Longitudinal Stability of Work-Family Enrichment and its Association with Well-Being and Personality Traits

Abstract

Are higher levels of work-family enrichment a consequence or manifestation of certain personality traits and individuals’ psychological functioning? Using random intercept cross-lagged panel models, this study examined the hypothesized stability of work-to-family enrichment (WFE) and family-to-work enrichment (FWE) over two 10-year intervals, the extent to which the within-person changes of WFE and FWE are associated with personality traits and psychological well-being (PWB), and possible gender differences. In this 20-year, longitudinal data analysis of employed adults (N=535), results indicated the robust nature of the stability of WFE/FWE. Our results suggest that personality traits were not associated with within-person change for either WFE or FWE, but PWB was associated with within-person change. Theoretically and conceptually, our findings provide strong evidence that work-family enrichment is not simply an “optimistic worldview” created by personality and well-being. The within-person results lend strong evidence that interventions that improve psychological well-being will also enhance work-family enrichment.

Counseling psychologists are keenly aware of the false distinction between career counseling and psychotherapy, and they have vigorously called for the integration of clients’ work lives into psychotherapy practice (e.g., Juntunen, 2006; Schultheiss, 2006). Work and family, regardless of the structure or form either takes, are inextricably linked major domains of life, making the work-family interface one of the most critical areas that counseling psychologists can identify and target in support of clients’ psychological and overall life functioning (e.g., Schultheiss, 2006). Indeed, everyday work and family life “constitute the backbone of human existence” (Aryee et al., 1999, p. 497), and some theorists contend that success at work and in the family domain is a primary task of successful adult development (Lachman & James, 1997).

A growing body of research across disciplines is focusing on work-family enrichment, one indicator of success at both work and in the family (e.g., Greenhaus & Powell 2006). Work-family enrichment is defined as “the extent to which experiences in one role or domain (e.g., work) improve the quality of life in another role or domain (e.g., family)” (Greenhaus & Powell, 2006, p. 73). Concrete examples of work-family enrichment take many forms. For example, consider the practicing therapist who has refined the skill of boundary setting through professional training and clinical experience, who then applies that skill at home to create boundaries for personal care as well as boundaries that protect the family. Similarly, consider the construction worker whose occupation allows access to specialized equipment, personal skills and professional connections that result in saved money and decreased stress on a home renovation. Another example of work-family enrichment reported in the literature are the skills and patience developed through parenting, helping managers better lead subordinates or help their employees problem-solve (Ruderman et al., 2002). Collectively these examples illustrate that work-family enrichment is possible in both directions (i.e., work to family and family to work), and it is possible across diverse occupational and family contexts.

Theoretical Framework

The theoretical origins of work-family enrichment emerged from inter-related strands of sociologically informed role theory. In contrast to the prevailing zeitgeist that multiple role occupancy (i.e., the idea of individuals participating in various role relationships) left individuals strained and over-committed (Goode, 1960), Sieber (1974) argued that multiple role occupancy, like women’s involvement in the workplace in addition to home responsibilities, benefitted individuals. Specifically, Sieber (1974) argued that involvement in multiple roles enabled acquisition of beneficial social and economic resources, which are useful for well-being. Later, Marks (1977) argued that additional resources acquired through role accumulation were, all else being equal, an advantage that offset additive burdens that might arise from multiple role occupancy. Subsequently formalized into role expansion theory (Barnett & Hyde, 2001), a body of research has accumulated to suggest that simultaneous occupancy of multiple, well-fitted roles in the work and family domain has several benefits to individuals, including enhanced psychological well-being (Wolfram & Gratton, 2014).

Work-Family Enrichment Literature

Indeed, work-family enrichment has been linked with a variety of beneficial outcomes. Cross-sectional and longitudinal evidence indicates that work-family enrichment is independently and negatively associated with depression (e.g., Hammer et al., 2005), anxiety (e.g., Grzywacz & Bass, 2003), and turnover intentions (e.g., McNall et al., 2010). Additionally, work-family enrichment is uniquely and positively associated with vigor at work (Cinamon & Rich, 2010) and thriving at work (Russo et al., 2018), each of which are associated with positive psychological functioning or “a life worth living.” (Urry et al., 2004). Work-family enrichment has also been directly associated with eudaimonic well-being (Grzywacz, 2000), suggesting that it is closely aligned with counseling psychology’s focus on prevention science, and building individuals’ strengths and positive attributes (Whiston et al., 2012).

Conceptual Concerns in Work-Family Enrichment Literature

Unfortunately, several conceptual issues require clarification before counseling psychologists can leverage the potential benefits that may arise from helping clients obtain greater work-family enrichment. The foremost conceptual issue is ambiguity related to the temporal stability of work-family enrichment. Evidence suggests that work-family enrichment is “relatively stable” over short periods like six months (Knecht et al., 2016) and one year (Hammer et al., 2005; Moazami-Goodarzi et al., 2015). Based on these existing studies, work-family enrichment, has shown to have a moderate to high levels of stability across time (r=.55–.64 and r=0.61–.74; Knecht et al., 2016; Moazami-Goodarzi et al., 2015, respectively), highlighting the construct’s robustness. Unfortunately, interpreting the meaning of work-family enrichment’s stability is difficult. On one hand, most individuals do not experience major life changes like marriage, childbearing, and job transitions within narrow observation periods like six months or one year (Rantanen et al., 2008), so stability might be expected. Although, on the other hand, the apparent stability may also be simple manifestation of enduring between-person attributes like personality. Indeed, a meta-analysis concluded that personality traits like extraversion, agreeableness, conscientiousness, and openness to experience underlie work-family enrichment (e.g., Michel et al., 2011). Further, recent trends in personality research theorize that broad personality traits produce more “narrow” phenotypes of personality (e.g., Costa & McCrae, 2017; McAdams & Pals, 2006), perhaps like appraisals of work-family enrichment. If true, the stability of work-family enrichment could simply mark between-person differences in personality and trait well-being, leaving open the possibility that personality or levels of well-being may lead to experiencing greater work-family enrichment.

