Crossover Of Resources And Well-Being Within Employee-Partner Dyads .

Lee et al.Page 4Author ManuscriptAuthor ManuscriptPartners’ perceptions about employees’ increased work schedule flexibility to handle familyresponsibilities may provide evidence that increases in one family member’s schedulecontrol can permeate into the family domain, as the W-HR model suggests (tenBrummelhuis & Bakker, 2012). At a very basic level, an employee who received aworkplace intervention may experience increased schedule control; this additional resourcemay be evident in the partner’s reports of the employee’s work schedule flexibility forfamily roles through cognitive empathy that involves understanding another person’ssituations (Bakker et al., 2009; Duan & Hill, 1996; Gladstein, 1983; Westman & Hamilton,2004). Furthermore, assessing partners’ perceptions about family time adequacy mayprovide evidence of crossover of resources such that an employee’s additional resource (i.e.,increased schedule control) can increase the partner’s perceived temporal resources forfamily and role performance. The first aim of this study is to examine the effects of aworkplace intervention on partner perceptions of employees’ work schedule flexibility forfamily and whether the effect is mediated by employee-reports of their increased schedulecontrol. We also examine the effect of a workplace intervention on partner perceptions offamily time adequacy and whether the effect is mediated by employees’ schedule control.These crossover of resources between employees and partners are depicted in Figure 1.Workplace Intervention Effect on Partner Perceived StressAuthor ManuscriptMoving beyond testing the link between employee resources and partner perceptions offamily resources, we also test whether partners’ well-being (measured by perceived stress) isaffected by employees’ increased schedule control and well-being. The W-HR model (tenBrummelhuis & Bakker, 2012) suggests that contextual resources (e.g., workplace support)can increase individual resources, which, in turn, influences home outcomes, includinghome-related happiness. Yet whether this benefit extends beyond the employee and to theemployee’s partner is, as of yet, unknown.Specifically, increased schedule control from a workplace intervention may decreaseemployees’ perceived stress. This within-person process explaining how employees’experiences in the work domain influence their well-being in the family and personaldomains is called spillover (Grzywacz & Marks, 2000). Crossover process involves anotherperson, such that increased schedule control from a workplace intervention may decrease theemployees’ perceived stress (spillover), which may, in turn, decrease their partner’sperceived stress (crossover of well-being).Author ManuscriptStudies have found evidence for crossover effects within couples. For example, duringmilitary downsizing in Russian army, marital dissatisfaction crossed over from the maleofficers to their wives (Westman & Hamilton, 2004). Most crossover studies have focused onnegative crossover rather than positive, such as crossover of job insecurity (Debus & Unger,2017) and crossover of work-to-family conflict (Hammer, Allen, & Grigsby, 1997; Pedersen,2014). A rare example of positive crossover is a study by Demerouti and colleagues (2012).They found that, when employees experienced more social support at work, their partnersreported more home resources, including home autonomy, social support, and time todevelop strengths. Building on these previous studies, we test positive crossover of well-Community Work Fam. Author manuscript; available in PMC 2020 September 25.

