|Home | About | Journals | Submit | Contact Us | Français|
Previous research suggests that discrepancies between work values and rewards are indicators of dissonance that induce change in both to reduce such dissonance over time. The present study elaborates this model to suggest parallels with the first phase of the extension- and-strain curve. Small discrepancies or small increases in extension are presumed to be almost unnoticeable, while increasingly large discrepancies are thought to yield exponentially increasing strain. Work satisfaction is a principal outcome of dissonance; hence, work value-reward discrepancies are predicted to diminish work satisfaction in an exponential fashion. Findings from the work and family literature, however, lead to the prediction that this curvilinear association will be moderated by gender and family roles. Using longitudinal data spanning the third decade of life, the results suggest that intrinsic work value-reward discrepancies, as predicted, are increasingly associated, in a negative curvilinear fashion, with work satisfaction. This pattern, however, differs as a function of gender and family roles. Females who established family roles exhibited the expected pattern while other gender by family status groups did not. The results suggest that gender and family roles moderate the association between intrinsic work value-reward dissonance and satisfaction. In addition, women who remained unmarried and childless exhibited the strongest associations between occupational rewards and satisfaction.
This paper elaborates a model of the work value system, which posits that the discrepancy between work values and work rewards influences the salience of both values and behavior over time (Porfeli & Vondracek, 2007). In this paper, the relationship between intrinsic work value-reward discrepancies and work satisfaction is tested using longitudinal data at two occasions spanning the third decade of life. During this period of life, families are typically formed and careers are established. Intrinsic work value-reward discrepancies are presumed to become increasingly detectable and aversive in a curvilinear manner; hence, the association between these discrepancies and work satisfaction are predicted to be curvilinear as well. Once the nature of the relationship between intrinsic work value-reward discrepancy and satisfaction is established, the potential moderating influence of the onset of family roles on this relationship is tested. Specifically, literature suggests that males and those who are married may perceive work as being more of a duty that fulfills the needs and desires of others than as an entitlement that fulfills personal needs and desires (MOW International Research Team, 1987). Gender and the establishment of family roles may, therefore, moderate the relationship between intrinsic value-reward discrepancies and work satisfaction.
The difference between what people have and want has been a topic of ongoing study inside and outside the field of vocational psychology for many years (for a review of social psychological literature, see Mortimer & Lorence, 1995). Within the field of vocational psychology, the Theory of Work Adjustment (Dawis & Lofquist, 1984) is a primary foundation of such investigation. Extensive research, typically framed within the broad concept of person-environment fit, finds that a better fit, or a smaller discrepancy between personal work orientations (e.g., interests and values) and job characteristics (e.g., job rewards and demands), is predictive of subjective satisfaction and satisfactoriness (Bizot & Goldman, 1993; Breiden, Mohr, & Mirza, 2006; Bretz & Judge, 1994; Chiocchio & Frigon, 2006; Feij, van der Velde, Taris, & Taris, 1999; Hesketh, McLachlan, & Gardner, 1992; Rounds, Dawis, & Lofquist, 1987), job tenure (Bretz & Judge, 1994), and turnover intentions (Lyons & O'Brien, 2006) with some variation across subpopulations (e.g., Chiocchio & Frigon, 2006; Lyons & O'Brien, 2006; Melchiori & Church, 1997).
During the process of vocational exploration, individuals discover their own interests and abilities and their work values and preferences (Patton & Porfeli, 2007; Zimmer-Gembeck & Mortimer, 2006). As adolescents make the transition to adulthood and acquire more intensive and stable employment, their growing understanding of what they value in work, coupled with their actual work experiences, presumably promote an increasing coordination (or association) between values and rewards (see Porfeli and Vondracek (2007) for an elaboration of this logic and a review of the pertinent literature). This process of becoming aware of one's values and coordinating them with behaviors and rewards may be becoming more extended in recent cohorts. As education is prolonged, individuals take longer to find a good match between their work orientations and their jobs, and to settle into stable full-time work. The extension of this highly formative period supports the utility of further study of person-environment discrepancies longitudinally, during the transition to adulthood, within the field of vocational psychology.
In a closely related field, I/O Psychology generally finds that discrepancy or congruence, variously defined, is associated with job satisfaction, organizational tenure and commitment, and the intent to leave an organization (Ostroff & Judge, 2007; Verquer, Beehr, & Wagner, 2003). Argyris (1957) has been credited with bringing the concept of person-environment fit to I/O psychology (Verquer et al), but clearly such a model was beginning to take shape in Munsterberg's work several decades earlier (Porfeli, 2008). Pertaining specifically to values, the fit (or lack of discrepancy) between personal and organizational values is associated with work satisfaction and tenure within an organization, and such value congruence tends to be a stronger predictor than other indicators of person-organization fit (Verquer et al).
Outside the field of vocational psychology, Multiple Discrepancies Theory (MDT; Michalos, 1985, 1991) is among the most prominent in the study of life satisfaction (Beckie & Hayduk, 1997; Cohen, 2000; Jacob & Brinkerhoff, 1999; Schulz, 1995). Michalos (1991) cites an extensive body of empirical research and theory from several literatures, including mental health, marital relationships, values, subjective well-being, education, economics, friendships, person-environment fit, and race- and sex-based equity studies to support MDT. This complex theory of discrepancies asserts that general and domain-specific life satisfaction is principally predicted by a host of perceived discrepancies with the self being the primary referent. These perceived discrepancies include those between the self and others, present and past self, present self and expected self, and between present self and what one believes s/he deserves or needs. The discrepancy between what one has and what one wants (e.g., work values and rewards) in the present moment is believed to have the greatest direct influence on satisfaction and behavior, and to exhibit a negative linear relationship with satisfaction. This study is a test of this assertion.