The role of gender is a second fundamental issue impeding the understanding and subsequent action around work-family enrichment. There is a long history of examining differences in work-family experiences between women and men because the worlds of work and family are often considered gendered (e.g., Cinamon & Rich, 2002; Powell & Greenhaus, 2010). Nevertheless, despite beliefs that “balancing” work and family is a “women’s issue” (Marshall, 1992), meta-analytic results found no gender differences in levels of work-family enrichment or its predicters (e.g., Lapierre et al., 2018). The observed similarity among women and men in reported levels of work-family enrichment can be interpreted in two basic ways. First, it can be interpreted as indicating that historic arguments of the gendered nature of work and family are no longer accurate. Similar levels of work-family enrichment among women and men can also be interpreted as indicating a shared tendency for personality or trait well-being to construct comparable experiences of work-family enrichment over time. However, a more nuanced perspective might be necessitated as some literature indicates that gendered differences exist amongst some correlates of work-family enrichment. Specifically, the relationship between work-family enrichment and various work-domain variables such as organizational affective commitment (i.e., an individual’s emotional commitment and identification with an organization; Allen & Meyer, 1990; Marques et al., 2015), work-life balance policies (e.g., Baral & Bhargava, 2011), and job satisfaction (e.g., Tang et al., 2014) were stronger for women than men. On the other hand, there is evidence that the relationship between work-family enrichment and marital status is only significant for men (e.g., Stoiko et al., 2016), while the relationship with various job characteristics (e.g., job autonomy, variety, significance, and feedback) is stronger for men (e.g., Baral & Bhargava, 2011). Taken together, these mixed findings suggest that there may be nuanced gender differences with various work and family domain variables, hinting at the possibly gendered nature of work-family enrichment as a construct.

Current Study

The overall goal of this study was to determine the long-term stability of work-to-family enrichment (WFE) and family-to-work enrichment (FWE) over two 10-year intervals. A protracted (20 year) observation period is particularly valuable because managing work and family is required throughout most of adulthood, and specific responsibilities within work and family can vary substantially throughout this period (Grzywacz et al., 2002). It is also consistent with counseling psychologists’ focus on lifespan development of clients. Delineating stability over time is a necessary first step to determining and understanding patterns of continuity and change in typical WFE and FWE across adulthood.

A central aim of this analysis is determining whether work-family enrichment is merely a narrow expression of between-person attributes like personality and trait well-being. In light of developmental theory positing that success in the work and family domains is a fundamental task of adulthood (Lachman & James, 1997), and contentions that working adults and their families develop adaptive strategies for meeting work and family responsibilities (Moen & Wethington, 1992), the core hypothesis tested is that both WFE and FWE will manifest as relatively stable experiences across adulthood. Since WFE and FWE are presumed to shape experiences, appraisals, and attributions of everyday work and family experiences, we hypothesized that both personality and psychological functioning will account for only a modest amount of stability in WFE and FWE, as per findings in the existing literature. Additionally, we hypothesized that there will be gender differences in the stability of work-family enrichment, including both WFE and FWE, across time.

Method

Participants

Participants were drawn from the first, second, and third waves of the Mid-Life Development in the United States (MIDUS; Brim et al., 1995–1996; Ryff et al., 2004–2006; Ryff et al., 2013–2014) survey. The MIDUS is a national survey, with a sample of over 7000 individuals aged 25–74, who were examined in areas of physical and psychological well-being and social factors. MIDUS participants were followed across three observations spanning approximately 20 years. The original sample was obtained through a random digit dialing procedure, which obtained household phone numbers via working telephone banks from the conterminous United States. Though it is uncertain if cell phones were used to contact participants during the baseline sample recruitment in 1995 (MIDUS I), once a respondent is recruited, further contact information (e.g., cell phones, landlines, email) was collected for each participant. All of this contact information was used to contact the participants for the second (MIDUS II) and third (MIDUS III) waves of the study. Data from the first wave was collected in 1995–1996 (Brim et al., 1995–1996) and included a computer-assisted telephone interview, lasting approximately 30 minutes, and a comprehensive self-administered questionnaire that was returned via mail. The second wave (Ryff et al., 2004–2006) and the third wave (Ryff et al., 2013–2014) were conducted approximately 10 years after the previous waves, using the same baseline survey procedures as the first wave. The attrition rate for each wave was 30% from wave 1 to wave 2 and 34% from wave 2 to wave 3. The sample in the current study consisted of employed adults in the United States. Furthermore, participants, who were age 45 or below at MIDUS I were used to capture the “prime time” that adults work in the course of their lifespan because we wanted to capture the majority of working time and average retirement age of 65 based on literature (e.g., Lumsdaine et al., 1996). This resulted in 535 adults (50.8% female and 49.2% male participants) who participated at all three time points. Participants’ age ranged from 24 to 45 (M = 36.34, SD = 5.72) at MIDUS-I. Detailed characteristics of the participants and their job characteristics can be found in Table 1.

Table 1.

Participant Characteristics.

Characteristics	N	%		Characteristics	N	%
Gender				Employment status
Women	272	50.8%		Full-time	480	89.7%
Men	263	49.2%		Part-time	52	9.7%
Race/ethnicity				Missing	3	0.6%
Caucasian	506	94.6%		Major industries (Time 1)
African American	16	3.0%		Professional and related services	196	36.6%
Other	7	1.3%		Manufacturing	86	16.1%
Native American	2	0.4%		Retail trade	49	9.2%
Multiracial	2	0.4%		Transportation, communications, and public utility	44	8.2%
Asian	1	0.2%		Public administration	43	8.0%
Missing	1	0.2%		Wholesale trade	27	5.0%
Marital status				Finance, insurance, and real estate	25	4.7%
Married	375	70.1%		Construction	22	4.1%
Never married	108	20.2%		Business and repair services	17	3.2%
Divorced	44	8.2%		Agriculture, forestry, fishing, and mining	8	1.5%
Separated	4	0.7%		Entertainment and recreational services	6	1.1%
Widowed	4	0.7%		Personal services	1	0.2%
Educational status				Major industries (Time 2)
Junior high school	2	0.4%		Professional and related services	212	39.6%
Some high school	7	1.3%		Manufacturing	93	17.4%
High school degree or equivalent	113	21.1%		Retail trade	36	6.7%
Some college (No degree)	116	21.7%		Transportation, communications, and public utility	41	7.7%
Associates degree	44	8.2%		Public administration	37	6.9%
College degree	150	28.0%		Wholesale trade	26	4.9%
Some graduate school	22	4.1%		Finance, insurance, and real estate	31	5.8%
Masters degree	59	11.0%		Construction	15	2.8%
Doctoral degree	21	3.9%		Business and repair services	11	2.1%
Missing	1	0.2%		Agriculture, forestry, fishing, and mining	5	0.9%
Income				Entertainment and recreational services	6	1.1%
$0–$9,999	46	8.6%		Personal services	6	1.1%
$10,000–$19,999	79	14.8%
$20,000–$29,999	140	26.2%
$30,000–$39,999	108	20.2%
$40,000–$49,999	65	12.1%
$50,000–$74,999	59	11.0%
$75,000–$99,999	19	3.6%
$100,000 or more	11	2.1%
Missing	8	1.5%