Lee et al.Page 5Author Manuscriptbeing effect, whether one partner’s additional work resources predict decreases in his/herperceived stress, which, in turn, predict decreases in the partner’s perceived stress (Figure 1).Potential Moderation by Contextual FactorsAuthor ManuscriptPrevious research reports that the extent of work-family conflict and time demands differ bygender and household characteristics, which suggests potential moderation by these factorsin crossover of work-family resources and well-being within couples. Working women tendto experience more work-family conflict than working men (Jacobs & Gerson, 2004;Maume, Sebastian, & Bardo, 2010); this may be associated with the couples’ increased needto pool resources together. Women also tend to exert more influence on their spouse thanmen. For example, Westman and Etzion (2005) found that wives’ social support buffered theassociation between family stress and work-family conflict for the husbands. Moreover,partner employment status (dual-earner vs. single-earner households) and childcareresponsibilities (living with children in the household) may be important factors contributingto how couples experience work-family conflict and maximize their resources and wellbeing across work and family domains (Winslow, 2005). In a study examining the effect of aworkplace intervention on employees’ perceived stress, the intervention was more beneficialin decreasing perceived stress for those with child care responsibilities, compared to thosewithout (Kossek et al., 2017). These studies motivate us to explore whether and howcrossover of resources and well-being between employees and their partners differ by theirgender, partner’s employment status, and living with children in the household.Present StudyAuthor ManuscriptWe draw upon data from a randomized workplace intervention, named “Support-TransformAchieve-Results (STAR),” implemented as part of the Work, Family & Health Study(WHFS; Bray et al., 2013; Kelly et al., 2014). The STAR workplace intervention wasdesigned to increase employees’ schedule control and supervisor support for family life –two workplace resources that the W-HR model (ten Brummelhuis & Bakker, 2012) argueshave implications for family resources and well-being. In this paper, we first examined theeffects of the STAR workplace intervention on two indicators of partner-reported familyresources employees’ work schedule flexibility to handle family responsibilities andfamily time adequacy. Our first set of hypotheses testing the crossover of resources inemployee-partner dyads was:Author ManuscriptHypothesis 1: The STAR intervention will increase partner perceptions ofemployees’ work schedule flexibility for family at the 12-month follow-up, throughemployees’ increased schedule control at the 6-month follow-up.Hypothesis 2: The STAR intervention will increase partner perceptions of familytime adequacy at the 12-month follow-up, through employees’ increased schedulecontrol at the 6-month follow-up.Second, we tested the crossover of well-being within employee-partner dyads, whether onepartner’s additional work resources predicted decreases in his/her perceived stress, which, inCommunity Work Fam. Author manuscript; available in PMC 2020 September 25.

Lee et al.Page 6Author Manuscriptturn, predicted decreases in the partner’s perceived stress. Our last hypothesis involving thetwo step paths was:Hypothesis 3: The STAR intervention will decrease partner’s perceived stress at the12-month follow-up. This effect will be explained by the effect of employees’increased schedule control on their perceived stress at 6 months (spillover; H3a),which will be carried over to employees’ perceived stress at 12 months that linkedto the partner’s perceived stress at 12 months (crossover of well-being; H3b).Findings from this study will contribute to translational research in the field of work andfamily by showing how a flexible work practice adopted by an industry sector can improvethe employees’ and their partners’ well-being.MethodAuthor ManuscriptParticipantsAuthor ManuscriptAuthor ManuscriptData came from the Work, Family, and Health Study (WFHS), a field trial testing the effectsof a workplace intervention on employees’, families’, and organizations’ well-being (Bray etal., 2013; Kelly et al., 2014). Researchers partnered with the information technology (IT)division of a U.S. company. Worksite recruitment was done in conjunction with the IndustryAdvisory Board. Selection criteria included adequate support from local management,worksite size and geographic proximity to other study sites, and ability to supportlogistically the intervention delivery and data collection. A total of 56 work teams all overthe country were identified and randomized to either the STAR or the control/Usual Practice(UP) condition (Kelly et al., 2014). Employees were eligible to participate if they wereregular employees (not contractors) located in the worksites where data collection occurred.Of 823 employees completing the baseline interview, 653 employees were married or livedwith a permanent romantic partner. All spouses or cohabiting partners of these employees(for at least one year) were eligible for study participation and invited to participate in thestudy at baseline and 12 months later. Of the 653 partnered employees, 455 employees’spouses/partners participated in the baseline survey. The 455 employees did not significantlydiffer from 198 (653–455) employees who were partnered but their partner did notparticipate in the surveys, in terms of employee age, gender, race, education, work hours,intervention condition, and baseline schedule control and perceived stress. Of the 455partners who provided baseline data, 334 partners were re-assessed at 12 months later. The334 partners also did not significantly differ from the attriters (n 121) in baselinecharacteristics, including age, education, employment status, intervention condition,perceptions of employees’ work schedule flexibility, family time adequacy, and stress,except for race; attriters were more likely to identify as non-white than white, non-hispanic.Among the 334 partners’ paired employees, 327 provided data both at 6 months and 12months. Thus the final analytic sample of the current study consisted of 327 employeepartner dyads, including 13 same sex couples. Their demographic information is provided inTable 1.Community Work Fam. Author manuscript; available in PMC 2020 September 25.