The discrepancy between a work value and reward is believed to produce dissonance or strain (Festinger, 1957). In this study, such discrepancy is conceptualized as an indicator of dissonance that becomes increasingly detectable and aversive. Small discrepancies between intrinsic values and rewards will yield no detectable dissonance (i.e., no relationship between the two), while larger discrepancies are predicted to be increasingly detectable and aversive (i.e., quadratic relationship between the two). Altogether, these predictions suggest that a quadratic function best explains the relationship between value-reward discrepancies and work satisfaction. This model, derived in part from a control theory of human behavior, is supported in empirical research (Kluger, Lewinsohn, & Aiello, 1994).
This predicted relationship can be illustrated with a common rubber band. The discrepancy between a value and obtained reward is akin to the extension of a rubber band and dissonance is indicated by the strain in the band. Small discrepancies or band deflections yield almost unnoticeable dissonance or strain, while larger discrepancies or deflections become increasingly noticeable and discomforting and are, therefore, predicted to yield exponential increases in dissonance or strain. When the presumed relationship is conceived in terms of discrepancies moving in positive (rewards exceeding values) and negative (values exceeding rewards) directions, the relationship conforms to a quadratic, u-shaped relationship. Presuming that dissonance and work satisfaction are inversely associated, the relationship between work value-reward discrepancies and work satisfaction is hypothesized to conform to an n-shaped (negative quadratic) function.
The two forms of dissonance may not be experienced in an identical fashion. Workers who obtain less than they want (i.e., a work reward deficit) are predicted to experience more dissonance than those who obtain more than they want (i.e., a work reward surplus). A work-value reward deficit is presumed to be more aversive than a work-value reward surplus, but a surplus may also be aversive depending on the value in question. As a consequence, the peak in satisfaction is generally predicted to be at the point where values equal rewards, but satisfaction may also peak when rewards slightly exceed values, or when a slight surplus condition exists.
We expect that age would moderate the effects of dissonance on satisfaction during the period of transition to adulthood, given that the decade of the twenties is a time of career exploration and establishment. As they gain work experience, young people may become more aware of what their values actually are, as they try out different jobs and discover the kinds of work tasks and conditions of employment that are most suited to their interests and capacities (Mortimer & Zimmer-Gembeck, 2007). This crystallization of work values is accompanied by increasing labor force attachment and stability. Youth often start off with “survival jobs” while they are still attending school, “flounder” between various short-term or other temporary jobs as they seek a better fit between their work orientations and the rewards that jobs have to offer, and eventually settle into “career jobs” (Huiras, Uggen, & McMorris, 2000). Expectations about the duration or permanence of these various work roles would likely moderate the effects of dissonance on job satisfaction. An individual would probably not expect a job that is pursued “just for the money” to satisfy personal values, and dissonance between values and rewards would likely not be so problematic. By the end of the twenties, however, work is likely to be more permanent and career-like. Discrepancies between occupational values and rewards at this age would likely be much more salient and worrisome, diminishing job satisfaction.
Gender and family roles may likewise influence these dissonance-producing processes because these roles shape the meaning of work in people's lives. This line of thinking is consistent with theoretical and empirical efforts suggesting that combinations of gender and family roles moderate the link between personal and work characteristics (Voydanoff, 2002). Accordingly, we suggest that gender and family roles also influence the association between intrinsic work value-reward dissonance and work satisfaction. This study is responsive to calls to further examine how family affects functioning in the workplace (Eby, Casper, Lockwood, Bordeaux, & Brinley, 2005).
A seminal international study of the meaning of work found several gender differences in the endorsement of work norms featuring entitlement versus obligation (MOW International Research Team, 1987). The entitlement norm “represents the underlying work rights of individuals and the work-related responsibility of organizations and society toward individuals,” while the obligation norm “represents the underlying duties of all individuals to society with respect to working” (MOW International Research Team, 1987). Thus, according to the entitlement norm, members of a society have the right to prepare for and enter into personally meaningful and gratifying work; the norm of obligation stresses that people should value and feel obliged to work as part of their role as a member of society. In comparison to females, males in the MOW study tended to view work as more of an obligation and a duty rather than an entitlement. Females tended to see work as less central, to endorse obligation less and entitlement more, and to conceptualize work more in terms of its social aspects. Other studies document that women have stronger intrinsic occupational reward values than men (Monica Kirkpatrick Johnson & Mortimer, 2000; Marini, Fan, Finley, & Beutel, 1996; Rowe & Snizek, 1995; Tolbert & Moen, 1998). There is thus reason to expect that value-reward discrepancies will have distinct effects on job satisfaction depending on gender and family status. Reviews of the literature (Eby, et al., 2005; Jacobs & Meltzer, 1988) also suggest that females more often than men place work second to family demands, view events in work and family contexts as influencing one another, assume more family responsibilities, sustain more career interruptions, and alter their work lives to accommodate their family demands.
