Daily Patterns of Stress and Conflict in Couples: Associations with Marital Aggression and Family-of-Origin Aggression

Abstract

For many married individuals, the ups and downs of daily life are connected such that stressors impacting one person also impact the other person. For example, stress experienced by one individual may “spill over” to negatively impact marital functioning. This study used both partners’ daily diary data to examine same-day and cross-day links between stress and marital conflict and tested several factors that make couples vulnerable to spillover. Assessment of 25 wide-ranging sources of daily stress included both paid and unpaid work, health issues, financial concerns, and having to make difficult decisions. Results showed that both husbands and wives’ experience of total daily stress were associated with greater same-day marital conflict and that conflict was greater on days both spouses experienced high levels of stress. Evidence of cross-day spillover was found only in those couples with high concurrent marital aggression and in couples where wives reported high family-of-origin aggression. These results highlight both the common, anticipated nature of same-day spillover and the potentially problematic aspects of more prolonged patterns representing failure to recover from stressors that occurred the previous day. The discussion focuses on how reactivity in one life domain puts that individual at risk for generating stress in another life domain and how current marital aggression and family-of-origin aggression are associated with difficulty recovering from stressful events.

Families and couples make up interconnected systems where events affecting one person can reverberate across the system and impact other family members (Cox & Paley, 1997). Such interdependence among people in close relationships can be both beneficial and detrimental to individual functioning. For example, social bonds appear to buffer stress responding in humans and animals (Cohen & Wills, 1985; Hostinar, Sullivan, & Gunnar, 2014) and protect against the development of various mental and physical health problems (Uchino, 2006; Robles, Slatcher, Trombello, & McGuinn, 2014). At the same time, this connectedness may put couples at risk if stress experienced by one individual spills over and adversely affects marital functioning. This process, known as “spillover,” refers to the transmission of problems across domains (Almeida, Wethington, & Chandler, 1999; Bolger, DeLongis, Kessler, & Wethington, 1989; Repetti, 1994). Most commonly, studies on spillover in couples have investigated how stress at work relates to marital behavior at home the same evening. In contrast, the current study examines same-day and bidirectional, cross-day links between stress and conflict and assesses spillover originating in a variety of life domains. We also test the interactive effect of wives and husbands’ daily stress to determine if high levels of stress in both spouses overload the system, resulting in heightened conflict. Moreover, it is unclear why some couples are more susceptible to spillover than others. To address this, the current study tests marital aggression and family-of-origin aggression (FOA) as risk-factors for daily spillover between stress and conflict.

Stress-Conflict Spillover in Daily Life

Stressors are common occurrences in married couples’ lives where couples must deal with various day-to-day hassles or problems, such as the car breaking down, traffic, or looming work deadlines. Chronic stress activation is thought to negatively impact physical and mental health (e.g., Juster, McEwen, & Lupien, 2010), with models of stress typically emphasizing its toll on the individual. However, stress is a highly dyadic process (Randall & Bodenmann, 2009), with connections between stress and relationship functioning unfolding in complex ways. For example, spouses can bring home stress experienced in outside domains (e.g., at work) to “share” with their partners (e.g., Repetti, 1989). Alternatively, couples can experience the same stressful event (e.g., child getting in trouble), which can have individual effects on each spouse. Marital problems can also spill outward to affect functioning in other areas (e.g., being irritable at work). Understanding the complex and dynamic links between individual and dyadic functioning can provide important information about how stress responses evolve and contribute to relationship problems or impaired individual functioning.

A number of studies have used diary data to document links between stress and marital functioning in daily life (e.g., Ben-Ari & Lavee, 2011; Bolger et al., 1989; Buck & Neff, 2012; Chan & Margolin, 1994; Doumas & Margolin, 2003; Heller & Watson, 2005; Repetti, 1989; Sanz-Vergel, Rodriguez-Muñoz, & Nielson, 2015; Roper & Yorgason, 2009; Schulz, Cowan, Cowan, & Brennan, 2004; Story & Repetti, 2006; Totenhagen, Curran, & Butler, 2012). As articulated elsewhere (e.g., Laurenceau & Bolger, 2005), daily data have several advantages compared to more traditional designs. Daily data increase ecological validity because they allow researchers to capture spillover as it naturally unfolds. Moreover, by investigating day-to-day associations between different events or behaviors, it is possible to understand how maladaptive relationship processes develop and are sustained across time. Early spillover research showed that husbands’ stressful workdays link to husbands’ withdrawal (Repetti, 1989). In couples with dual-earners, work stress has been associated with combinations of anger, negative mood and withdrawal the same evening (Schulz et al., 2004; Story & Repetti, 2006). Evidence of spillover across days, which is far less common, shows that stress at home is linked to stress experienced on the next workday (e.g., Bolger et al., 1989; Chan & Margolin, 1994).

Thus far, our understanding of stress spillover has been largely informed through the examination of associations between work and home (e.g., Heller & Watson, 2005; Sanz-Vergel et al, 2015), although there also has been attention to certain types of stressors associated with specific life circumstances (e.g., activity limitations due to illness, Roper & Yorgason, 2009; security-political concerns, Ben-Are-& Lavee, 2011). Only a few studies have assessed multiple stressors, such as family demands or financial difficulties (e.g., Bolger et al., 1989; Buck & Neff, 2012). In line with more general stress research (Cohen, Kessler & Gordon, 1995), our goal is to include multiple dimensions of stress originating in a variety of life domains. We also broaden our conceptualization of spillover to include stressors occurring in spouses’ joint lives, which may have individual effects on each spouse. Stressors occurring in the home domain may be especially relevant to marital functioning because of their proximity to home life and because of the possibility for spouses to unite under stress or, conversely, to blame and criticize each other.

Though extant research has demonstrated clear connections between stress and marital functioning in daily life, important questions still remain. In particular, more information is needed on the systemic effects of both spouses’ stress experiences. To date, studies on daily stress spillover in couples have typically focused on two processes: within-person associations across settings (spillover; e.g., wife experience at work and wife behavior at home) or cross-person associations across settings (crossover; e.g., wife work experience and husband behavior at home). Moving beyond this framework, this study tests how marital functioning is affected when one versus two spouses have especially stressful days. Does the relationship absorb the stress of one spouse but reach a tipping point when both spouses are stressed? The interactive effect of husband and wife stress is tested to determine if conflict is heightened on days when both partners contribute strain to the marital system.