Measures

Work-Family Enrichment

Items were developed specifically for the MIDUS survey to measure work-to-family enrichment (WFE) and family-to-work enrichment (FWE). Each construct was assessed at all three time points using the same eight items; four assessing WFE (e.g., “The things you do at work make you a more interesting person at home”) and four assessing FWE (e.g., “Talking with someone at home helps you deal with problems at work”). Participants responded to each item on a scale ranging from 1 (“all the time”) to 5 (“never”). Item responses were coded and summed to create separate variables reflecting WFE and FWE, such that higher scores indicated greater enrichment (Grzywacz & Marks, 2000). The WFE and FWE scales show evidence of concurrent and construct validity, such that work-family enrichment scales were related to better physical and mental health (Grzywacz, 2000; Grzywacz & Bass, 2003). The scales have been shown to have positive correlations with constructs such as family satisfaction, job satisfaction, agreeableness, and conscientiousness (e.g, Wayne et al., 2004), further highlighting the scales’ construct validity. Other studies using the same scales in MIDUS samples have found the work-to-family (Cronbach’s α = 0.72–0.74) and family-to-work (Cronbach’s α = 0.68–0.70) enrichment scales to have adequate internal consistencies (e.g., Grzywacz, 2000; Grzywacz & Marks, 2000; Wayne et al., 2004). The coefficient omega of WFE and FWE ranged from .65 to .72 over the three timepoints, and the Cronbach’s α for assessments of WFE and FWE ranged from 0.60 to 0.71 over time in the current sample.

Personality Traits

Personality traits were assessed using an inventory created for the MIDUS (Lachman & Weaver, 1997). The inventory began with creating a list of all the personality dimension adjectives cited within the literature (at the time), along with those that appeared as the most consistent markers of specific personality traits. Adjectives displaying the highest item-to-total correlations or factor loadings in a pilot study were selected and included in the final inventory used in the MIDUS I and MIDUS II. This inventory consisted of four items assessing neuroticism (e.g., “Moody”, “Worrying”), five items measuring extraversion (e.g., “Outgoing”, “Friendly”), seven items assessing openness to experience (e.g., “Creative”, “Imaginative”), four items measuring conscientiousness (e.g., “Organized”, “Responsible”), and five items each measuring agreeableness (e.g., “Helpful”, “Warm”), and agency (e.g., “Self-confident”, “Forceful”). For each of the items, participants rated their responses on a 4-point scale, ranging from 1 (“a lot”) to 4 (“not at all”). Each item set was coded and averaged, such that higher scores reflected greater levels of the respective personality trait.

The MIDUS personality assessments demonstrate good evidence of construct validity. Agreeableness, openness to experience, and extraversion have been found to be negatively correlated with daily negative affect, positively correlated with positive affect (Leger et al., 2016), and positively correlated with subjective well-being (Weiss et al., 2008). Across different MIDUS samples, agreeableness (Cronbach’s α = .70–.81), extraversion (Cronbach’s α = .71–.78), and agency (Cronbach’s α = .80–.81) have all displayed adequate to good reliability (Dunkel et al., 2015; Figueredo et al., 2004). Meanwhile, openness to experience (Cronbach’s α = .65–.77) and neuroticism (Cronbach’s α =.62–.75) have shown modest to acceptable levels of reliability, and the internal consistency of conscientiousness has been below conventionally accepted values (Cronbach’s α = .58–.68; Dunkel et al., 2015; Figueredo et al., 2004). In the current study, the coefficient omega and Cronbach’s α for neuroticism, extraversion, openness to experience, conscientiousness, agreeableness, and agency in were .74, .78, .73, .54, .80, and .82, and .70, .77, .74, .53, .79, and .81 in MIDUS I and .74, .79, .75, .54, .83, and .84 and .73, .77, .76, .53, .82, and .83 in MIDUS II, respectively.

Psychological Well-Being

Participants’ psychological well-being was assessed in the MIDUS I and MIDUS II using the 18-item version of Ryff’s Scales of Psychological Well-Being (RPWB; Ryff & Keyes, 1995), which measures psychological functioning across six domains. The 18-item RPWB has demonstrated adequate convergent and discriminant validity as it was correlated with single-item measures of happiness and life satisfaction, and all of its subscales were negatively correlated with multiple measures of depression (Ryff & Keyes, 1995). Following Ryff’s and Keyes’ recommendation, psychological well-being was treated as a second-order unidimensional construct in the current study. Participants rated how much they agreed with each item (e.g. “Maintaining close relationships has been difficult and frustrating for me”) on a scale from 1 (“strongly agree”) to 7 (“strongly disagree”). Responses were recoded as needed and summed so that higher scores indicated greater well-being (Ryff et al., 2003). Other studies have found the 18-item scale to have acceptable to good internal consistency (Cronbach’s α = 0.75–0.82) in MIDUS samples (e.g., Lee et al., 2016; Mishra et al., 2019). In the present study, the coefficient omega and the Cronbach’s α estimates were .80 and .81 for Time 1 and .84 and .83 for Time 2.

Data Analysis

Latent variables were formed using individual items except for psychological well-being, for which subscales were used as indicators. In addition, one of the loadings in each latent variable was fixed to one for model identification, and we allowed the residual variances among the corresponding indicators to associate over time.

When examining whether WFE and FWE are a narrow form of trait-like attributes, it becomes critical to separate within-person dynamics from between-person differences by including random intercepts (Hamaker et al., 2015). In the current study, distinguishing within-person from between-person components was done by conducting a random-intercept cross-lagged panel model (RI-CLPM; Hamaker et al., 2015), which allows to separately estimate how much individuals fluctuate around their specific scores (i.e., within-person autoregressive paths), the dynamics of these fluctuations between WFE and FWE (i.e., within-person cross-lagged paths), and to what extent individuals will remain at their stable, trait-like level across time (i.e., between-person variation). The advantage of conducting RI-CLPM over the traditional CLPM is that RI-CLPM allows researchers to disaggregate within- between-person sources of variance. Following the common procedure in multi-level modeling, the intra-class correlations (ICC) were calculated in examining whether there is sufficient variance at the between-person and within-person level. For WFE, the ICC was .363 and .432 for women and men, respectively. This implies that 36.3% and 43.2% of the variance in WFE can be explained by differences between individuals. Likewise, the ICC for FWE was .403 and .459 for women and men, respectively. These results indicate that a substantial part of the variance in both WFE and FWE were due to stable differences between individuals.

Measurement invariance was assessed by testing a sequence of factorial invariances across both time and gender. Although previous studies have tested factorial invariance of the work-family enrichment items across gender and two time points (e.g., Babic et al., 2017; Cho et al., 2013), the process was repeated because those previous analyses were undertaken with only two waves of the MIDUS or among employees in Belgium. Furthermore, considering that the purpose of the current study was to examine the longitudinal associations between WFE and FWE, we did not examine the residual invariance model (i.e., constraining residuals to be equal over time), and therefore, the residuals were freely estimated in all models (Little, 2013; Vandenberg & Lance, 2000). Following the recommendation for longitudinal models that intercepts and variances of each indicator for a latent construct should not differ across all measurement occasions (Little, 2013; Widaman & Thompson, 2003), the longitudinal null model for WFE and FWE was fit so that each indicator had one variance, all covariances were fixed to zero, and the variances and means were constrained to be equal over time. This information was used to calculate the Comparative Fit Index (CFI), which is one of the goodness of fit indices used in the current study.