When asked to discuss their work lives, women more frequently speak about their family and less frequently about their work than do men (Montgomery, Panagopoulou, Peeters, & Schaufeli, 2005). Men with traditional attitudes toward work and family experience more guilt when family interferes with work (Livingston & Judge, 2008), and traditional women typically diminish work rather than family demands to cope when family interferes with work (Somech & Drach-Zahavy, 2007). Research on work-family conflict finds that people who have more and younger children experience more work-family conflict, and mothers experience more work-family conflict than fathers (Byron, 2005; Eby, et al., 2005). These findings suggest that males and females ascribe different meanings about and take different behavioral approaches to work that may contribute to differences in how personal work value-reward discrepancies relate to work satisfaction. Given that males tend to view work as being more of an obligation and a duty, discrepancies between their personal intrinsic work values and rewards are predicted to have a weaker influence on their work satisfaction than for females who tend to cast work more in entitlement terms.
As family roles are assumed, the primary referent in assessing the meaning of work may shift from an individualistic frame of reference to one oriented to the new collectivity. What was once a primary mode of satisfying one's own personal intrinsic needs and desires becomes a means to support, or interfere with, familial needs. Consistent with this expectation, there is evidence from the Youth Development Study that marriage fosters declines in intrinsic reward values among both men and women (Monica Kirkpatrick Johnson, 2005). The aforementioned international study of the meaning of work also found that married adults downplayed the social aspects of work, and less frequently defined work as an entitlement than their unmarried age mates (MOW International Research Team, 1987). Other research finds that increased perceived family responsibilities consistently predict an increased commitment toward work (Eby, et al., 2005). In a similar vein, the relationship between perceived social pressure and job seeking behavior is stronger for people with marital and parenting roles (Van Hooft, Born, Taris, & van der Flier, 2005). As adults establish family roles and feel the pressures associated with these roles, their work satisfaction may increasingly depend on the extent to which their work can support a family rather than the satisfaction of their personal values. As a result, extrinsic work values increase in importance as men become fathers and as unmarried women, who must shoulder the burden of child support, become mothers (Monica Kirkpatrick Johnson, 2005). Family formation, may, therefore, diminish the association between intrinsic work value-reward discrepancies and work satisfaction.
Prior studies of work and family strongly indicate that gender moderates the effects of family roles. Specifically, combinations of gender and family status may lead to different associations between intrinsic work value-reward discrepancies and satisfaction. Males who marry and/or become a father are predicted to exhibit the weakest associations between intrinsic work value-reward discrepancies and satisfaction, given that this group is predisposed to assume the most obligatory orientation toward work. Females who do not establish family roles are predicted to exhibit the strongest associations between intrinsic work value-reward discrepancies and satisfaction given that this group is predisposed to maintain the most self-gratifying view of work.
Ongoing debate surrounds the most appropriate way to model discrepancies (Cattell, 1983; Cronbach & Furby, 1970; Williams, Zimmerman, & Mazzagatti, 1987). While most agree that simple differences are problematic from a reliability perspective and that residualized indicators offer a slight improvement, no clear consensus exists on how to mathematically estimate discrepancies between two or more variables. As a result, both simple and residual differences are employed in the present study and the results will be compared.
Ongoing research suggests that discrepancies may also be modeled with polynomial regression terms and this approach may have advantages over both simple and residualized discrepancy estimates (Edwards, 1994; Edwards & Cable, 2009; Edwards & Parry, 1993). Combining polynomial regression terms with a response surface interpretative framework permits a more refined depiction of the influence of discrepancies on an outcome. Figure 1 depicts a hypothetical relationship between work value-reward discrepancy and satisfaction. The top frame is a response surface defined by a polynomial regression model including reward, reward2, value, value2, and reward X value. This response surface indicates that increasing discrepancies between a value and reward in a surplus (reward exceeding value) or deficit (value exceeding reward) direction predict precipitous declines in satisfaction (i.e., the dissonance axis). The top frame also suggests that satisfaction is greatest when a value and reward are equal and this peak remains constant regardless of the level of the value and reward (i.e., the consonance axis). The bottom frame is a contour map (akin to a topographic map) of the same response surface. If one could stand above the top frame and look downward at the response surface it would appear like the contour map in the bottom frame. Lighter colors represent higher points on the response surface, indicating greater satisfaction. These two frames indicate that satisfaction will be greatest when a value and reward are consonant; satisfaction declines as a value and reward become increasing dissonant in both surplus and deficit directions.
Longitudinal data including two occasions of measurement and spanning the third decade of life were drawn from the Youth Development Study (Mortimer, 2003). The participants were approximately 21–22 (1995) and 28–29 (2002) years old at the two occasions, which reflects seven years between the occasions. We decided to focus on this age span to lessen the likelihood that respondents would still be in school and to increase the likelihood that we would capture the shift from less to a more stable attachment to the labor force.
