workforce (Ali & Akhter, Reference Ali and Akhter2009; Guy & Fenley, Reference Guy and Fenley2014). Over the past years, there has been a large increase in women’s status in the workforce, and particularly in women obtaining high profile and leadership roles that were historically male-dominated (Guy & Fenley, Reference Guy and Fenley2014). With the increase in workforce equality, both men and women may now face comparable stress that stems from dissatisfaction at work.
The stress process perspective also indicates that the effects of stress on well-being may vary by social resources (Pearlin & Bierman, Reference Pearlin, Bierman, Aneshensel, Phelan and Bierman2013), and research indicates that an important social resource is an individual’s frequency of social contact (Berkman, Glass, Brissette, & Seeman, Reference Berkman, Glass, Brissette and Seeman2000; House et al., Reference House, Umberson and Landis1988). This study examined social contact with friends and family separately, and showed that social contact with friends significantly buffered the relationship between work dissatisfaction and trouble falling asleep and staying asleep. This buffering effect may occur because social support that is derived from social contact may help people cope with stress from work dissatisfaction (Rote et al., Reference Rote, Hill and Ellison2012; Turner & Turner, Reference Turner, Turner, Aneshensel, Phelan and Bierman2013). In addition, frequent levels of social contact outside of work allow an individual to be more invested in non-work domains, thereby decreasing the salience of stress experienced in the work role (Greenhaus & Beutell, Reference Greenhaus and Beutell1985; Wiley, Reference Wiley1991).
The buffering effects of social contact do seem limited by type of social relationship, though, because social contact from family did not provide the buffering effects that social contact from friends did. One explanation as to why only social contact with friends significantly moderated the relationship between work dissatisfaction and sleep problems, but family contact did not, is that contact with friends is seen as a reciprocal relationship (Takizawa et al., Reference Takizawa, Kondo, Sakihara, Ariisumi, Watanabe and Oyama2006; Walters et al., Reference Walters, Lenton, French, Eyles, Mayr and Newbold1996). Friendships are formed from a joint decision (Russell et al., Reference Russell, Bergeman and Scott2012), and both sides often partake in receiving and providing support, forming a reciprocal relationship (Russell et al., Reference Russell, Bergeman and Scott2012; Takizawa et al., Reference Takizawa, Kondo, Sakihara, Ariisumi, Watanabe and Oyama2006). On the other hand, family contact can create greater social demands that enhance the stress process (Umberson et al., Reference Umberson, Crosnoe and Reczek2010). As there is more likely to be a lack of reciprocal relationship in family contact than friend contact due to the obligations created by family relationships, social contact with family may often provide less potent buffering compared to social contact with friends.
The current study also examined whether women, more than men, would benefit from the buffering effects of social contact. This expectation was based on evidence showing that women tend to be more expressive of feelings when interacting with friends and also tend to utilize social resources more often than men (Kessler & McLeod, Reference Kessler and McLeod1984; Rosenfield & Mouzon, Reference Rosenfield, Mouzon, Aneshensel, Phelan and Bierman2013). However, analyses showed no significant gender difference in the buffering effects of either source of social contact. This pattern of results therefore suggests that it is the engagement in a non-work role facilitated by social contact that facilitates these buffering effects, rather than the support, which this engagement provides. Overall, though, this set of results underscores the importance of social contact with friends as a buffering resource because it appears to be an equally effective buffer for both men and women.
Limitations
The sample was taken from a national survey; however, the data were not 100 per cent nationally representative because the territories of Canada were excluded. Therefore, being able to generalize the results for all of Canada remains an issue. Furthermore, the sample was cross-sectional. Although controlling for work schedule helped to demonstrate that work dissatisfaction was influencing sleep, rather than that sleep problems were in turn influences of work satisfaction, this control variable cannot take into account all factors that may lead sleep to influence work problems. Additional analyses based on longitudinal data would better support the causal ordering and validity of the relationships observed in this research. Moreover, additional longitudinal analyses would help to ensure that health problems are not confounding the association between work satisfaction and sleep problems.