Assessment of structural models was done after measurement invariance was established. Because the chi-square statistic is sensitive to sample size, we also assessed other fit indices. Values greater than .95 for CFI and Tucker-Lewis Index (TLI) and standardized root mean squared residual (SRMR) values less than .08 are considered acceptable (Hu & Bentler, 1999). We also interpreted values greater than .90 for CFI and TLI as acceptable (Little, 2013). In addition, values of root mean square error of approximation (RMSEA) < .05 indicate good fit, and values between .05 – .10 indicate adequate fit (MacCallum et al., 1996). All analyses were conducted with Mplus, Version 8.0 (Muthén & Muthén, 1998–2017).

In summary, invariance testing was conducted to examine a potential influence of gender or time on the item level of WFE and FWE where an item may function differently based on time or gender. RI-CLPM was conducted to examine the following hypotheses: 1) WFE and FWE will manifest as stable experiences across adulthood, and 2) personality and psychological functioning will account for a modest amount of stability in WFE and FWE. Furthermore, potential gender differences in stability of work-family enrichment was also examined within a RI-CLPM framework.

Results

Missing responses totaled 0.3% of all item responses. When examining qualitative differences between participants with missing and no missing responses, no differences were found for age, WFE, FWE, and psychological well-being across the 3 time points, agreeableness, and agency (p > .05). However, those with missing responses had lower levels of extraversion (t (531) = 2.41, p = .016) and openness to experience (t(528) = 2.35, p = .019). No differences were found in terms of gender (χ²(1) = 1.68, p = .195) and race (χ²(5) = 7.09, p = .214). These results suggest the importance of controlling for levels of extraversion and openness to experience to minimize the influence of missingness in our models. For all models, missing data was handled using Full Information Maximum Likelihood (FIML).

Lilliefors’ (1967) test was used to assess for univariate normality of all individual items in each of the three waves. This test indicated that all the variables across three waves violated the normality assumption (p < .05). Similarly, when inspecting the multivariate normality using the TECH13 option in Mplus (Muthén & Muthén, 1998–2017), both multivariate skewness (p < .001) and kurtosis (p < .001) were statistically significant, indicating violations of the multivariate normality assumption. To deal with non-normality, the maximum likelihood estimation with robust standard errors (MLR) was used. As a result, the Satorra-Bentler (S-B) scaled chi-square difference test (Satorra & Bentler, 2001) was used to compare nested models.

Means, standard deviations, Cronbach’s α’s, and zero-order correlations based on the observed scores are presented in Table 2.

Table 2.

Correlations, Means, Standard Deviations, and Reliability Coefficients.

	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18
1. PosWF(T1)	1.00
2. PosFW(T1)	.33***	1.00
3. PosWF(T2)	.43***	.15**	1.00
4. PosFW(T2)	.20***	.45***	.32***	1.00
5. PosWF(T3)	.35***	.17***	.41***	.17***	1.00
6. PosFW(T3)	.17***	.38***	.19***	.48***	.42***	1.00
7. PWB (T1)	.19***	.35***	.10*	.25***	.14**	.20***	1.00
8. PWB (T2)	.16***	.22***	.19***	.27***	.24***	.27***	.59***	1.00
9. N (T1)	−.09*	−.23***	.02	−.06	−0.05	−.08	−.46***	−.28	1.00
10. E (T1)	.15***	.22***	.07	.20***	.13**	.19***	.46***	.32***	−.11**	1.00
11. O (T1)	.23***	.15**	.17***	.14**	.21***	.11*	.38***	.31***	−.13**	.46***	1.00
12. Agr (T1)	.09*	.15**	.07	.15**	.06	.14**	.29***	.26***	.02	.45***	.30***	1.00
13. Agy (T1)	.22***	.13**	.12**	.19***	.14**	.15**	.35***	.29***	−.06	.52***	.42***	−.002	1.00
14. N (T2)	−.02	−.11*	−.02	−.04	−.13**	−.08	−.35***	−.43***	.64***	−.08	−.09*	−.03	−.06	1.00
15. E (T2)	.13**	.15**	.13**	.13**	.17***	.19***	.35***	.45***	−.03	.68***	.35***	.36***	.40***	−.12**	1.00
16. O (T2)	.25***	.20***	.22***	.18***	.24***	.14**	.33***	.41***	−.11*	.32***	.70***	.19***	.36***	−.15**	.46***	1.00
17. Agr (T2)	.07	.11*	.07	.11*	.04	.08	.22***	.31***	−.03	.27***	.20***	.66***	−.01	−.13**	.47***	.29***	1.00
18. Agy (T2)	.15**	.11*	.15***	.18***	.21***	.18***	.25***	.36***	−.02	.38***	.30***	−.03	.69***	−.04	.46***	.42***	−.04	1.00
M	11.32	13.35	11.57	13.63	11.13	13.12	103.69	100.30	2.31	3.15	2.98	3.38	2.65	2.11	3.02	2.87	3.33	2.52
SD	2.57	2.85	2.47	2.66	2.68	2.68	13.21	13.74	0.63	0.55	0.48	0.50	0.64	0.61	0.57	0.52	0.54	0.67
α	0.69	0.68	0.63	0.61	0.71	0.60	0.81	0.83	0.70	0.77	0.74	0.79	0.81	0.73	0.77	0.76	0.82	0.83
omega	0.70	0.70	0.65	0.65	0.72	0.65	0.80	0.84	0.74	0.78	0.73	0.80	0.82	0.74	0.79	0.75	0.83	0.84

Note. WFE = Work-to-family enrichment; FWE = Family-to-work enrichment ; PWB = Psychological well-being; N = Neuroticism; E = Extraversion; O = Openness to experience; C = Conscientiousness; Agr = Agreeableness; Agy = Agency.

^*p < .05,

^**p < .01,

^***p < .001.

Random Intercept Cross-Lagged Panel Models

Measurement Invariance

Multiple-group longitudinal measurement invariance was examined to assess whether latent factors were invariant between gender and across three-time points. The first model was a configural model where both genders were included in a joint analysis, but no equality constraints were placed on either “gender” or “time.” Then, a series of models with additional constraints were fit and the overall model fit and degree of decrement in model fit compared to the previous model was assessed. In evaluating measurement invariance, the decrement was assessed by goodness of fit statistics of the constrained model relative to a previous model without constraints. Change in the Comparative Fit Index (ΔCFI) < .01 (Cheung & Rensvold, 2002) was used to indicate whether a significant decrease in the model fit was observed from less to more constrained models.