Risk Factors for Daily Stress-Conflict Spillover

Another important next-step is to determine what factors make the boundaries between individual functioning and dyadic functioning more or less permeable. In particular, certain ongoing or prior relationship experiences might increase the risk for experiencing stress spillover. Past research indicates that distressed couples have greater connectivity in their daily negative experiences. For example, stress spillover is stronger in high-conflict couples and those with less marital satisfaction (e.g., Story & Repetti, 2006). Another important risk factor for stress spillover might be aggression; both marital aggression and FOA have been linked to behavioral, emotional, and physiological regulatory problems. In particular, aggressive people or those exposed to FOA may be sensitized to stress or may interpret events in a more negative fashion (Cummings, El-Sheikh, Kouros, & Buckhalt, 2009; Davies, Sturge-Apple, Cicchetti, Manning, & Zale, 2009; Ford, 2005; Roberton, Daffurn, & Bucks, 2011). Such individuals may have more difficulty compartmentalizing or turning off responses to stress and may experience prolonged or heightened stress reactions that set the stage for spillover. Specifically, spillover may occur when spouses have difficulty recovering from stressful events or when high reactivity in one life domain puts that person at risk for generating stress in another domain. Here, we test the idea that stress spillover is characteristic of relationships in which there is high aggression or when spouses were exposed to high aggression in their family-of-origin.

Present Study

The present study uses diary methodology to investigate same-day and cross-day links between stress and couple conflict in married dyads. With work-home distinctions increasingly blurred in modern life and with females and males often experiencing a different balance between paid versus unpaid work, the present study accounts for wide-ranging stressors that adults encounter on a daily basis that can affect and be affected by relationship processes. As exploratory analyses, we first test how 25 different daily stressors (e.g., problems associated with paid and unpaid work, financial pressures, or stress emanating from other family members) are linked to marital conflict. We then test links between cumulative daily stress and same-day marital conflict. Beyond testing same-day associations, we also test links across days. By testing spillover from one day to the next, we can investigate directional effects and also ascertain if spillover processes sustain or dissipate after a night's rest. We also examine the interaction between wife and husband stress to determine if marital conflict is greater on days when both spouses are highly stressed, perhaps by overwhelming the supportive capacity of the marital system. Finally, we test factors that make couples vulnerable to spillover by testing the proximal factor of marital aggression and the more distal factor of FOA as moderators of same- and cross-day links between stress and conflict.

First, we hypothesize that stress will be associated with heightened marital conflict on the same day and that stress and marital conflict will be bidirectionally linked across days, with stress on one day predicting next-day conflict and vice versa (Hypothesis 1). Second, we hypothesize that marital conflict will be intensified when spouses experience high levels of stress on the same day (Hypothesis 2). Finally, we hypothesize that greater marital aggression (Hypothesis 3) and greater FOA (Hypothesis 4) will be associated with heightened spillover on the same day and bidirectionally across days.

Method

Overview

Data for the current project are part of a larger longitudinal study investigating family aggression and developmental outcomes in adolescence and young adulthood. The first cohort (N = 119 families) was recruited using word-of-mouth, flyers, and other advertisements posted in the community. To participate, parents were required to have been living together for at least 3 years, to have a child age 9 to 10, and to be able to complete study measures in English (see Margolin, Vickerman, Oliver, & Gordis, 2010 for details). A second cohort (N = 70 families) was recruited using similar procedures 4 years later at the third wave; youth in this cohort were 13 to 14 years old to be comparable in age to the original sample. At the first laboratory visit for each cohort, husbands and wives provided retrospective reports of exposure to FOA before the age of 18. Reports of marital aggression over the past year were provided at the wave-4 in-lab visit (time point 4 for cohort 1 and time point 2 for cohort 2). Diary data were collected at home following the wave-4 visit. The IRB approved all procedures; couples provided consent prior to their participation for each wave, and families were compensated $10 a day for participating.

Participants

Participants include 114 couples who provided daily reports of their stress and marital functioning over a 2-week period following a laboratory visit. Across both cohorts, 169 families were invited to participate in the wave four procedures (20 families had moved out of the area or could not be reached), and 140 families provided at least partial data at this wave (e.g., filling out online questionnaires only due to scheduling difficulties). Of these 140 families, 126 participated fully in the in-lab procedures and thus were eligible to provide home data; in total, 125 wives and 120 husbands completed diary questionnaires. To be included in the present study, couples were required to be living together and to have provided at least one day of overlapping data, resulting in 114 couples included in the present study. We conducted tests of attrition to determine if there were any differences between the original, complete sample (N = 189) and those included in the current study in terms of family income, education, age, ethnicity, years married, and FOA; no significant differences were found (all p-values > .25).

Wives’ ages ranged from 33.4 to 59.0 years old (M = 45.7; SD = 6.0), and husbands’ ages ranged from 32.8 to 71.9 years old (M = 48.3; SD = 7.0). Years of education ranged from 7 to 20 for wives and from 8 to 20 for husbands (for wives: M = 14.7; SD = 2.4; for husbands: M = 15.0; SD = 2.6). Couples had been married for 15.9 years on average (SD = 6.1). In total, 36.6% of wives identified as Caucasian, 21.5% Hispanic/Latino, 19.4% African American, 11.8% Asian American, and 9.7 multi-ethnic. For husbands, 39.8% identified as Caucasian, 21.5% African American, 19.4% Hispanic/Latino, 10.8% Asian American, 7.5 multi-ethnic, and 1.1 other. Family income (Mdn = $95,000) was consistent with the cost of living in the urban recruitment area (U.S. Census Bureau, 2010); 19.8% earned ≤ $50,000, 35.6% ≥ $51,000 and ≤ $100,000, 27.8% ≥ $101,000 and ≤ $150,000, and 16.8% earned ≥ $151,000. Among husbands, 82.5% of worked full time, 3.5% worked part time, and 14% were not employed. Among wives, 47.4% worked full time, 25.4% worked part time, and 27.2% were not employed.