A randomly selected sample of 9th grade students was originally drawn from the St. Paul (Minnesota, USA) School District in the fall of 1987, and 1,000 respondents completed surveys in the spring of 1988. Fifty-two percent of the sample is female and 74% is Caucasian. The panel has been surveyed nearly annually since 1988, with a retention rate of 72 percent through the 2002 data collection (age 28–29). The initial sample (N=1,000) well represented adolescents enrolled in the St. Paul Public Schools at the beginning of the study (e.g., Mortimer, Finch, Shanahan, & Ryu, 1992), although attrition has been greater among men, minorities, and respondents whose parents were not employed at the outset (Mortimer 2003; Staff and Mortimer, 2007). Comparing the St. Paul community to the nation as a whole at the start of the study, St. Paul was a slightly less affluent community and had a smaller minority population. These differences, as well as panel attrition, may limit the generalizability of the results of this study.
Only a subsample of participants who submitted complete data was included in this study. The number of people who participated at the first and second occasions was 780 and 721 respectively. Varying numbers of participants submitted complete work value, reward, and satisfaction data, as well as family role information, across the two occasions (1st occasion (N = 587); 2nd occasion (N = 585)); 378 participants (217 females and 161 males) submitted complete data for both occasions. A meaningful fraction of the difference between survey participation and the provision of complete data is attributed to the 276 participants who were not employed at one or both occasions. As a consequence, these participants were not asked to respond to questions about work rewards. In order to examine shifts in relationships between value-reward discrepancies and work satisfaction over time as family roles are assumed, this study included the fraction of participants who submitted complete data and who had not established family roles by the first occasion (N = 300). Thus, at the first occasion, all included participants had not married or become a parent.
To assess differences in socioeconomic characteristics between the survey participants at ages 21–22 and 28–29 who were part of our analytic sample and those who were not (because of lack of employment, the failure to answer key questions, or establishing family roles before the designated period of the study), we divided participants into two groups: those who met the inclusion criteria outlined above and those who did not. In general, excluded participants were from a lower socioeconomic status as demonstrated by their parents' lower income (t(727) = −3.47, p < .01), and the participants' lower hourly pay (t(696) = −7.30, p < .01) and fewer hours of work per week (t(645) = −9.16, p < .01) as reported during the second occasion of this study. Participants who submitted complete data also obtained more education than those who did not (χ2 (7, N = 716) = 61.24, p < .01). The means and frequencies from these analyses suggest that the typical person included in this study earned $18.61/hour (SD = 8.73) and worked 38.9 hours/week (SD = 10.1); by the second occasion 74.3% had obtained at least an Associate's degree and 42.3% had received at least a Bachelor's degree. The typical participant not included in this study by virtue of incomplete data earned $12.53/hour (SD = 12.15) and worked 28.5 hours/week (SD = 19.5); by the second occasion 52.4% had obtained at least an associate's degree and 19% had received at least a Bachelor's degree. These differences are consistent with the more unstable work histories of the non-participants and suggest that the generalizability of these results is limited to stably employed people with demographic characteristics similar to those of participants in this study.
The family role variable is an indicator of the transition to marital and/or parental roles between the first and second occasions. Based on their marital and parental statuses, participants were divided into two groups. The first group never reported being married or having children (N = 138). The second group reported a marriage and/or having a child between the ages of 21–22 and 28–29 and never divorced (N = 162). The respondent's status, as continuously unmarried and not a parent, or continuously married, was confirmed by examination of annual data obtained between these two waves. Respondents who had a child at any time between waves were presumed to be continuously in a parent role. We explored the possibility of distinguishing the effects of marriage and parenting roles, but the limited sample size precluded such an analysis. The final groups included males who acquired family roles (N = 68), males without family roles (N = 67), females who acquired family roles (N = 94), and females without family roles (N = 71) by 2002.
It should be emphasized that participants were placed into the two family role groups based upon their status between two waves of observation. One group never married or had children through their late twenties. The second group either married and/or had children during their twenties. If the establishment of family roles were permitted to vary by the occasion of measurement, then over time we would see large fractions of people migrating from the no family roles group to the family roles group. Comparisons of the models for these two groups over time would be confounded by persons in the no family roles group at an earlier occasion becoming members of the family roles group at a later occasion. If the predictive power of the models diverged, then this could be attributed to the acquisition of family roles or to the migration of people with distinct characteristics from one group to the next, or both.
The idea of including a timing component was also explored (e.g., established a family role before versus after the first occasion predicting differences in model fit at the second occasion). However, the statistical power of the family role X gender X timing groups was insufficient in certain subgroups (i.e., the sample sizes were too small, given the other bases of exclusion) and particularly for those groups who established family roles before the first occasion (N = 78) versus after (N = 300).
Family status subgroups were assessed to determine if they differed on a range of socioeconomic indicators (i.e., participants' parents' educational attainment and occupational prestige or the participants' rate of pay, job prestige, hours worked per week, or educational attainment at the second occasion of measurement). The results suggest that the participants are more alike than different. The participants differed in their hourly rate of pay with those with family roles earning $19.67 per hour as compared to $17.37 per hour for those who did not establish family roles by 28–29 years old (t(289) = −2.26, p < .01). The only other difference suggests that participants without family roles have parents with higher educational attainment than those who establish family roles (t(246.85) = 2.14, p < .01). In other words, there was some indication that family of origin and participants' income are associated with family formation, but the two groups only exhibited differences in two of the six variables tested and those differences were small.