For the configural model, all factors could covary. The chi-square test produced a significant result, χ²₍₄₂₆₎ = 689.48, p < .001, and inspection of fit indices suggested acceptable fit: SRMR = .065, CFI = .924, and RMSEA = .048, (90% Confidence Interval (CI) [.041, .055]). While inspecting the modification indices, the cross-loading of item 3 of the WFE (i.e., “Having a good day on your job makes you a better companion when you get home.”) factor on the FWE factor was standing out for both genders. Based on empirical reports discussing differing levels of job satisfaction for men and women over time, (e.g., Boswell, Shipp, Payne, & Culbertson, 2009), among women, this cross-loading was added at all three time points, and among men, the cross-loading was added at Time 1. This significantly improved model fit: χ²₍₄₂₂₎ = 630.69, p < .001, and inspection of fit indices suggested acceptable fit: SRMR = .058, CFI = .940, and RMSEA = .043, (90% CI [.036, .050]). The subsequent model tested metric invariance by adding equality constraints between genders and over time to the model specification. The chi-square test produced a significant result, χ²₍₄₅₇₎ = 696.91, p < .001. Other fit indices maintained good model fit with SRMR = .071, CFI = .931, and RMSEA = .044 (90% CI [.038, .051]). Compared to the configural model, ΔS-Bχ²₍₃₅₎ = 64.30, p = .002; however, ΔCFI = .009, suggesting that the model did not significantly change using the criterion of ΔCFI < .01. The next model (scalar invariance) retained the constraints from the metric invariance model and added equality constraints between genders and across time for item intercepts. The chi-square test was significant, χ²₍₄₈₉₎ = 855.66, p < .001, SRMR = .079, CFI = .894, and RMSEA = .053 (90% CI [.047, .059]). When compared to the metric invariance model, ΔS-Bχ²₍₃₂₎ = 171.32, p < .001, and ΔCFI = .037, indicating significant change in the scalar invariance model.

To achieve partial scalar measurement invariance across gender and time, five intercepts were allowed to vary across gender and time. The adjustments were made on item-level analyses indicating whether each item would function differently based on time or gender. These decisions were all based on theoretical and empirical findings in lifespan and gender differences in gendered work (Cinamon & Rich, 2002; Powell & Greenhaus, 2010), job satisfaction (e.g., Boswell et al., 2009), and problem-solving strategies (Strough et al., 1996). Whether an intercept should be freely estimated across time or gender or both gender and time was determined by conducting a chi-square difference test between the less constrained model (i.e., releasing intercepts for both time and gender) and the more constrained model (i.e., releasing an intercept for either time or gender; see Table 3). The final measurement model allowed five distinct intercepts for individual items to vary. First, based on empirical reports discussing differing levels of job satisfaction over time, the final measurement model allowed the intercept of item 3 of WFE (i.e., “Having a good day on your job makes you a better companion when you get home.”) to vary at Time 3 (e.g., Boswell et al., 2009). Next, consistent with gender and lifespan differences in problem-solving and salience of interpersonal concerns (e.g., Strough et al., 1996), the final measurement model allowed the intercept of item 1 of FWE (i.e., “Talking with someone at home helps you deal with problems at work.”) to vary at Time 3 and between gender. Finally, based on the traditional view that work and family are gendered and that working women engage in “double duty” (Cinamon & Rich, 2002; Powell & Greenhaus, 2010), the final measurement model allowed the intercept of item 4 of FWE (i.e., “Your home life helps you relax and feel ready for the next day’s work.”) at Time 1 to vary between gender. Adding these modification indices resulted in: χ²₍₄₈₃₎ = 756.90, p < .001, SRMR = .074, CFI = .921, and RMSEA = .046 (90% CI [.040, .052]). Compared to the metric invariance model, ΔS-Bχ²₍₂₆₎ = 62.14, p < .001; however, ΔCFI < .01. We accepted this partial scalar invariance model, and the measurement portion of subsequent models used the same restrictions from this model.

Table 3.

Goodness of Fit Statistics for All Tested Measurement Models.

Model	S-Bχ 2	df	ΔS-Bχ2	Δdf	p	SRMR	CFI	ΔCFI	TLI	RMSEA		ΔS-Bχ2	Δdf	p	Comparison
										Estimate	90% CI
0. Null	4105.21	632	-	-	-	-	-	-		-	-	-	-	-	-
1a. Configural	689.48	426	-	-	-	.065	.924	-	.895	.048	.041, .055	-	-	-	-
1b. Configural (item 3 of WFE loading on FWE)	630.69	422	-	-	-	.058	.940	-	.916	.043	.036, .050	-	-	-	-
2. Weak invariance	696.91	457	64.30	35	.002	.071	.931	.009	.911	.044	.038, .051	-	-	-	-
3. Strong invariance	855.66	489	171.32	32	< .001	.079	.894	.037	.873	.053	.047, .059	-	-	-	-
3.1a. Release item 3 of T3WFE (time)	818.62	488	129.66	31	< .001	.078	.905	.026	.885	.050	.044, .056	2.56	1	.110	3.1c.
3.1b. Release item 3 of T3WFE (gender)	847.80	488	162.47	31	< .001	.078	.896	.035	.875	.053	.047, .058	45.01	1	< .001	3.1c.
3.1c. Release item 3 of T3WFE (both)	816.22	487	126.95	30	< .001	.076	.905	.026	.885	.050	.044, .056	-	-	-	-
3.2a. release item 1 of T3FWE (time)	802.85	487	112.71	30	< .001	.075	.909	.022	.890	.049	.043, .055	8.46	1	.004	3.2c.
3.2b. Release item 1 of T3FWE (gender)	818.11	487	129.42	30	< .001	.076	.905	.026	.885	.050	.044, .056	23.79	1	< .001	3.2c.
3.2c. Release item 1 of T3FWE (both)	795.30	486	104.42	29	< .001	.076	.911	.020	.892	.049	.043, .055	-	-	-	-
3.3a. Release item 4 of T1WFE (time)	786.94	485	95.43	28	< .001	.075	.913	.018	.894	.048	.042, .054	10.93	1	.001	3.3c.
3.3b. Release item 4 of T1WFE (gender)	778.17	485	85.37	28	< .001	.075	.916	.015	.897	.048	.041, .054	0.04	1	.848	3.3c.
3.3c. Release item 4 of T1WFE (both)	778.06	484	85.52	27	< .001	.075	.915	.016	.897	.048	.041, .054	-	-	-	-
3.4a. Release item 1 of T1FWE (time)	773.67	484	80.58	27	< .001	.075	.917	.014	.898	.047	.041, .053	8.91	1	.003	3.4c.
3.4b. Release item 1 of T1FWE (gender)	765.22	484	71.27	27	< .001	.074	.919	.012	.901	.047	.040, .053	0.00	1	.954	3.4c.
3.4c. Release item 1 of T1FWE (both)	765.08	483	71.30	26	< .001	.074	.919	.012	.901	.047	.040, .053	-	-	-	-
3.5a. Release item 4 of T1FWE (time)	762.73	483	68.59	26	< .001	.074	.919	.011	.902	.047	.040, .053	6.47	1	.011	3.5c.
3.5b. Release item 4 of T1FWE (gender)	756.90	483	62.14	26	< .001	,074	.921	.0098	.904	.046	.040, .052	0.37	1	.542	3.5c.
3.5c. Release item 4 of T1FWE (both)	756.47	482	61.89	25	< .001	.074	.921	.010	.903	.046	.040, .052	-	-	-	-

Note. All chi-square difference tests for invariance testing among configural, metric invariance, and scalar invariance models were significant at p < .001. Bold model indicates the best fitting model for releasing constraints for intercepts among time, gender, or both time and gender based on the Satorra-Bentler scaled chi-square difference tests. FWE = Family-to-work enrichment ; WFE = Work-to-family enrichment; T1 = Time 1; T2 = Time 2; T3 = Time 3.