Home Data Collection Procedures and Compliance

All families that completed the wave-4 in-lab portion of data collection were invited to provide daily diary data for the 2-week period following the laboratory visit. Husbands and wives completed the first report at the laboratory visit (reporting on the previous day) and then completed a questionnaire for the day of the laboratory visit that night. Couples had the choice of completing the diary assessments on paper (n = 66 for mothers; n = 50 for husbands) or online. For those using the online system, questionnaires were emailed to the participants at 5:00 pm on the evening of reporting and responses were uploaded to an online server. Couples completing paper questionnaires were given addressed envelopes and instructed to mail the questionnaires each evening. Electronic time stamps and postage stamps were used to assess compliance. On average, wives completed 12.7 (SD = 2.9) days of data, and husbands completed 12.4 (SD 3.2) days (Mdn = 14 days for both wives and husbands). Out of 1,596 possible days (14 days × 114 couples), wives reported on 1,448 (90.7%) days, and husbands reported on 1,416 (88.7%) days. For the daily records that were completed, wives completed 95.0% within 24 hrs of the day of reporting (97.2 % in 48 hrs); husbands completed 94.1 % within 24 hrs (98.1% in 48 hrs). Wives and husbands completed questionnaires on the same day for 92.9% of the recorded days.

Measures

Daily stress

Husbands and wives each reported on 25 items per day to evaluate wide-ranging sources of stress. Examples of items include: making mistakes doing paid work; feeling overloaded from unpaid work; issues of safety for you and your family; health problems of someone in family; argument with child; heavy traffic or bad drivers; appliances or the car not working. The internal consistency of all daily measures were computed using equations from Shrout and Lane (2012) that account for the multilevel structure of the data. Internal consistency was estimated to .77 for husbands and .77 for wives. For each source of stress, participants rated the extent to which they “felt concerned about...” using response options ranging from 0 (not at all) to 3 (a lot). Scores across the items were averaged with higher values indicating higher levels of stress. The full listing of items appears in Table 1. The five most commonly endorsed sources of stress that were endorsed for at least one day by both husbands and wives were: having too many things to do; finances; having to make difficult decisions; traffic or bad drivers; and not having enough money. The results also showed several differences between husbands and wives in their mean level of endorsement of specific stressors, with wives more frequently endorsing: having too many things to do; health of family members; feeling pressure from unpaid work; overloaded with unpaid work; being criticized doing unpaid work; and having a child argue with me. In contrast, husbands had higher endorsement of one item: conflict with coworker doing paid work. We also calculated an average stress score for paid work (5 items; e.g., conflict with coworker during paid work), unpaid work (5 items; e.g., pressure from unpaid work), and other (15 items; e.g., finances); results showed one significant difference: wives had higher endorsement of unpaid work (M = .30) than husbands (M = .19), t(113) = 3.06, p < .01.

Table 1.

Descriptive Statistics for Items Included in the Daily Stress Measure

	Wives				Husbands
	M (SD)	% Days	% People	r	M (SD)	% Days	% People	r
Finances	1.00 (1.01)	58.8	89.5	.03	0.99 (1.02)	59.9	88.6	.07
Too many things to do	1.07 (1.04)	62.4	93.9	.09*	0.86 (0.92)	56.2	93.0	.06
Difficult decisions	0.56 (0.88)	35.0	80.7	.14*	0.50 (0.79)	34.0	82.5	.11
Own health	0.46 (0.78)	31.7	66.7	.11*	0.39 (0.70)	28.0	64.0	.05
Health family members†	0.51 (0.86)	31.7	71.9	.17*	0.40 (0.76)	27.9	65.8	−.02
Victimized by crime	0.08 (0.37)	5.6	31.6	.10	0.07 (0.35)	4.8	21.9	.14
Traffic or bad drivers†	0.38 (0.69)	28.0	80.7	.13*	0.45 (0.75)	31.1	78.1	.03
Noise	0.26 (0.57)	20.2	63.2	.09*	0.28 (0.58)	21.4	57.9	.03
Safety	0.18 (0.53)	12.2	50.9	.06	0.18 (0.53)	13.1	47.4	.09
Appliances or car not working	0.20 (0.58)	13.5	42.1	.07	0.24 (0.67)	14.4	45.6	.04
Not finding something you need	0.28 (0.65)	18.7	57.9	.09	0.19 (0.54)	14.3	53.5	.06
Not having enough money	0.81 (1.00)	49.1	77.2	.05	0.83 (0.98)	50.8	79.8	.02
Pressure – paid work	0.78 (0.97)	47.8	64.2	.09	0.77 (0.96)	47.8	77.6	.01
Overloaded – paid work	0.92 (1.07)	50.2	66.3	.07	0.86 (1.02)	50.1	76.6	.01
Making mistakes – paid work	0.21 (0.52)	16.5	37.9	.13*	0.19 (0.51)	14.8	48.6	.05
Criticized – paid work	0.13 (0.42)	9.9	29.5	.07	0.19 (0.55)	13.9	39.3	−.01
Conflict coworker – paid work	0.10 (0.36)	7.8	30.1	.03	0.20 (0.55)	14.9	42.1	.02
Pressure – unpaid work	0.45 (0.75)	31.3	65.0	.17*	0.30 (0.62)	23.6	55.7	.10
Overloaded – unpaid work	0.56 (0.81)	38.4	72.8	.15*	0.36 (0.69)	26.5	55.7	.09
Making mistakes – unpaid work	0.12 (0.39)	9.7	30.1	.09	0.10 (0.38)	8.1	28.3	.21
Criticized – unpaid work	0.16 (0.43)	9.5	27.2	.16*	0.08 (0.22)	5.5	22.6	.29*
Conflict coworker – unpaid work	0.10 (0.32)	4.6	15.8	.11	0.06 (0.20)	3.9	17.0	.33*
Hearing gossip about self	0.11 (0.24)	7.7	36.0	.03	0.09 (0.23)	5.3	26.3	.06
Child irritable, sullen, or angry	0.26 (0.31)	18.7	72.8	.15*	0.25 (0.32)	18.2	68.4	.18*
Child arguing with me	0.23 (0.34)	17.6	57.0	.17*	0.16 (0.32)	10.1	51.8	.30*

Note.