Lee et al.Page 7ProceduresAuthor ManuscriptTrained interviewers obtained informed consent and then conducted computer-assistedpersonal interviews with the employees at the workplace at baseline and again at the 6month follow-up. Partners were recruited to participate in a telephone interview throughcontact information provided by the employee, and through recruitment communicationgiven to the employee to provide to the spouse or partner directly. Partners who completedthe baseline interview and were still living with the employee at 12-months were asked toparticipate in a 12-month partner interview by telephone. The workplace interviews withemployees lasted on average 58 minutes per assessment, and the telephone interview timewith partners averaged about 32 minutes per assessment. Employees and their partners eachreceived 20 at baseline and additional 20 at each follow-up for their participation.The STAR Workplace InterventionAuthor ManuscriptAuthor ManuscriptFollowing baseline interviews, a randomized workplace intervention (“STAR”), designed toincrease employees’ schedule control and supervisor support for family and personal life(Bray et al., 2013), was implemented (n 172 intervention, n 155 control). The interventionconsisted of a 3-month structural and cultural change process, including two types of mainactivities. The first was facilitator-led sessions for all employees in the interventioncondition (with their managers present) to identify new work practices to help transitionfrom rigid work schedules to giving employees more control over when and where theywork. Examples include cross-training to provide back-ups, possibility of remote work, andself-scheduling. The second was training managers/supervisors in the intervention conditionto demonstrate support for employees’ personal and family lives while also supportingemployees’ job performance. Specifically, managers/supervisors completed computer-basedtraining that taught them about (1) personal and business reasons for reducing work-lifeconflicts, (2) top management support for this initiative, and (3) example behaviors andstrategies that demonstrate both professional and personal support to their employees.Participatory sessions for employees lasted eight hours, and managers attended an additionalfour hours of training (for details on STAR procedures see Kelly et al., 2014; Kossek et al.,2014). All procedures were conducted in accordance with established ethical guidelines, andapproved by appropriate Institutional Review Boards.Author ManuscriptThe STAR intervention has proven to be effective in increasing employees’ schedule controlas the intervention intended (Kelly et al., 2014). Previous research has also reported that theSTAR intervention significantly increased employees’ daily time spent with children, dailytime adequacy for personal life, and nightly sleep duration measured by actigraphy (Davis etal., 2015; Lee et al., 2016; Lee, McHale, Crouter, Kelly, et al., 2017; Olson et al., 2015).These intervention effects were found across different work sites with lack of variance at thework site level (Almeida et al., 2018; Lee et al., 2016; Lee, McHale, Crouter, Kelly, et al.,2017). Consistency across different work sites was assured by using a facilitator’s guide andsemi-structured scripts for the facilitator-led sessions. Moreover, the STAR interventionutilized virtual training sessions and the company intranet, because many work groups werenot co-located in the same building. This allowed employees working in different locationsto participate in the same intervention conversations with their managers and team members.Note that the WFHS includes another sample of employees in extended-care industry whoCommunity Work Fam. Author manuscript; available in PMC 2020 September 25.