Structural Model

Given that partial scalar measurement invariance was established between genders and over time, between-person variation, within-person autoregressive paths, and within-person cross-lagged paths were examined. This model was a good fit to the data: χ²₍₄₉₁₎ = 769.02, p < .001; SRMR = .075; CFI = .914; TLI = .904; RMSEA = .046 (90% CI [.040, .052]). To compare the gender difference, this unconstrained model was compared to the model with all within-person autoregressive and cross-lagged paths and covariances (i.e., random intercepts and within-person variances of WFE and FWE) constrained to be equal across gender (χ²₍₅₀₃₎ = 785.73, p < .001; SRMR = .077; CFI = .913; TLI = .905; RMSEA = .046 (90% CI [.040, .052]). The chi-square difference test produced a non-significant difference between these models (ΔS-Bχ²₍₁₂₎ = 16.77, p = .159), indicating no significant differences in the regression paths and covariances by gender. As a result, all subsequent models collapsed gender into a single model (χ²₍₂₃₅₎ = 411.56, p < .001; SRMR = .053; CFI = .942; TLI = .932; RMSEA = .037 (90% CI [.031, .043]). At the between-person level, the random intercept variances were significant for both WFE (σ² = .13, p < .001) and FWE (σ² = .10, p < .001), capturing the stability of WFE and FWE, and indicating the presence of significant variability in the levels of WFE and FWE across three timepoints between individuals. Furthermore, there was a significant positive correlation between the stable traits of WFE and FWE (r = .55, p = .001), indicating that individuals who reported higher WFE across the three timepoints also reported higher FWE across the three timepoints. When inspecting the within-person autoregressive and cross-lagged paths, none of these paths were statistically significant (p > .05), suggesting no significant within-person fluctuation for either WFE or FWE (see Figure 1).

Figure 1. Structural Model Before Adding Covariates.

Note. Regression paths shown are standardized regression paths. Covariances shown are standardized latent correlations. Although not shown in the figure, the residual variances among the corresponding indicators over time were allowed to associate. B-FWE = Between-person family-to-work enrichment; B-WFE = Between-person work-to-family enrichment; W-FWE = Within-person family-to-work enrichment; W-WFE = Within-person work-to-family enrichment; T1 = Time 1; T2 = Time 2; T3 = Time 3.

** p < .01, *** p < .001.

Covariate Effects

A separate RI-CLPM was fit to determine how personality traits, psychological well-being, employment at Time 1 (full-time = 1, part-time = 0), and two dummy coded major industries (i.e., Professional and related services and Manufacturing) were associated with within-person changes in WFE and FWE. This model was a good fit to the data: χ²₍₁₀₆₄₎ = 1838.13, p < .001, SRMR = .057, CFI = .918, TLI = .902, RMSEA = .037 (90% CI [.034, .040]).

The patterns of the regression paths and correlations were like the model without covariates. The variances of the random intercepts remained significant for both WFE (σ² = .08, p = .029) and FWE (σ² = .06, p = .010). All the within-person autoregressive and cross-lagged paths remained non-significant. However, the significant positive correlation between the stable traits of WFE and FWE changed to non-significant (r = .64, p = .089; see Figure 2).

Figure 2. Structural Model After Adding Covariates.

Note. Regression paths shown are standardized regression paths. Covariances shown are standardized latent correlations. Although not shown in the figure, the residual variances among the corresponding indicators over time were allowed to associate. Significant covariates included psychological well-being, manufacturing, and professional and related services (see Table 4). B-FWE = Between-person family-to-work enrichment; B-WFE = Between-person work-to-family enrichment; W-FWE = Within-person family-to-work enrichment; W-WFE = Within-person work-to-family enrichment ; T1 = Time 1; T2 = Time 2; T3 = Time 3.

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

When inspecting the covariate effects, previous levels of PWB predicted greater within-person increases in FWE (β = .43, p = .024 for Time 1 to Time 2; β = .40, p = .006 for Time 2 to Time 3). Similarly, the previous level of PWB predicted greater within-person increases in WFE from Time 2 to Time 3 (β = .28, p = .037), but did not reach statistical significance from Time 1 to Time 2 (β = .31, p = .074). Individuals working in a manufacturing industry at Time 1 experienced greater within-person increases in FWE at Time 2 (β = .13, p = .013), and those working in professional and related services at Time 2 experienced greater within-person increases in WFE at Time 3 (β = .15, p = .016). None of the personality traits were associated with within-person changes in both WFE and FWE (p > .05; see Table 4).

Table 4.

Covariate Effects on Endogenous Variables in the Final Structural Model (Figure 2).

Construct	T1
	PWB	N	E	O	Agr	Agy	Prof	Manu	Full-Time
W-FWE (T2)	.43*	.10	−.04	−.09	.11	.14	.09	.13*	−.07
W-WFE (T2)	.31	.14	−.06	.12	−.01	−.02	.14	.09	−.03
Construct	T2
	PWB	N	E	O	Agr	Agy	Prof	Manu	Full-Time
W-FWE (T3)	.40**	.01	.03	−.06	−.02	.03	.05	−.02	.01
W-WFE (T3)	.28*	−.06	.02	.10	−.10	.11	.15*	.07	−.01

Note. PWB = Psychological well-being; N = Neuroticism; E = Extraversion; O = Openness to experience; Agr = Agreeableness; Agy = Agency; Prof = Professional and related services; Manu = Manufacturing; W-FWE = Within-person family-to-work enrichment; W-WFE = Within-person work-to-family enrichment; T1 = Time 1; T2 = Time 2; T3 = Time 3.

^*p < .05,

^**p < .01.