^*p < .01. r = daily associations with marital conflict. Boxes = gender difference in means.

^†= gender difference in correlations.

Daily marital conflict

Husbands and wives separately completed 12 items assessing daily marital conflict. Scores per item ranged from 0 (not at all) to 3 (a lot). Six items assessed the actions of the reporter (i.e., I was angry at my spouse; I was annoyed with my partner/spouse; I yelled at or criticized my partner/spouse; I felt distant or withdrawn from my partner/spouse; I nagged my partner/spouse; I flew off the handle or exploded at my partner/spouse). An identical six items assessed the actions of the spouse (My partner was angry at me; My partner/spouse was annoyed at me, etc.). Across reporters, wife and husband reports were significantly correlated (r = .48, p < .001). Within reporters, reports of self and partner behavior were highly correlated for both wives (r = .88, p < .001) and husbands (r = .80, p < .001), which is to be anticipated given the dyadic nature of conflict. The internal consistency of conflict was .89 for wives’ reports and .88 for husbands’ reports. For parsimony and in light of the high correlations across the actors and reporters, we calculated one score for conflict per day by taking the average across all items.

Marital aggression

Marital aggression over the past year was measured using 18 items from the Conflict Tactics Scale and the Domestic Conflict Inventory (CTS; Straus, 1979; DCI; Margolin, John, & Foo, 1998). In 9 items, participants reported on their own behaviors toward the spouse and, in an identical 9 items, participants reported on the spouse's behaviors towards them. Examples of items include: insulted or swore at your spouse; did something to spite your spouse; pushed, grabbed, or shoved your spouse; slapped your spouse; threw something at your spouse; hit or tried to hit your spouse with something; kicked, bit, or hit your spouse with a fist). Scores for husband and wife aggression were obtained by taking the maximum value per item across the two reporters. Husband and wife aggression were highly correlated (r = .62, p < .001). In total, 71.5% of couples reported aggression in both spouses, 11.5% reported aggression in neither spouse, and 16.9% reported aggression in one spouse and not the other. Given the high correlations across husband and wife aggression and to increase parsimony in the analyses, we added scores for husbands and wives to obtain one index of total marital aggression per couple. Cronbach's alpha for total marital aggression was .95.

Family-of-origin aggression

In a measure designed for the larger study, husbands and wives separately reported on their exposure to FOA before the age of 18 (see Arbel, Rodriguez, & Margolin, 2015). Items assessed parent-to-child and interparental verbal and physical aggression on a 7-point scale, ranging from 0 (never) to 6 (more than 50 times). For parent-to-child aggression, 2 items assessed verbal aggression (e.g., how often did your parents scream or yell at you), and 2 assessed physical aggression (e.g., how often did your parents physically injure or bruise you). An identical set of items assessed interparental aggression (e.g., how often did your parents scream or yell at each other, etc.). Scores were averaged so that higher scores reflected greater exposure to FOA. Cronbach's alpha was .81 for wives and .88 for husbands.

Overview of Analyses

Hypotheses were tested using multilevel modeling to account for dependency in the data resulting from the repeated measurements across days. All daily measures were validated via multilevel factor analysis to show adequate model fit and discriminant validity across the study constructs. We first present descriptive statistics and correlations; multilevel modeling was used to obtain daily correlations between level 1 variables (e.g., daily stress and daily conflict) and correlations between level 1 and level 2 variables (e.g., daily stress and marital aggression). We examined associations between individual stressors and conflict and then tested the cumulative impact of these different stressors on marital functioning. Hypotheses 1, 3, and 4 were tested twice, first on the same day (i.e., to test if stress and conflict were linked on the same day) and then across days (i.e., to test if stress predicts conflict on the next day and vice versa). To conduct the cross-day analyses, we created lagged versions of the variables and then used the lagged variables as predictors of next-day marital conflict and stress. All cross-day models controlled for previous-day levels of the dependent variable. In addition, all wife models included husband stress as a covariate and vice versa. Because daily stress may be impacted by whether participants worked that day, we included whether it was a weekday as a covariate. We also included the day of reporting as a covariate because endorsement may change over time as participants become fatigued. Missing data were handled using full information maximum likelihood estimation in Mplus Version 7 (Enders, 2010; Mutheń & Mutheń, 1998–2012).

Figure 1 depicts the models for each study hypothesis. To examine spillover between stress and conflict (Hypothesis 1), we tested same-day and cross-day effects simultaneously. The first model tested links from stress to marital conflict (Row 1, Column 1); when testing links from marital conflict to stress, we used separate models to test wife stress as the outcome (Row 1, Column 2) and husband stress as the outcome (Row 1, Column 3). Tests of same- and cross-day links from stress to conflict were conducted using the equation presented below (Hypothesis 1; Row 1, Column 1):

$$\text{Level 1}:{\text{daily marital conflict}}_{\mathrm{ij}}={\beta }_{0\mathrm{j}}+{\beta }_{1\mathrm{j}}\left({\text{same-day wife stress}}_{1\mathrm{ij}}\right)+{\beta }_{2\mathrm{j}}\left({\text{same-day husband stress}}_{2\mathrm{ij}}\right)+{\beta }_{3\mathrm{j}}\left({\text{previous-day wife stress}}_{3\mathrm{ij}}\right)+{\beta }_{4\mathrm{j}}\left({\text{previous-day husband stress}}_{4\mathrm{ij}}\right)+{\beta }_{5\mathrm{j}}\left({\text{weekend}}_{5\mathrm{ij}}\right)+{\beta }_{6\mathrm{j}}\left({\text{day of reporting}}_{6\mathrm{ij}}\right)+{\mathrm{r}}_{\mathrm{ij}}$$

$$\text{Level 2}:{\beta }_{0\mathrm{j}}={\gamma }_{00}+{\mathrm{u}}_{0\mathrm{j}}$$