Intrinsic work values were assessed at each occasion of measurement with four items. The work value items were preceded by the following prompt, “When you are looking for work, how important are the following things to you?” Intrinsic work value items included “A chance to make my own decisions at work,” “A job where I have a lot of responsibility,” “A job that uses my skills and abilities,” and “A chance to learn a lot of new things at work.” The response set ranged from 1 (“not at all important”) to 4 (“extremely important”). The items for this scale were summed and then standardized. Intrinsic value scales have been found to exhibit acceptable psychometric characteristics in other data sets (Monica Kirkpatrick Johnson, Mortimer, Lee, & Stern, 2007). In the present study, internal consistency reliabilities of the scale for the two occasions of measurement were .78 and .72.
Intrinsic work rewards were assessed at each occasion of measurement with five items. The items assessed various intrinsic aspects of work including the freedom to think of new ways of doing work tasks or solving problems, the chance to use one's skills and abilities, the chance to learn new things, and feeling that one's work is meaningful and important. The response set varied depending on the item with 2 items employing a response set from 1 (“never”) to 5 (“almost always”), 2 items employing a response set from 1 (“not at all true”) to 4 (“very true”), and the final item ranging from 1 (“not at all challenging”) to 4 (“very challenging”). The items for this scale were summed, and then standardized. Internal consistency reliabilities of the scale for the two waves were .81 and .81 respectively.
Given ongoing debate in the literature concerning the estimation of discrepancy scores (Cattell, 1983; Cronbach & Furby, 1970; Williams, et al., 1987), simple, residualized, and polynomial indictors of discrepancies are employed and compared in this study. Simple intrinsic work value-reward discrepancies were computed by subtracting intrinsic work values from intrinsic rewards (rewards minus values) at each occasion. Both aspects of the discrepancy (values and rewards) were standardized; hence, the result is a discrepancy variable in which a negative number reflects a reward deficit relative to values and a positive number reflects a surplus. The residualized difference score was created by employing OLS regression to predict rewards from values at each occasion. The residuals that spring from this regression model indicate the difference between actual and predicted rewards based upon one's work value. Those people who obtain more work rewards than their work values would receive positive numbers, reflecting a reward surplus; those who obtain fewer work rewards than their work values receive negative numbers, reflecting a reward deficit. Finally, a polynomial approach (Edwards, 1994; Edwards & Cable, 2009; Edwards & Parry, 1993) was employed to test the presumed quadratic effect of work value-reward discrepancies on work satisfaction. The dissonance-satisfaction response surface is defined by a polynomial regression model including reward, reward2, value, value2, and reward X value predicting work satisfaction.
Work satisfaction was assessed with one item, “All things considered, how satisfied are you with your job as a whole?” Participants responded on a scale from 1 (“extremely dissatisfied”) to 6 (“extremely satisfied”). The means for the panel across the occasions were fairly stable at 4.3 and 4.5 and skewness was within acceptable limits. A preliminary analysis employing ANOVA found only one statistically significant difference when work satisfaction was examined by gender, family status, or gender X family status groups across the three occasions of measurement. Those who established family roles exhibited more work satisfaction (M = 4.66) at the second occasion, age 28–29, than did those who had not (M = 4.40) (F (1, 300) = 5.20, p < 0.05), but the effect size was quite small (eta2 = .02).
The first hypothesis is that the relationship between work value-reward dissonance and satisfaction assumes a negative quadratic relationship. As a work value-reward discrepancy increases in the positive or negative direction, it is predicted to yield an exponential increase in dissonance. Along the consonance axis (see Figure 1) where values equal rewards, satisfaction is predicted to be highest regardless of the level of the value and reward. The second hypothesis asserts that the association between work value-reward discrepancies and work satisfaction varies as a function of sex and family roles, with males who established family roles and females who did not exhibiting the weakest and strongest associations, respectively, over time.
Regression models were computed for the entire sample, and then by gender, family role status, and gender by family role status to test the possible moderating effects of these groups. A few notes are in order before proceeding to the statistics. First, a multivariate outlier was identified and removed given the known sensitivity of curvilinear regression models to outliers. Second, work satisfaction at each occasion was predicted with 3 different concurrent estimates of work value-reward discrepancies (i.e., simple, residual and response surface approaches). Given the limitations of simple and residualized value-reward discrepancy scores (Cattell, 1983; Cronbach & Furby, 1970), a response surface approach (Edwards & Cable, 2009) was also employed. Third, both linear and quadratic coefficients were included in all simple and residualized value-reward discrepancy models, regardless of the significance of the betas, to assure comparability of the r-square estimates across models. If certain models contained the quadratic term and others did not, then the r-square estimates would not be directly comparable. Similarly, the response surfaces and contour maps included non-significant betas because the response surfaces cannot be drawn appropriately without them (i.e., the shape artificially changes if one is omitted).
The findings are shown in Tables 1 (simple difference predictors), predictors),22 (residualized predictors) and and33 (response surface). Overall, the results generally suggest that, for the entire sample, work value-reward discrepancies exhibit an increasing association with work satisfaction across the third decade of life. Employing simple difference scores, the association between work value-reward discrepancies and satisfaction is quadratic in the late 20's, but not in the early 20's. Focusing on the first column for each occasion in Table 1, the squared difference is not statistically significant at 20–21 years old but is significant at 28–29 years old. The simple difference regression equations predict that the peak in satisfaction occurs on the surplus side of the discrepancy when the participants were 28–29 years old (i.e., the regression equation predicts that a difference of 1.5 will yield a peak in satisfaction of 4.89). The curvature of the regression line suggests that increasing work value-reward deficits translate into an accelerating decline in work satisfaction and increasing surpluses are associated with declining satisfaction.