Furthermore, to examine the association between the covariates and the stability of WFE and FWE, another model where the covariates were allowed to covary with the between-person levels of WFE and FWE was conducted (χ²₍₁₀₁₈₎ = 1746.52, p < .001, SRMR = .054, CFI = .923, TLI = .904, RMSEA = .037 (90% CI [.034, .039]). PWB, extraversion, openness to experience, and agency were positively associated with greater stability for both WFE and FWE (r = .15 to .61, p < .05). Agreeableness was only associated with greater stability for FWE (r = .22 to .29, p < .01) but not WFE (p > .05). Neuroticism was negatively associated with the stability of WFE (r = −.20, p = .022) and FWE (r = −.23 to −.19, p < .01). Regarding industry, working in professional and related services were associated with greater stability for WFE (r = .29 to .32, p < .001) and FWE (r = .17, p = .017), but working in manufacturing or full-time were not associated with the stability of WFE and FWE (p > .05; see Table 5).

Table 5.

Associations Between Covariates and Within-Person Levels FWE and WFE.

Construct	PWB		N		E		O		Agr		Agy		Prof		Manu		Full-Time (T1)
	T1	T2	T1	T2	T1	T2	T1	T2	T1	T2	T1	T2	T1	T2	T1	T2
B-FWE	.49***	.61***	−.19**	−.23**	.29***	.31***	.15*	.25***	.29***	.22**	.25***	.29***	.11	.17*	.13	−.05	−.02
B-WFE	.30**	.47***	−.06	−.20*	.21**	.29***	.40***	.46***	.09	.07	.24**	.36***	.29***	.32***	.08	.02	−.04

Note. PWB = Psychological well-being; N = Neuroticism; E = Extraversion; O = Openness to experience; Agr = Agreeableness; Agy Agency; Prof = Professional and related services; Manu = Manufacturing; B-FWE = Between-person family-to-work enrichment; B-WFE = Between-person work-to-family enrichment; T1 = Time 1; T2 = Time 2.

^*p < .05,

^**p < .01,

^***p < .001.

Discussion

Considering the profound influence of work and family on psychological health, work-family research has proliferated much work in the past several decades, spanning the disciplines of vocational psychology, organizational behavior, industrial and organizational psychology, and family science. Most research efforts on work-family enrichment focus on examining possible antecedents and consequences. Little attention has been given to the fundamental question about the longitudinal stability of work-family enrichment over and above personality and trait well-being. The implications of such a possibility could heighten our understanding of work-family enrichment, and how it might be leveraged by those in the caring professions like counseling psychology.

As hypothesized, work-to-family and family-to-work enrichment demonstrate substantial stability over time in the RI-CLPM. This trend was somewhat inconsistent with the findings from the repeated-measures ANOVAs, revealing a decrease in both work-to-family and family-to-work enrichment from Time 2 to Time 3. However, if an individual’s level of work-to-family and family-to-work enrichment changes at a different rate, the results from the repeated-measures ANOVA are known to be inaccurate (Kwok et al., 2008). As a result, in the current study, we discussed the results from the RI-CLPM. Like previous research spanning narrow periods of time like six months to one year (Babic et al., 2017; Cho et al., 2013; Hammer et al., 2005), results indicated strong evidence of stability in both work-to-family and family-to-work enrichment data, spanning 20 years of adulthood. Further, there was strong evidence of similarity in the stability of work-to-family and family-to-work enrichment among working women and men. Collectively these findings are consistent with early contentions that working women and men, and their families develop adaptive strategies to meet their work and family responsibilities (Moen & Wethington, 1992). Further, the idea that working adults and their families devise strategies to achieve work and family goals, likely within the constraints of dynamic work and family responsibilities across time, brings empirical support to the theoretical argument that success at work and in family life is a major task of adult development (Lachman & James, 1997).

Our results suggest that personality traits were not associated with within-person change for either work-to-family or family-to-work enrichment, but that psychological well-being was associated with within-person change. Specifically, previous levels of psychological well-being predicted greater within-person increases in family-to-work enrichment (from Time 1 to Time 2) and work-to-family enrichment (from Time 2 to Time 3). This result, if replicated, would suggest that enhancements to trait-based well-being has the potential to enable work-family enrichment, and potentially create a virtuous cycle wherein the psychological benefits of work-family enrichment (Hammer et al., 2005) accentuate trait well-being and subsequent experiences of work-family enrichment. Since most studies in this area only report between-person results (e.g., Hammer et al., 2005), our within-person results lend strong evidence that interventions that improve psychological well-being will also contribute to enhancements in work-family enrichment. Collectively our results contribute to emerging theories of work-family enrichment positing that personal resources like well-being are beneficial for enabling work-family enrichment (Greenhaus & Powell, 2006) and their subsequent ripple effects through the work and family systems (e.g., Grzywacz et al., 2007).

It is also worth noting that personality and well-being were associated with between-person stability of work-family enrichment. Like previous studies implicating personality in work-family enrichment (e.g., Michel et al., 2011), greater extraversion, openness to experience, and agency were associated with greater stability for both work-to-family and family-to-work enrichment, whereas greater neuroticism was associated with less stability of work-to-family and family-to-work enrichment. An additional contribution to the literature is our evidence indicating that trait-based well-being was also associated with stability in work-to-family and family-to-work enrichment.

One notable result of this study is the robust stability of work-to-family and family-to-work enrichment, even after controlling for personality and well-being. This finding has profound conceptual implications that have not been documented before. Foremost, the robust stability of work-to-family and family-to-work enrichment over time supports the fundamental conceptualization of work-family enrichment as a process that is primarily given shape by contextual resources and demands (Greenhaus & Powell, 2006). Second, the virtually imperceptible change in work-to-family and family-to-work enrichment stability coefficients, after adjusting for personality and well-being, offers guidance into the nature of documented linkages between personality traits with work-family enrichment (Michel et al., 2011). That is, the linkages likely reflect alternative forms of “causation” as opposed to simple co-occurrence, suggesting one is the manifestation of the other (e.g., Costa & McCrae, 2017; McAdams & Pals, 2006). Indeed, as some have argued (Grzywacz et al., 2007), some personality traits may enable differential ability to extract greater benefit from opportunities in the workplace or family to create work-family enrichment. This is an exciting area for future research. Finally, the results provide strong evidence that work-family enrichment, despite being a multistage process, can be measured using self-reported measures (e.g., Carlson et al., 2006) without apparent influence by personality, thereby providing additional evidence of measurement validity.

Occupational type (largely white collar “professional and related services” vs. blue collar “manufacturing work”) was differentially associated with changes in work-family enrichment. Employment in manufacturing was associated with greater within-person increases in family-to-work enrichment, while those working in professional and related services experienced greater within-person changes in work-to-family enrichment. These occupation-specific patterns are a unique contribution to the literature and are reminiscent of the “socialization of work” explanation for the recreation of social class. That is, adult experiences in the workplace fundamentally differ by social class, and those differential experiences in the workplace spillover into the home in terms of values and expectations parents pass on to children (Kohn, 1976). This is a critical area for future research.