The same basic equations were used to test cross-day links in the opposite direction, with conflict predicting next-day wife (Column 2) and husband (Column 3) stress. Hypothesis 2 examined if spillover was greater on days when both partners were highly stressed. To test this hypothesis, we used the following equation (Hypothesis 2; Column 1, Row 2):

$$\text{Level 1}:{\text{daily marital conflict}}_{\mathrm{ij}}={\beta }_{0\mathrm{j}}+{\beta }_{1\mathrm{j}}\left({\text{same-day wife stress}}_{1\mathrm{ij}}\right)+{\beta }_{2\mathrm{j}}\left({\text{same-day husband stress}}_{2\mathrm{ij}}\right)+{\beta }_{3\mathrm{j}}({\text{same-day wife stress}}_{1\mathrm{ij}}\times {\text{same-day husband stress}}_{2\mathrm{ij}})+{\beta }_{4\mathrm{j}}\left({\text{weekend}}_{4\mathrm{ij}}\right)+{\beta }_{5\mathrm{j}}\left({\text{day of reporting}}_{5\mathrm{ij}}\right)+{\mathrm{r}}_{\mathrm{ij}}$$

$$\text{Level 2}:{\beta }_{0\mathrm{j}}={\gamma }_{00}+{\mathrm{u}}_{0\mathrm{j}}$$

The moderating effects of marital aggression (Hypothesis 3) and FOA (Hypothesis 4) were tested in the same models to determine if FOA is associated with spillover above and beyond current marital aggression. Husband and wife effects were tested in separate models (wives: Row 3; husbands: Row 4). For both husbands and wives, we tested three models: in the first, we tested same-day effects (Column 1); in the second, we tested links from stress to conflict (Column 2); in the third, we tested links from conflict to stress (Column 3). The following equation represents the link between wife stress and same-day conflict moderated by marital aggression and FOA (Hypotheses 3 and 4; Row 3, Column 1):

$$\text{Level 1}:{\text{daily marital conflict}}_{\mathrm{ij}}={\beta }_{0\mathrm{j}}+{\beta }_{1\mathrm{j}}\left({\text{same-day wife stress}}_{1\mathrm{ij}}\right)+{\beta }_{2\mathrm{j}}\left({\text{same-day husband stress}}_{2\mathrm{ij}}\right)+{\beta }_{3\mathrm{j}}\left({\text{weekend}}_{3\mathrm{ij}}\right)+{\beta }_{4\mathrm{j}}\left({\text{day of reporting}}_{4\mathrm{ij}}\right)+{\mathrm{r}}_{\mathrm{ij}}$$

$$\text{Level 2}:{\beta }_{0\mathrm{j}}={\gamma }_{00}+{\gamma }_{01}\left({\text{wife FOA}}_{1\mathrm{j}}\right)+{\gamma }_{02}\left({\text{marital aggression}}_{2\mathrm{j}}\right)+{\mathrm{u}}_{0\mathrm{j}}$$

$${\beta }_{1\mathrm{j}}={\gamma }_{10}+{\gamma }_{11}\left({\text{wife FOA}}_{1\mathrm{j}}\right)+{\gamma }_{12}\left({\text{marital aggression}}_{2\mathrm{j}}\right)+{\mathrm{u}}_{1\mathrm{j}}$$

Similar equations were used to test cross-day links with wife stress and for the husband models.

Figure 1.

Hypothesized associations being study constructs. Black lines = hypotheses; Gray dotted lines = covariates; Bolded words = dependent variable in each model; FOA = family-of-origin aggression.

Results

Descriptive Statistics

In addition to the descriptive statistics for the individual stress items presented in Table 1, the final column for wives and husbands shows the daily associations between each source of stress and marital conflict, tested at a significance level of p < .01 due to the large number of associations examined. For wives, marital conflict was significantly associated with: having too many things to do; having to make difficult decisions; own health; health of family members; traffic or bad drivers; noise; making mistakes doing paid work; pressure from unpaid work; feeling overloaded with unpaid work; being criticized doing unpaid work; child being irritable, sullen, or angry; and child arguing with me. For husbands, daily marital conflict was significantly associated with: being criticized doing unpaid work; having conflict with a coworker doing unpaid work; child being irritable, angry, or sullen; and child arguing with me. To determine if correlations between stress and conflict differed across husbands and wives, we used Wald Tests of Parameter Constrains for all items where (1) spillover was significant for both husbands and wives or (2) spillover was significant in husbands but not wives or vice versa. Wives had greater spillover for two items: health of family members and traffic or bad drivers.

Table 2 presents descriptive statistics and correlations for daily wife stress, daily husband stress, daily marital conflict, marital aggression, and FOA. Results showed that daily wife stress and husband stress were positively correlated with each other and that daily marital conflict was positively correlated with daily wife stress and husband stress. Marital aggression was positively associated with daily wife stress, husband stress, and marital conflict. By and large, FOA was not correlated with any of the study variables, with the exception of an inverse association between wife FOA and husband stress. There were no significant differences between husbands and wives in their mean levels of daily stress, conflict, or FOA (all p-values > .17). We also tested if age, years of education, number of years married, ethnicity, and family income were associated with the study variables and found no significant associations (all p-values > .11). However, weekends were associated with greater conflict (b = .04, p = .03) and with less stress in wives (b = −.04, p < .01) and husbands (b = −.04, p = .04). Results also showed that endorsement of conflict (b = −.08, p < .01) and wife stress (b = −.05, p = .04) decreased over time. Neither full-time nor part-time employment predicted mean levels of stress or conflict (all p-values > .11).

Table 2.

Descriptive Statistics and Bivariate Correlations between Main Study Variables

	M (SD)	Number of Days Endorsed	% Participants Endorsing Any (Min-Max)	1.	2.	3.	4.	5.
1. Daily wife stress	0.42 (0.28)	1,346	100 (.02-1.59)	--
2. Daily husband stress	0.39 (0.29)	1,287	100 (.01-1.61)	.10**
3. Daily marital conflict	0.20 (0.24)	818	96 (0-1.60)	.26***	.18**
4. Marital aggression	0.88 (1.05)	--	90 (0-2.56)	.48**	.43***	.67***
5. Wife FOA	2.72 (1.22)	--	100 (1.00-6.25)	.01	−.20*	.08	−.14
6. Husband FOA	2.77 (1.38)	--	100 (0.63-6.25)	−.08	.12	.06	.04	.12

Note.