Employing residualized discrepancy scores (first column of Table 2), the relationship between the value-reward discrepancy and satisfaction is linear during the early twenties, but by the late 20's, the results point to a quadratic relationship, as indicated by the statistically significant squared term. Peak satisfaction occurs on the surplus side of the discrepancy between values and rewards when the participants were 28–29 years old (i.e., the regression equation predicts that a residual difference of 2.8 will yield a peak in satisfaction of 5.30). Both the simple and residual difference functions suggest that increasing discrepancies are associates with satisfaction in a negative quadratic fashion, but only during the late 20's.
The response surface approach yields some similarities and important differences. Like the previous models, the r-square estimates generally increase from age 21–22 to age 28–29. Like the residualized discrepancy score model, only linear effects were statistically significant during the early 20's (see the first column of statistics in Table 3), but by late 20's the beta weights tied to quadratic change and the value-reward discrepancy have a statistically significant effect. Unlike the previous two models, the response surface model has the capacity to distinguish the relative predictive power of rewards and values. Intrinsic work rewards are the only consistent predictors of satisfaction during the early 20's and this effect was moderate to strong (see the first column in Table 3). During the second occasion two differences emerge. The quadratic reward term for rewards becomes statistically significant as does the reward by value interaction.
Using the results from the response surface model, we plotted the response surface for all participants (see Figure 2). Like the simple and residualized difference model, a peak in satisfaction occurs mainly on the surplus side of the dissonance axis, but the response surface model reveals more detail about this phenomenon. The response surface (first panel) and contour maps (second panel) for all participants in Figure 2 demonstrates the peak in satisfaction deviates from the consonance axis and toward dissonance on the surplus side of the map. Satisfaction is highest when the work reward and value are high and fairly equal (top left portion of the consonance axis), with a tendency toward satisfaction remaining higher when rewards exceed values (i.e., a surplus condition). At lower levels of values and rewards, satisfaction remains slightly higher when a surplus condition exists than when a deficit condition exists. This latter phenomenon is well illustrated by the shift from the highest to the second highest category of satisfaction within the contour map (second lightest color, satisfaction = 4 to 6) as one moves from the top left to the bottom right hand corner of the map along the consonance axis. In sum, this contour map suggests that people are generally more satisfied with work value-reward discrepancies when the discrepancy reflects a surplus condition (i.e., rewards exceed values) as compared to discrepancies of equal magnitude that reflect a deficit condition (i.e., values exceed rewards). When a consonance condition exists, those people with more intrinsic rewards and greater intrinsic values are more satisfied than are those with lower, but still consonant levels of both.
The trends by sub-groups suggest that the relationship between work value-reward discrepancies and satisfaction varies by gender, family role status, and configurations of both (see the columns labeled accordingly in Tables 1–3). As a note of caution, this subdivision substantially reduces the sample sizes within each group, limiting statistical power.
The results from the simple difference and residualized gain models (see Tables 1 and and2)2) suggest that, in their early 20's, the work satisfaction of females, those males and females (combined) who would establish family roles during their 20's, and females who would establish family roles were more responsive to intrinsic value-reward discrepancies relative to their comparison groups. The lack of statistically significant quadratic terms during the early 20's in Tables 1 and and22 suggests that the association between work value-reward discrepancies and work satisfaction is linear for the subgroups with the only exception being those who established family roles. In this latter case, the negative quadratic relationship is consistent with the hypothesized function. In other words, discrepancies seemingly are associated with work satisfaction, but during the early 20's, the association is not consistent with the extension and strain curve model, which is made manifest as a negative quadratic relationship between work value-reward discrepancies and work satisfaction.
The results from Tables 1 and and22 suggest some stability and change in model fit across the third decade of life and between the subgroups. By the late 20's, females continued to exhibit better model fit than males, but now the relationship between work satisfaction and the discrepancy was stronger for those participants who did not establish family roles relative to those who did. The greatest differences in model fit were exhibited in the gender by family role groups. The simple and residualized difference models show that females without family roles exhibited the largest r-square at 28–29 years old and the betas suggest a linear association between the discrepancy and work satisfaction. Females with family roles and males without family roles exhibited the hypothesized quadratic relationship during their later 20's, but this was only true for the simple difference models.
The sum of the results from the simple and residualized difference regression models across subgroups and time suggest that females continued to show a stronger relationship between intrinsic work value-reward discrepancies and work satisfaction, but a shift occurred with respect to family roles. At age 21–22, participants who would be establishing family roles between the first and second occasion exhibited a stronger relationship than those who would remain unmarried and childless. But by age 28–29, those that did not establish family roles exhibited a stronger relationship than those who had begun family formation.
The work satisfaction of females without family roles became more responsive to intrinsic work value-reward discrepancies during the late 20's but they did not exhibit the predicted quadratic relationship. Females with family roles and males without family roles exhibited the predicted quadratic relationship in the simple difference models (Table 1), but this pattern was not supported in the residualized difference models (Table 2). Absent the results springing from the response surface approach, the general conclusion would be that discrepancies between intrinsic values and rewards are predictive of work satisfaction, the strength of the relationship increases across the third decade of life and across subgroups, and may become quadratic over time for certain subgroups.