Other findings are also noteworthy. First, it is interesting that work-to-family and family-to-work enrichment were significantly associated with each other across all three time points, but the within-person change of family-to-work enrichment did not predict a change in work-to-family enrichment, and vice versa. These results further support the position that work-to-family and family-to-work enrichment are not reciprocally associated and that they have their own antecedents and outcomes (Grzywacz & Butler, 2005; Hammer et al., 2005). Second, the finding of no significant gender differences in the structural model is noteworthy because of persisting views that gender plays a fundamental role in work-family experiences (e.g., Powell & Greenhaus, 2010). The findings supportive of gender differences are frequently obtained using convenience methods, leading some to contend that varying misrepresentation of women in research relative to the general population is the source of problem (Grzywacz et al., 2013). Our findings add to the small number of studies that have examined gender differences in work-family enrichment, using a general population sample, and demonstrated no gender differences in the stability of the work-family enrichment.

Limitations

The contributions of this study need to be considered in light of its limitations. Less than ideal estimates of reliability, both with several indicators of work-to-family enrichment and one personality trait (i.e., conscientiousness), must be acknowledged. The potential misfit in our analyzed models, where some of the model fit indices indicated marginal fit, may be due to the issues in internal consistency or model complexity in the RI-CLPM (Hamaker et al., 2015), increasing a chance in model mis-specification. However, recognizing that modest internal consistency will attenuate associations (Schmidt & Hunter, 1996), the observed stability estimates are more likely to be underestimated than overestimated.

Furthermore, although the RI-CLPM allowed us to differentiate within-person dynamics from between-person differences among concepts of interest, we were not able to disentangle the potential heterogeneity among these within-person changes. For instance, the within-person differences in work-to-family and family-to-work enrichment by occupation may differ by other between-person factors (e.g., race) or time-varying covariates (e.g., schedule flexibility). Such refined analyses await further research. Another weakness of this study is the lack of racial and ethnic diversity. Over 90% of study participants identified themselves as European American or White, thus pointing to the results’ potential lack of external validity to minority groups in the U.S. In general, there is a paucity of studies concerning ethnic and racial minorities in work-family studies and even fewer regarding work-family enrichment. Those work-family conflict studies with ethnic minority populations show that racial and ethnic minorities tend to report lower levels of work-family conflict (e.g., Grzywacz et al., 2002) and higher levels of work-to-family enrichment than Whites (Voydanoff, 2004). The literature suggests that one important reason may be because the conceptualization and value of family is likely different among racial/ethnic minority groups, and thus the boundary between work and family may be less salient among ethnic/racial minorities. For example, Grzywacz et al. (2005) found that many immigrants, rather than seeing work as a means for self-actualization, view work as means to enhancing family life. Another study by Grzywacz et al. (2007) reported that Latinx experience work-family conflict infrequently because work and family are conceptually integrated; they view work as a means to provide for their family. Future studies are warranted to examine these racial and ethnic differences in the conceptualization of the work-family interface. Furthermore, this study, like many, is guilty of conceptualizing gender as binary and heteronormative. Studies focusing on gender and sexual minorities (Kim et al., 2019) are imperative to addressing the critical social justice question of who has or does not have jobs that facilitate mental and physical health, a knowledge that psychologists are not yet equipped to answer (Whiston et al. 2012).

Implications for Practice, Advocacy, Education/Training, and Research

The results of this study have practical implications for psychologists and other mental health professionals, supporting working adults. The substantial stability of work-family enrichment over time, independent of personality and general well-being, should encourage clinicians to avoid the well-documented tendency to avoid career issues in psychotherapy (Juntunen, 2006; Schultheiss, 2006). Instead, clinicians could target work-family enrichment as a legitimate treatment focal point for pursuing improvements in clients’ psychological health. As an example, when working with clients attributing certain personality characteristics to difficulties managing work and family responsibilities, clinicians can empower clients to view enrichment between these two domains as a skill to be developed and dispel the myth that work-family enrichment is dependent upon having “the right attitude.” Instead, counseling psychologists could help clients recognize how mood, values, skills, and behaviors experienced in one domain can actually complement the other domain. Work-family enrichment could also be promoted by helping clients reframe the skills, behaviors, and experiences in either their work or personal lives to be adapted into the other domain, and thereby help them minimize dissonance between two domains. In the context of work-family interpersonal capitalization, sharing positive emotions with a spouse or partner by discussing positive work events or experiences can not only increase job satisfaction, but also relationship satisfaction (Ilies et al., 2011).

The results can also be used in developing interventions to promote work-family enrichment. Evidence-based approaches to targeting work-family experiences predominantly focus on reducing work-family conflict and the available interventions include supervisor training, self-monitoring (Hammer et al., 2011), and mindfulness-based training intervention as a cognitive emotional segmentation strategy (Michel et al., 2014). Few interventions exist in specifically targeting ways to increase work-family enrichment, rather than reducing conflicts. Our results, demonstrating the stability of work-family enrichment over a 20-year period, illustrates the robust stability people have with regards to experiencing work-family enrichment across adulthood. Because of this stability above and beyond personality traits and partially psychological well-being, it is even more critical for clinicians to advocate for the importance of work-family enrichment early on as a prevention point for young adults in managing multiple roles. For example, university counseling center psychologists can provide psychoeducation, as well as create training modules for mental health trainees on how skills and values learned in school can be used in their work or home domains and vice versa. Training like this will help young adults recognize, experience, and foster work-family enrichment as they enter the workforce and move into later adulthood.

Findings from the current study open several interesting future research avenues. Given that the current study examined more internal factors, future research can potentially focus on the role of environmental factors in the experience and stability of work-family enrichment. It is possible that environmental factors may be antecedents, with certain kinds of environments (within family and work domains) enabling greater work-family enrichment (Carlson et al., 2006). In fact, certain environmental factors, in addition to individual factors, may be tied to the use, transfer, and movement of resources from one domain to another (Carlson et al., 2006).

In sum, our research is the first study to examine the longitudinal stability of work-family enrichment over and above well-being and personality traits, investigating both within and between person changes. We found that, even after controlling for previous levels of psychological well-being and personality, work-to-family and family-to-work enrichment has substantial stability across a 20-year period for both women and men. The stability of work-family enrichment over time is consistent with its conceptual nature of being a process, and the theoretical view that individuals and families execute strategies to achieve success in the work and family domains as a core task of adult development. Further, the results clearly convey that work-family enrichment is not simply a narrow manifestation of broader enduring traits like personality or trait well-being. In light of these findings, counseling psychologists are encouraged to develop and refine clinical strategies that help working adults experience greater work-family enrichment, and potentially better mental and physical health.

Significance of the Scholarship to the Public:

The findings clearly convey that work-family enrichment is not simply a manifestation of specific personality traits or an optimistic worldview. In light of these findings, clinicians are encouraged to target work-family enrichment as a legitimate treatment focal point for pursuing improvements in clients’ psychological health