Lee et al.Page 8Author Manuscriptprovide direct care to older residents in nursing homes. The STAR intervention was alsoimplemented in the extended-care settings with industry-specific customization, but therewas lack of intervention effects, potentially due to different nature of work (see Kelly et al.,2014; Marino et al., 2016, for more discussions).MeasuresAuthor ManuscriptEmployee-reported schedule control.—At baseline and the 6-month post-interventionfollow-up, employees reported how much choice they had over eight dimensions of controlover work schedule (Thomas & Ganster, 1995). The items are, “How much choice do youhave over (1) when you take vacations or days off, (2) when you can take off a few hours,(3) when you begin and end each work day, (4) the total number of hours you work eachweek, (5) doing some of your work at home or at another location, instead of at theworkplace, (6) the number of personal phone calls you make or receive while you work, (7)the amount or times you take work home with you, and (8) shifting to a part-time schedule(or full-time if currently part-time) while remaining in your current position if you wanted todo so?” Responses ranged from 1 (very little) to 5 (very much). We used the mean of the 8items for the schedule control scale where employees responded to at least 6 items out of the8 items. The Cronbach’s α of schedule control was .78 and .84, at baseline and 6 months,respectively.Author ManuscriptEmployee-reported perceived stress.—At baseline and the 6-month follow-up,employees responded to 4 items from the Perceived Stress Scale (PSS; Cohen, Kamarck, &Mermelstein, 1983). PSS is the most widely used scale of stress appraisals and has beenfound to be predictive of many adverse physical and mental health outcomes. The itemsread, “During the past 30 days, how often have you felt (1) that you were unable to controlthe important things in your life, (2) confident about your ability to handle your personalproblems, (3) that things were going your way, and (4) difficulties were piling up so highthat you could not overcome them? Responses were coded as 1 (never) to 5 (very often). Weused the mean of the 4 items where partners responded to at least 3 items. The Cronbach’s αof employees’ perceived stress was .74, .75, and .77 at baseline, 6 months, and 12 months,respectively.Author ManuscriptPartner-reported employees’ work schedule flexibility for family.—Partnersreported the level of flexibility the employee had in work schedule to handle familyresponsibilities at baseline and 12 months. Based on Neal and Hammer’s (2007) spousalcrossover work, the following item was asked: “How much flexibility does your spouse/partner have in his/her work schedule to handle family responsibilities? Responses rangedfrom 1 (no flexibility at all) to 4 (a lot of flexibility). This item was validated in Hammer etal.’s (1997) study of bank employees and their spouses/partners.Partner-reported family time adequacy.—Partner perceived family time adequacy wasmeasured by 2 items, which were derived from part of the larger Family Resource scalerevised (Van Horn, Bellis, & Snyder, 2001) for family roles. The items were: “To whatextent is there enough time (1) to be with your spouse/partner and (2) to care for the needs ofother family members?” Responses were coded as 1 (never) to 5 (all of the time). We usedCommunity Work Fam. Author manuscript; available in PMC 2020 September 25.

Lee et al.Page 9Author Manuscriptthe mean of the 2 items where partners responded to both items. The Cronbach’s α betweenthe 2 items were .52 and .50 at baseline and 12 months, respectively. Family time adequacyquestions have been used in several studies separately for each domain (e.g., time adequacyto be with partner) or as a scale (DePasquale et al., 2017; Soomi. Lee et al., 2015; SoomiLee, McHale, Crouter, Kelly, et al., 2017).Partner-reported perceived stress.—We used the 4 items from the Perceived StressScale (PSS; Cohen, Kamarck, & Mermelstein, 1983), the same items used to assessemployees’ perceived stress. Responses were coded as 1 (never) to 5 (very often). We alsoused the mean of the 4 items (allowed 1 item missing), such that higher scores reflectedhigher perceived stress. The Cronbach’s α of partners’ perceived stress was .73 and .79 atbaseline and 12 months, respectively.Author ManuscriptCovariates.—We considered employees’ and partners’ background characteristics aspotential covariates that may influence their perceptions about resources and stress (Almeidaet al., 2018; Soomi Lee, McHale, Crouter, Kelly, et al., 2017). Those were employees’ andpartners’ gender (0 women, 1 men), age (in years), race (0 white, 1 non-white), andeducational level (0 college graduate or more, 1 less than college graduate). We alsoconsidered marital status (0 married, 1 cohabiting) and number of children (coupleswithout a child had 0 value). In addition, we included employees’ work hours and partners’employed status (0 employed, 1 not employed) at baseline. All continuous covariates werecentered at the sample mean.Analytic StrategyAuthor ManuscriptWe estimated the models via Structural Equation Modeling (SEM) using MPlus Version 7.2.We used maximum likelihood estimation with robust standard errors. Model goodness-of-fitwas evaluated using Hu and Bentler’s (1999) criteria: (a) nonsignificance in the result of chisquare test; (b) the root mean square error of approximation (RMSEA) no more than .06; (c)the root mean squared residual (SRMR) no more than .08; and (d) the comparative fit index(CFI) and the Tucker-Lewis index (TLI) closer to .95 or greater.Author ManuscriptTo test whether the workplace intervention predicted changes in partner perceptions offamily resources and stress through changes in employees’ schedule control and stress, weused residualized gain approach. This approach can take into account differences in changedue to baseline levels

The Work-Home Resources (W-HR) model suggests the interconnectedness of resources across work and family domains, such that resources obtained from the work domain may also increase resources in the family domain (ten Brummelhuis & Bakker, 2012). Specifically, the W-HR model asserts that contextual resources, such as family-supportive