The results from the response surface models pertaining to the subgroups demonstrate that (a) work rewards are the principal driver in the prediction model, exhibiting a linear association during the early 20's and a quadratic association during the later 20's, (b) work values generally add no independent predictive power in the model, (c) the model fit varies by subgroups, and (d) work value-reward discrepancies, as reflected by the value X reward interaction term, only become influential at the aggregate level during the later 20's, and only for one subgroup, namely females with family roles (see Table 3). An examination of the betas suggests that work rewards are a moderate to strong predictor of work satisfaction on both occasions and work values add no predictive power. Females and those who would establish family roles exhibited slightly better model fits as indicated by the r-squares during the early 20's. While females continued to exhibit a better model fit than their male counterparts during the late 20's, the difference across the family roles and family roles X gender groups changed. Those participants without family roles and females without family roles exhibited the largest r-squares by the late 20's. The stability and change in model fit for the response surface models are generally consistent with the simple and residualized difference models in this regard. In Table 3, the quadratic terms associated with work rewards were statistically significant for females, participants without family roles, and females with and without family roles, suggesting a curvilinear association between work rewards and satisfaction by the late 20's for these groups. The females with family roles subgroup was the only one to exhibit a statistically significant association between work value-reward discrepancies and work satisfaction as reflected by the reward X value interaction term and akin to Figure 2. The latter results suggest that the work satisfaction of females with family roles appears to hinge on the work value-reward discrepancy, but this is not the case for any of the other subgroups.
The results from the response surface models echo much of what was found in the simple and residualized difference models and add to our understanding in ways that the simple and residualized difference models cannot. Intrinsic values generally play a marginal role in the prediction of work satisfaction. As compared to intrinsic values, intrinsic work rewards play an increasingly influential role in work satisfaction across the third decade of life. The increase in the role of intrinsic work rewards is particularly underscored by the shift from non-significant to statistically significant quadratic terms across the third decade of life for the response surface models. The response surface for all participants in Figure 2 reveals the quadratic nature of the relationship between intrinsic work rewards, values and satisfaction and suggests that higher levels of rewards show a peak in satisfaction and diminishing rewards are associated with precipitous declines in satisfaction. This response surface also shows that increasing work value-reward dissonance is associated with precipitous declines in satisfaction, particularly when values exceed rewards (e.g., the deficit side of the dissonance axis).
The results suggest that the association between intrinsic work value-reward dissonance and work satisfaction increases across the third decade of life and may be expressed as increasingly quadratic. This trend seems consistent with the population-level tendency for adults to seek, obtain, and become established in career-track jobs during the early 20's; hence, the importance of intrinsic value-reward dissonance grows in strength as young adults become established workers. This general trend is, however, highly conditional on gender and family roles.
Consistent with previous research on the meaning of work, the results generally suggest that gender socialization in combination with the socializing forces tied to establishing family roles may contribute to defining work as a means of supporting others more than as a means of satisfying personal values. As the groups progressed through their twenties, males and the group who established family roles exhibited a much weaker association between intrinsic work rewards and satisfaction than females and those who did not establish family roles. Consistent with research on the meaning of work, these trends suggest that the transition to family life may promote a duty-oriented view of work over a more self-gratifying view.
Although gender and family roles appear to have a moderating effect on the link between intrinsic rewards and satisfaction, the combination of both reveals that one group stands out from the rest when the focus shifts to work value-reward dissonance. The discrepancy between intrinsic values and rewards, as indicated by the interaction term in the response surface model, is a statistically significant predictor of work satisfaction for females with family roles at the end of the third decade of life. The job satisfaction of women without family roles, as well as men irrespective of their family roles, is primarily driven by their occupational rewards.
This pattern is contrary to our predictions, based on the MOW study, that both men and women with family roles would see work as an obligation, and therefore be less bothered by discrepancies between the rewards obtained in the workplace and their personal values. While this may be the case for men (as demonstrated here) and for women in later phases of the life course, the situation may be different for young women in their late twenties, who are working while also caring for young children. As a result of our sample inclusion criteria, the portion of the family roles group who had become mothers at age 28–29 would be caring for children mostly of preschool age. Though there is considerable cultural affirmation for equal occupational achievement for men and women, mothers still have primary responsibility for the care of children. For example, in the nationally-representative Monitoring the Future Study, male and female high school seniors in 1995 (about 4 years younger than the YDS cohort), were much more likely to endorse the traditional family structure (husband works full time, wife does not work) as the preferred arrangement when there are preschool children than the situation in which husbands and wives both work full time. Husbands working full-time and wives part-time was also considered highly acceptable for both genders (M. K. Johnson, Oesterle, & Mortimer, 2001). Though the small sample sizes preclude comparison of the effects of full-time versus part-time employment, it could be that working mothers of very young children are particularly bothered by the lack of satisfaction of their intrinsic values given the sacrifices they are making at home to go to work. Though we would prefer to compare married women to married and unmarried mothers and married mothers to single mothers, who are subject to different economic pressures to work, the size of the sample also made such fine-tuned investigation impossible.