^*p < .05

^**p < .01

^***p < .001. FOA = family-of-origin aggression.

Daily Stress-Conflict Spillover and Moderation Analyses

Table 3 presents results for the wife and husband models testing same- and cross-day associations from stress to conflict (Column 1) and from conflict to stress (wife stress: Column 2; husband stress: Column 3; Hypothesis 1). As seen in Column 1, both wife and husband stress were significantly associated with greater conflict on the same day, but no cross-day links emerged from wife and husband stress to next-day conflict. Results also showed no cross-day links from conflict to next-day wife stress (Column 2) or conflict to next-day husband stress (Column 3). For completeness, effects for same-day spillover were included in all three models; the same-day effects were significant in all models tested.

Table 3.

Same-Day and Cross-Day Associations between Stress and Marital Conflict

1. Same- and cross-day links: stress → conflict	b (SE)	2. Same- and cross-day links: conflict → wife stress	b (SE)	3. Same- and cross-day links: conflict → husband stress	b (SE)
Intercept	.19 (.02)***	Intercept	.42 (.03)***	Intercept	.39 (.03)***
Same-day wife stress	.26 (.04)***	Same-day marital conflict	.18 (.04)**	Same-day marital conflict	.13 (.05)**
Same-day husband stress	.20 (.05)**	Same-day husband stress	.05 (.06)	Same-day wife stress	.04 (.05)
Previous-day wife stress	.06 (.06)	Previous-day marital conflict	−.02 (.03)	Previous-day marital conflict	−.01 (.02)
Previous-day husband stress	.03 (.05)	Previous-day husband stress	.01 (.04)	Previous-day wife stress	−.03 (.04)
Previous-day marital conflict	.12 (.03)***	Previous-day wife stress	.13 (.05)**	Previous-day husband stress	.13 (.05)*

Note.

^*p < .05

^**p < .01

^***p < .001. Boxes = tests of hypotheses. Covariates weekday and report number not shown. FOA = family-of-origin aggression. Column 1: links between (1) same- and previous-day wife stress and conflict and (2) same- and previous-day husband stress and conflict; Column 2: links between same- and previous-day conflict and wife stress; Column 3: links between same- and previous-day conflict and husband stress.

Hypothesis 2 tested whether marital conflict was greater on days that both partners reported high levels of stress. To test this hypothesis, we entered wife stress, husband stress, and the interaction between wife and husband stress as predictors of same-day conflict. In this model, the interaction effect was significant (b = .20, p < .001); we then tested simple slopes for (1) the association between wife stress and marital conflict at high and low husband stress and (2) the association between husband stress and marital conflict at high and low wife stress. For all moderation analyses, simple slopes were tested at the 15^th and 85^th percentiles. Results showed that at both low and high husband stress, the link between wife stress and conflict was significant (low husband stress: b = .20, p < .01; high husband stress, b = .32, p < .001). In contrast, the link between husband stress and conflict was not significant at low wife stress (b = .09, p < .07) but was significant at high wife stress (b = .27, p < .001).

Results for tests of Hypotheses 3 and 4 are presented in Table 4 for wives and Table 5 for husbands. For wives, both marital aggression and wife FOA were associated with greater same-day links between wife stress and conflict (Table 4, Column 1). Couples with high marital aggression and wives exposed to high FOA (Figures 2a-b) showed high levels of spillover, though spillover was still significant at low values of marital aggression and wife FOA. Across days, marital aggression also moderated the link between wife stress and conflict in both directions (Table 4, Columns 2-3). As shown in Figures 2c-d, only those couples with high marital aggression showed significant cross-day links between wife stress and marital conflict. Although the interaction between marital aggression and next-day wife stress was significant, the simple slope reached significance only at values of marital aggression at the 95^th percentile. Wife FOA moderated the link between marital conflict and next-day wife stress but not vice versa (Table 4, Columns 2-3; Figure 2e). Specifically, only wives with high FOA showed cross-day spillover from conflict to stress. For husbands, cross-day links from conflict to stress were significant only for couples with high marital aggression (Table 5, Column 3; Figure 3f). No other interaction effects for the husbands were significant. As follow up analyses, we tested if wife FOA moderated husband spillover and vice versa and found no significant effects.

Table 4.

Wife FOA and Marital Aggression as Moderators of Daily Wife Stress Spillover

1. Same-day wife stress and conflict	b (SE)	2. Previous-day wife stress → conflict	b (SE)	3. Previous-day conflict → wife stress	b (SE)
Intercept	.20 (.02)***	Intercept	.19 (.02)***	Intercept	.41 (.03)***
Same-day wife stress	.26 (.04)***	Previous-day wife stress	.08 (.05)	Previous-day marital conflict	−.06 (.03)
Marital aggression	.15 (.04)***	Marital Aggression	.15 (.04)***	Marital Aggression	.13 (.03)***
Wife FOA	.03 (.02)	Wife FOA	.02 (.02)	Wife FOA	.02 (.02)
Marital aggression × same-day wife stress	.06 (.02)**	Marital aggression × previous-day wife stress	.07 (.02)**	Marital aggression × previous-day marital conflict	.06 (.02)***
Wife FOA × same-day wife stress	.14 (.05)**	Wife FOA × previous-day wife stress	.07 (.04)	Wife FOA × previous-day marital conflict	.06 (.02)**
Same-day husband stress	.24 (.07)***	Same-day wife stress	.25 (.04)***	Same-day husband stress	.03 (.07)
		Same-day husband stress	.20 (.05)***	Same-day marital conflict	.17 (.04)***
		Previous-day husband stress	.06 (.05)	Previous-day wife stress	.12 (.05)*
		Previous-day marital conflict	.11 (.03)***	Previous-day husband stress	−.01 (.03)

Note.

