9Stress
RENA REPETTI, DARBY SAXBE, AND SHU-WEN WANG
We may all wonder how the ups and downs of life outside the home affect what goes on inside the family. Do stressful experiences permeate the domestic sphere? If a father has a bad day at work, will he treat his wife and children differently that evening? When we examined the behavior and physiology of CELF family members after work, we did find evidence that job stressors influenced life at home. However, the effects were not “one-size-fits-all.” First, the nature of work-to-home spillover differed for mothers and fathers. Among fathers job stress appeared to have a visible effect on their outward social behavior at home, whereas among mothers the impact was more internal, with evidence of lingering stress hormones in the evening after work. Even within each gender, the consequences of job stress varied depending on characteristics of the parent and the family. The responses of fathers who had trouble regulating their negative emotional states or who were unsatisfied with their marriages differed from the responses of fathers who generally maintained an even emotional keel or were in happier marriages. Mothers who described their marriages as less satisfying or performed a highly disproportionate share of the household labor seemed to recover less readily from the physiological effects of a stressful day at work.
WHAT WE ALREADY KNEW
Our analysis of the CELF data was guided by previous research on how parents’ experiences at work shape family life. One of the ways that psychologists have addressed those questions is by looking closely at the lives of people who hold different kinds of jobs. Employees who say they are satisfied with their jobs, or who describe greater autonomy at work, seem to have happier family lives. They show more warmth and responsiveness to their children and describe their marriages as more satisfying. Of course, correlation does not necessarily imply causation. But longitudinal studies follow the same group of people over months or years and are therefore able to tease out the effects of stable individual characteristics, such as personality traits, that shape life both at work and at home.1 These studies indicate that experiences of achievement and fulfillment at work can, over time, carry over into the family and have a positive impact on relationships. On the flip side is chronic job stress, which can lead to signs of personal distress, including symptoms of depression, and can also impair marital and parent-child relations. Employees whose dealings with coworkers and supervisors are conflictual and unsupportive seem to have less affectionate and angrier interactions with their children and to experience more marital tension and arguments. Similarly, workers who face constant demands and time pressure on the job report more interpersonal tensions at home, get into more conflicts with family members, and know less about their children's daily activities and whereabouts.2
Some psychologists have considered how behavior and emotion in the same family differ from one day to the next. For example, prior to the CELF project, Rena Repetti and her students studied air traffic controllers in Southern California, mothers employed in white-collar jobs in New York City, and a different sample of dual-earner middle-class couples in Los Angeles.3 In each study, we measured the stressfulness of each day at work for several consecutive days, asking our employed participants to rate their workload and the quality of their social interactions with coworkers and supervisors. We wanted to see if the family's interactions differed following a particularly jam-packed day at work or one marred by a tense exchange with a coworker. To assess how the employed person interacted with family members on each day after work, we sometimes used videotapes and we always asked family members to describe social interactions in their family each evening.
Across studies, when we compared these two sets of measures—daily job stress and daily family interactions—two interesting patterns emerged. On the one hand, more stressful days at work were often followed by more expressions of anger and irritability at home, compared to the same employed person's behavior after less stressful days. For instance, when describing marital interactions after the stressful workdays, both the employed person and his or her spouse were more likely to say that the employee was “impatient,” or “sarcastic,” or “a little touchy” that night, or that he or she started an argument or disagreed with something the spouse said. Interactions with a child were more likely to involve the use of disciplinary tactics such as reminding, yelling, or even punishing the child. This pattern is called “negative emotion spillover,” based on the idea that the frustrations and pressures of the workday spilled over into the home, where they continued to be expressed in the mood and behavior of the employed parent or spouse.
The second observed pattern is social withdrawal: more stressful days at work followed by an overall decline in the employee's expression of all emotion, both positive and negative, and an overall decline in the amount and intensity of his or her social behavior. Compared to the slower, easier, or more socially serene workdays, after stressful days at work the employed person was more likely to say that he or she “wanted to be alone,” “was too tired to interact,” “didn't want to talk,” “wanted some quiet time to be alone,” or simply “ignored” the spouse. Spouses’ ratings were in agreement with the description of a less involved, more distracted family member. When we looked at videotapes taken at a work site day care center of mother-child reunions, we saw that after fast-paced and demanding days at work mothers spoke less and their attention was less likely to be completely focused on their preschool-aged child.
How can we reconcile two such seemingly different responses to job stressors? One paints a picture of an employed person backing off from family members and the other suggests a more intense and more irritable person. We wondered how these patterns would appear without the filtering of individual perception and reporting; what would they look like in the raw? Would we find that some people respond to job stress with social withdrawal and that a different group responds with irritability? Alternatively, after a difficult day on the job might the same stressed-out and fatigued person try to avoid social interaction and then become impatient and disagreeable if a spouse's request for assistance or a child's question interferes with that goal? What does wanting to be alone or feeling too tired to interact with others look like to an outside observer who doesn't know what the individual is thinking or feeling?
CELF PARENTS’ SOCIAL INTERACTIONS IN THE FIRST HOUR HOME
The CELF study was uniquely positioned to shed new light on this subject; it could show us how employed parents actually behaved at home after work. Researchers would finally see what husbands and wives, mothers and fathers, actually say and do when they are with their loved ones at home after work. When psychologists study social behavior, they quantify—often in minute detail—what they see. They rate qualities such as degree of responsiveness to another person, facial expressions, and emotional reactions to others; the codes they use often require a painstakingly detailed analysis of social behavior every few seconds. With this type of approach, it would be easy to get lost in the over 1,600 hours of video that CELF collected. We had to devise a unique strategy to describe social interaction in the CELF families. Belinda Campos, a researcher at CELF, hit upon the idea of systematically sampling and coding thin slices of video.