On a more general level, these results suggest that increasingly large discrepancies between values and rewards may be associated with some acceleration in the decline in satisfaction for certain types of workers. On the basis of our analyses, females appear to be more affected by the establishment of family roles than males. This group (females with family roles) was the only one to manifest a consistent increase in the link between value-reward discrepancies and satisfaction over time. The establishment and maintenance of family roles may act as a destabilizing force as it pertains to the link between the satisfaction of work values and work satisfaction broadly. Given that females generally feel a greater pull away from work and toward family relative to males, this result seems to further support the conditioning effect of family roles on the link between work value-reward dissonance and satisfaction. Although interesting, these results and interpretation pertaining to women with family roles should be treated with caution given that only two occasions are being used to demonstrate possible quadratic changes in the target relationship over time and the fairly small size of this portion of the sample.
While general trends and subgroup comparisons across the simple, residualized, and response surface discrepancy score approaches yielded similar patterns for the groups across time, there were a few notable exceptions. The r-square estimates were generally higher for the response surface approach relative to the other two. This approach also revealed that intrinsic work rewards are the only significant predictor of satisfaction until age 28–29, and at this age, it remains the strongest predictor of satisfaction. These patterns suggest that young adults may employ other criteria aside from personal work values when considering what constitutes more or less of a reward. Michalos (1991) suggested a broad array of criteria that people employ when considering their relative work rewards. Future research could examine the relative predictive power of discrepancies between work rewards and personal values, what others receive, what one expects to receive now or in the future, and/or what one deserves. Theory generally asserts that discrepancies have a meaningful influence on satisfaction (Dawis & Lofquist, 1984; Michalos, 1985, 1991), but this study finds that this is mainly true for values as the referent during the later 20's and leaves open the question as to what, if any, criteria are employed to define discrepancies from obtained rewards before that time.
On a very broad level, the response surface results for the second occasion suggest that notions of value-reward consonance (an example of occupational fit) and dissonance (an example of occupational misfit) may need revision. Dissonance of the surplus kind appears to be linked to greater job satisfaction than equivalent levels of dissonance on the deficit side of the contour. These dissonance differences point to a possible threshold effect, whereby a person is generally satisfied with their work when they are at least meeting their values and increasingly satisfied up to a moderate surplus. More is better up to a point. The pattern of satisfaction along the consonance axis suggests that consonance as a result of high levels of values and rewards is associated with greater satisfaction than consonance born from low levels of each. There are probably several reasonable explanations for this phenomenon. One is that people who exhibit low levels of intrinsic values and rewards may have settled for their lot in life rather than having sought out a work situation to satisfy low levels of intrinsic work rewards. In contrast, people who highly value intrinsic rewards may be more apt to seek out work that offers more intrinsic rewards. Future research should be conducted to test the proposed model of work-value dissonance and consider the implications of this model as it pertains to more general theories of occupation fit.
It should be noted that the family status groups were generally equivalent on a range of socio-demographic variables pertaining to their educational and occupational attainment (with the exception being parents' educational attainment and respondent's pay rate), which reduces the possibility that the observed group differences were confounded by such factors. Despite the demographic similarities of the two groups, other unmeasured exogenous factors may distinguish who does and does not establish family roles and these variables may also contribute to the differences in the predictive nature of work value-reward dissonance. The reasons for the observed group differences may be manifold and future research should continue to explore and test explanations offered here as well as other possibilities. Several measurement limitations are worth noting as well. First, this study does not account for people in cohabiting relationships and same-sex unions. As a result, some fraction of those participants who are labeled as without family roles may in fact have quite intensive family roles. Second, the work satisfaction variable included only one indicator. Although this is a weakness, many studies employ a similar measurement strategy (Holtzman & Glass, 1999; Kalleberg & Mastekaasa, 2001). Finally, the reliability of the intrinsic work values scale was in the borderline range. These limitations call for continued research employing improved measures to ensure the validity of the results.
This study also employed discrepancy estimates with known limitations. An ongoing debate continues to question the validity of discrepancy scores variously computed with no single estimate gaining a strong and consistent endorsement (Cattell, 1983; Cronbach & Furby, 1970; Edwards, 1994; Edwards & Cable, 2009; Edwards & Parry, 1993; Williams, et al., 1987). This study, therefore, included three different estimates. On the one hand, models employing the three estimates demonstrated very similar findings. All three converged to support the general finding that the dissonance model demonstrates improved fit across time and for certain subgroups. Future research, like that pertaining to the response surface model, should continue to identify discrepancy estimates that address the noted concerns with existing estimates. This study adds to that ongoing work by contrasting the results of three discrepancy estimates and finding some important similarities and differences across the three.
Notwithstanding these limitations, the present study provides evidence that work-value reward discrepancies become more powerful predictors of job satisfaction during the decade of the twenties, and particularly for women who establish family roles during this period. Further investigation is necessary to replicate the pattern found here, and to discover the processes that underlie the interactions of time, gender, and family status that are suggested by these analyses.
The Youth Development Study is supported by a grant from the National Institute of Child Health and Human Development (HD 44138), “Work Experience and Mental Health: A Panel Study of Youth.” The National Institute of Mental Health (MH 42843) provided previous support for this research. The content is solely the responsibility of the authors and does not represent the official views of the National Institute of Child Health and Human Development or the National Institutes of Health.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.