^*p < .05

^**p < .01

^***p < .001. Boxes = tests of hypotheses. Covariates weekday and report number not shown. FOA = family-of-origin aggression. Column 1: links between same-day wife stress and conflict moderated by marital aggression and wife FOA. Column 2: links between previous-day wife stress and conflict moderated by marital aggression and wife FOA; Column 3: links between previous-day conflict and wife stress moderated by marital aggression and wife FOA.

Table 5.

Husband FOA and Marital Aggression as Moderators of Daily Husband Stress Spillover

1. Same-day husband stress and marital conflict	b (SE)	2. Daily husband stress → daily marital conflict	b (SE)	3. Daily marital conflict → daily husband stress	b (SE)
Intercept	.20 (.02)***	Intercept	.19 (.02)***	Intercept	.38 (.02)***
Same-day husband stress	.16 (.05)**	Previous-day husband stress	−.01 (.05)	Previous-day marital conflict	−.04 (.03)
Marital aggression	.14 (.04)***	Marital Aggression	.14 (.04)***	Marital Aggression	.23 (.07)**
Husband FOA	.01 (.01)	Husband FOA	−.01 (.01)	Husband FOA	.02 (.02)
Marital aggression × same-day husband stress	.03 (.03)	Marital aggression × previous-day husband stress	.02 (.03)	Marital aggression × previous-day marital conflict	.12 (.04)**
Husband FOA × same-day husband stress	.02 (.05)	Husband FOA × previous-day husband stress	−.04 (.03)	Husband FOA × previous-day marital conflict	−.01 (.02)
Same-day wife stress	.29 (.05)***	Same-day wife stress	.25 (.04)***	Same-day wife stress	.02 (.06)
		Same-day husband stress	.19 (.05)***	Same-day marital conflict	.12 (.04)**
		Previous-day wife stress	.07 (.06)	Previous-day wife stress	−.02 (.04)
		Previous-day marital conflict	.12 (.03)***	Previous-day husband stress	.11 (.04)**

Note.

*p < .05

^**p < .01

^***p < .001. Boxes = tests of hypotheses. Covariates weekday and report number not shown. FOA = family-of-origin aggression. Column 1: links between same-day husband stress and conflict moderated by marital aggression and husband FOA. Column 2: links between previous-day husband stress and conflict moderated by marital aggression and husband FOA; Column 3: links between previous-day conflict and husband stress moderated by marital aggression and husband FOA.

Figure 2.

Same-day and next-day associations between daily stress and marital conflict moderated by marital aggression and FOA. FOA = family-of-origin aggression. ¹For this interaction, the association between marital conflict and next-day wife stress was significant only for couples with marital aggression at the 95^th percentile or above.

Discussion

This study examined same- and cross-day links between stress and marital conflict and also tested factors that put couples at risk for experiencing spillover processes. Consistent with Hypothesis 1, we found evidence of spillover of daily marital conflict and same-day wife and husband stress. However, cross-day links between stress and marital conflict were not significant in either direction. As predicted, conflict was greatest on days when both spouses experienced high stress (Hypothesis 2). Couples with high levels of marital aggression showed heightened spillover of wife stress and marital conflict on the same day and bidirectionally across days; couples with high levels of marital aggression also showed cross-day links from marital conflict to next-day husband stress (Hypothesis 3). Same-day spillover between wife stress and marital conflict was heightened for wives exposed to greater FOA. Wife FOA also moderated the cross-day link from marital conflict to next-day wife stress (Hypothesis 4). It is noteworthy that cross-day links only emerged for couples with either high marital aggression or high wife FOA.

Hypothesis 1: Spillover between Stress and Marital Conflict

These findings replicate research suggesting that stress can impact marital functioning on a daily basis (e.g., Buck & Neff, 2012; Heller & Watson, 2005; Repetti, 1989). An explanation for these results is that stress takes up emotional, physical, and cognitive resources, which interferes with the ability to regulate behavior and emotion. On the one hand, stress may leave people depleted of the energy needed to engage in a meaningful way with their partner (Doumas et al., 2003). On the other hand, because stress activates the “fight or flight” response, stressed spouses may be primed to act aggressively, interpret ambiguous situations negatively, and be more likely to respond with confrontational behavior. Consistent with past research (e.g., Buck & Neff, 2012), we did not find evidence of direct effects for cross-day spillover. Thus, most couples may experience a daily “reset,” where stress and marital conflict experienced the day before do not impact next-day behaviors or experiences. This “resetting” may be adaptive by preventing spillover from spiraling out of control and negatively impacting marital functioning.

Hypothesis 2: Heightened Spillover when Spouses Are Also Stressed

The couple-as-a-system approach represented in our analyses reveals three important findings related to the simultaneous examination of both partners’ stress. First, the results remained significant even when including husband and wife stress in the same model, suggesting there are independent effects of husband stress and wife stress on daily marital functioning. Second, the interaction between husbands and wives’ stress indicates that marital conflict is especially likely on days that both spouses experience high stress. Although we still need more detailed description of the actual interaction processes on those days, it is possible that each person acts negatively towards his or her partner and also responds more intensely to the behaviors of the other person. That is, neither is able to modulate the negative tone. This finding gives credibility to what couples anecdotally report about difficulty interacting or handling other stressful tasks when both spouses have experienced stress. Third, it appears that spillover processes are especially dependent on the stress states of wives; that is, wife spillover occurred regardless of husband stress levels whereas husband spillover only occurred if wives were also highly stressed. Interestingly, this result is parallel to other research testing stress crossover in couples using longitudinal data (Neff & Karney, 2007).