Social science research generally frames sleep problems as an intervening variable between social conditions and health problems (Jarrin, McGrath, Silverstein, & Drake, Reference Jarrin, McGrath, Silverstein and Drake2013), suggesting that controls for health conditions in this research would bias the analyses by indirectly controlling for sleep problems. Yet, health conditions may also affect sleep problems (Kaufmann et al., Reference Kaufmann, Mojtabai, Hock, Thorpe, Canham, Chen and Spira2016), and additional longitudinal research that holds baseline health statuses constant and examines the association between work satisfaction and sleep problems would help to ensure that poor health is not adversely affecting both work satisfaction and sleep problems. The one-item nature of the outcome variable may also raise reliability concerns, but research has repeatedly used this measure as a valid indicator of insomnia (Hon & Nicol, Reference Hon and Nicol2010; Sutton et al., Reference Sutton, Moldofsky and Badley2001). At the same time, the use of diary measures or other measures of sleep that provide more details on sleep would be useful in identifying the way that work dissatisfaction is related to specific characteristics of sleep problems.
Practical and Intervention Implications
This research suggests that dissatisfying work increases an individual’s likelihood of having trouble falling asleep. Future research should examine the specific factors that lead to work dissatisfaction among adults in the latter half of life, as a means of designing interventions to prevent deleterious influences on sleep. In addition, work dissatisfaction can have such potent deleterious consequences for sleep because the stress caused by work dissatisfaction often leaves an individual alert through cognitive and hormonal stimulation (Burgard & Ailshire, Reference Burgard and Ailshire2009; Hall et al., Reference Hall, Buysse, Nowell, Nofzinger, Houck, Reynolds and Kupfer2000; Harvey et al., Reference Harvey, Tang and Browning2005). The effect of being alert from stress in the work domain often spills over into the home domain and could cause further family conflict (Bolger, DeLongis, Kessler, & Wethington, Reference Bolger, DeLongis, Kessler and Wethington1989; Wethington, Reference Wethington, Thye and Lawler2000). Hence, an additional way to address the potential effects of work dissatisfaction is, rather than changing job conditions, instead to educate individuals on relaxation techniques or ways to manage stress to reduce stress from spilling over into other life domains.
Future Directions and Conclusion
The findings from this study suggest several areas for future research. Beyond replicating this research using longitudinal analyses, research should examine whether there are similarities or differences between men and women regarding the extent to which additional stressors are associated with sleep problems. In addition, future research should more closely examine the potential social contact mechanisms, such as social support and investment in non-work domains, to identify whether these putative mechanisms explain how social contact buffers the stress of work dissatisfaction. Future research which examines the mechanisms responsible for buffering by social contact could also provide a greater understanding as to why some sources of social contact buffer stress more than others. An additional area for future research is that of age differences in the association between work dissatisfaction and sleep problems. Longitudinal research indicating that sleep problems increase throughout the latter half of life suggests that the consequences of work dissatisfaction for sleep problems will be stronger as individuals age. Yet, recent research indicates that birth cohort effects can play a substantial role in the stress process (Bierman, Reference Bierman2014), with the result that cross-sectional analyses which examine age differences without taking these cohort effects into account often confound age and cohort effects (e.g., Mirowsky & Ross, Reference Mirowsky and Ross2008). However, longitudinal methods can be used to differentiate age and birth cohort effects, and an additional important area for longitudinal research is in examining how the association between work stress and sleep problems varies by both age and birth cohort.
In conclusion, research shows that work dissatisfaction is adversely associated with sleep problems among Canadians in the latter half of life, and that there are no significant gender differences in the relationship between work dissatisfaction and sleep problems. However, social contact with friends is important for weakening the relationship between work dissatisfaction and sleep problems. Furthermore, the extent of the buffering effects provided by social contact with friends does not vary by gender. This research, therefore, contributes to the knowledge of sleep problems in the latter half of life by showing that work dissatisfaction is substantially related to increased risk of sleep problems, but that social contact with friends can help to buffer this relationship for both men and women.