Hypotheses 3 and 4: Marital Aggression and FOA as Moderators of Daily Spillover

Although same-day links between stress and marital conflict appear fairly normative in this sample, stress spillover across days was related to specific risks associated with experiences of aggression. Couples who report high aggression during the past year appear to have particular difficulty recovering from the stress and marital conflict experienced the previous day and may get caught in cycles where stress increases the likelihood of next-day conflict and vice versa. Aggressive couples’ difficulties exiting and recovering from negative interactions have been evidenced in moment-to-moment behavioral sequences in the home (Burman, Margolin, & John, 1993), as well as delayed cortisol recovery from conflict discussions in the lab (Feinberg, Jones, Granger, & Bontempo, 2011). The data here suggest another failure to recover in terms of cross- day, bidirectional links between stress and conflict. Whereas the spillover processes dissipate over night for the majority of couples, a night's rest does not appear to interrupt stress spillover for aggressive couples. More detailed analysis is needed to understand what transpires in the aggressive couples. Do arguments persist, disrupting evening tasks or delaying sleep, both of which could increase stress the following day? Does previous-day stress lead to next-day disrupted diurnal hypothalamic-pituitary-adrenal patterns, resulting in reduced capacity to handle stress or conflict? A key point of our data is that in addition to overall links between aggression and stress, aggression potentiates prolonged patterns of spillover between stress and conflict.

The long-term legacy of wives’ FOA as a risk-factor for cross-day links from conflict to wife stress suggests that childhood experiences play an important and unique role in spillover processes. As compared to marital aggression, which is recent or even ongoing, FOA occurred decades ago and is unrelated to the reports of marital aggression. In another paper with this same sample (Arbel et al., 2015), wives’ FOA also was linked to greater cortisol activity in a conflict discussion with their spouse and adolescent child, which led to our speculation that women's early experiences hold more relevance in future stressful interpersonal situations. This, of course, needs to be replicated in other samples but does suggest that FOA may influence the interpretation of interpersonal stress. Perhaps connectivity between stress and conflict in females exposed to FOA reflects their tendency to take a “tend and befriend” approach when stressed, whereby interpersonal relationships and stress regulation are more intertwined (Taylor et al., 2000). Future research testing specific behaviors (e.g., criticism, withdrawal) and coping styles will help to more fully understand these gender differences.

Implications

Although major events may negatively impact relationship functioning, this study points to the importance of daily stressors. Our findings suggest that spillover is a common occurrence but that most couples do not experience spillover across days. Cross-day spillover may reflect maladaptive relationship processes where negative events in different domains are reinforced and sustained over time. The identification of maladaptive relationship patterns is a common target in couple interventions. Frequently, the targeted patterns are momentary communications between partners (e.g., Jacobson & Christensen, 1996), although some intervention approaches also address larger daily patterns, such as repeated conflicts surrounding couples’ reunification after separate workdays (e.g., Shoham, Rohrbaugh, & Cleary, 2008). The data here highlight the relevance of daily patterns involving everyday events, as well as the importance of how spouses interpret one another's reactions to those events. Increasing awareness of stress spillover processes and examining each partner's role could be productive directions in therapy.

Strengths and Limitations

This study used daily diary methodology to examine associations between stress and marital conflict on the same day and bidirectionally across days. Our same-day results replicate a small body of research documenting daily associations between stress and marital functioning in couples (e.g., Bolger et al., 1989; Buck & Neff, 2012; Heller & Watson, 2005; Repetti,1989; Schulz et al., 2004; Story & Repetti, 2006). We extended the literature by assessing spillover originating in both work and home domains, testing how stress in both partners on the same day impacts marital functioning at the systems level, and by examining marital aggression and FOA as risk factors for spillover processes. The daily data design allowed us to model everyday marital functioning in a naturalistic setting and test temporal links between stress and conflict. Moreover, our findings indicate that spillover in couples is related not only to concurrent marital functioning but also to experiences in the family-of-origin. In addition, because our measure of stress was designed to capture paid and unpaid work across traditional and non-traditional settings and to capture wide-ranging stress associated with health, financial worries, and time-pressured lives, we were able to test broadly construed life stressors—both individually and as an aggregate. The items, however, were selected for couples in an urban environment with adolescent children; specific social, cultural, technological and geographic circumstances also should be considered, e.g., racial aggression, technology problems. In addition, analyses testing links between conflict and specific stressors were exploratory; conducting more systematic comparisons across individual stressors and stressor types (e.g., shared or unshared; controllable or uncontrollable) is an important direction for future work.

Despite these strengths, our results should be interpreted in light of several important limitations. First, because the participants completed one diary assessment per day, we cannot determine the direction of associations between stress and conflict within the same day. Second, though we examined effects across one day, we did not test if the spillover processes persisted for longer timeframes for some couples. Third, despite bringing the analysis to the timeframe of daily influences, we still lack information about explanatory mechanisms. Other researchers (Buck & Neff, 2012; Story & Repetti, 2006), for example, have shown that negative mood and self-regulatory depletion mediate the link between daily stressors and conflict. The identification of mechanisms is particularly important for translating these results into clinical interventions or other applications. Fourth, some measurement issues deserve note: Because our stress measure inquired “how concerned were you about...”, we cannot distinguish between the occurrence of and concern about stressors, although, as a positive, we did capture how the stressors were uniquely experienced. Relatedly, we did not differentiate between types of conflict behaviors. Withdrawal has been shown to be an important component in husbands’ responses to stress (e.g., Repetti, 1989), and although represented here through one item, it was not tested separately due to power issues. We also could not determine whether wives or husbands initiated the conflict; this distinction would provide a more a nuanced examination of spillover. Fifth, because the couples in this study have adolescent children, they might face unique stress related to parenting adolescents and have specific ways of dealing with marital conflict, e.g., postponing discussing issues until children are in bed; these factors limit generalization to other samples. Nonetheless, spillover may be particularly relevant to this sample: Though not examined here, it is possible that parents’ spillover between stress and conflict is apparent to the children and influences them.

Conclusion

This study is an extension of research testing links between stress and marital conflict in daily life; our results lend further evidence to the idea that everyday stressors can impact couple functioning. While daily links between stress and conflict may be common, extended stress spillover may be indicative of maladaptive relationship processes. Aggressive couples may fail to recover from stress and conflict experienced the day before, putting them at risk for greater stress and conflict the next day. Additionally, negative family-of-origin experiences may impact stress responding and set the stage for maladaptive relationship processes. In sum, this study points to how day-to-day stress can spark conflict between spouses and suggests that interrupting stress spillover might be an effective intervention for improving couple functioning.