Working with another CELF researcher, Tatyana Plaksina, Repetti, and Campos devised a coding system that would be applied to thirty-second segments of videotape sampled every ten minutes from the continuous footage taken in the homes of the families in our study. The sampling started as soon as a parent returned home from work and ended when the cameras stopped rolling at night. Before long, a larger team, which included graduate students Shu-wen Wang and Jacqueline Sperling, and ten UCLA undergraduates, was assembled. We worked for a full year applying the detailed coding scheme to the mothers’ and fathers’ interactions captured on videotape on the two weekday afternoons and evenings. By examining video recordings of family interaction at home at ten-minute intervals and characterizing whom the parents were with and what they were saying and doing, we were able to sketch a picture of their social behavior after work. The result after a year of coding was an archive consisting of long strings of numbers that quantified the social and emotional qualities of each parent's interactions with family members on two weekday afternoons and evenings.
We can only put faith in these strings of numbers as reliably reflecting meaningful aspects of social behavior if the codes they represent were applied in an equivalent way by all coders. Recall that a rather large team, more than ten coders, was involved in the coding effort; how did we ensure that Coder A's coding was comparable to that of Coder B's . . . and Coder C's . . . and Coder D's? This is where inter-rater reliability comes in. Inter-rater reliability reflects the consistency with which the coding system was implemented. To ensure a high level of inter-rater reliability, all members of the CELF coding team were intensively trained on practice video clips and required to reach a high level of inter-rater reliability (we used the criterion of an intraclass correlation coefficient = .80) before being allowed to code. Afterwards, all coders overlapped on 10 percent of the video, and inter-rater reliability was computed regularly to ensure the quality of the coding (i.e., that intraclass correlation coefficients remained above .80). Crucial to this process were regular weekly meetings in which all members of the coding team gathered to discuss and resolve coding concerns and questions.4
CELF researchers, led by Wang, used the codes to see how experiences at work related to social behavior at home in the evening.5 Reasoning that the effects of job stress would be most readily observed soon after leaving work, we decided to focus our analysis on the social interactions that occurred in the first hour after each parent arrived home. Based on prior research showing that job stress is associated with social withdrawal and negative emotion spillover, the analysis honed in on three of the codes. Two describe how vigorous the individual's social behavior was during each thirty-second slice of video. One of the “social vigor” codes, scored on a 0–3 scale, measures the intensity of response to another's initiation for interaction. For example, if a spouse asks, “How was your day?” the person whose behavior is being coded could simply ignore the question (and receive a 0 for being “not at all” responsive), grunt (and be assigned a score of 1 for low response intensity), or say, “Good, how was your day?” (which would be coded a 2 for moderate response intensity); an effusive and detailed response that included details of the day would get a high response intensity score of 3. The other social vigor code was talking, and it was also scored on a 0–3 scale. Complaining about work continuously at the dinner table would earn a score of 3, whereas complete silence would result in a score of 0 for that thirty-second clip. A few words would receive a score of 1 for minimal talking, and a conversation in which the target asks a couple of questions while also responding to another's questions would be rated a 2 for moderate talking.
Although the two social vigor codes might seem similar, they actually tapped different dimensions of social behavior. First, a response intensity rating could be assigned only if another person made some bid for attention during the thirty-second clip. The other social vigor code, talking, was scored regardless of any particular behavior on the part of others who were present. However, even when both codes were applied, the ratings could differ quite a bit. For example, a parent could do plenty of talking while ignoring an initiation from a child, which would earn a high score for talking and a low score for response intensity. Alternatively, a parent who said little but hugged her child when she described a bad day at school would merit a low score for talking and a high response intensity score.
Because the two social vigor codes reveal nothing about the emotional tone of an interaction, the other key variable was a measure of negative emotion displayed by the parent. The scores, which could range from 1 (none at all) to 7 (many), reflected both the effusiveness and the quantity of negative emotion behaviors, such as facial movements (e.g., frowns, raised eyebrows), body language (e.g., turning away), and voice quality (e.g., angry, stern). Taken together, the social vigor and negative emotion codes told us quite a bit about how active and intense, and how negative, the parent was during the first hour after returning home. This information allowed us to recognize social withdrawal and negative emotion spillover if they were present.
The sheer descriptive power of these codes—response intensity, talking, and negative emotion—paint a picture of how working parents interact with spouse and children during their first hour home after work. We observed that the men and women in our study showed similar amounts of talking (low to moderate), whether they were interacting with their spouse or with their school-aged child. Interestingly, although men and women generally expressed little negative emotion, women displayed more negative emotion in their first hour interactions with husband as well as school-aged child. Men, on the other hand, responded to their wives with greater intensity than wives did to their husbands; however, there was no difference between men and women in levels of response intensity to their children in their first hour interactions.
Of course, we were primarily interested in how experiences on the job would be linked with social behavior after work. In addition to video recordings of their family interactions, twice each day at work—once before lunch and once before leaving at the end of the day—CELF parents told us about their work experiences. One set of questions asked about the demands on their time and the pressure to perform tasks quickly. Another set of questions asked them to rate how they felt—resentful? tense? distant?—during interactions with coworkers and supervisors that day. The validity of both measures is documented in our prior work; there are significant correlations between these scales and objective measures of workload, social support at work, and satisfaction with work relationships.6 Likewise, both measures demonstrate adequate internal consistency in prior research,7 as well as in our CELF families. With only two weekdays of video, the CELF study was not well designed to compare the same person's social behavior on more stressful and less stressful days. After all, the chances were low that the two days that we happened to videotape would capture one high-stress workday and one low-stress work-day from each member of a couple. Therefore, each individual's job ratings were averaged across the two days and then, given the high correlation between the two measures, aggregated to form a composite score, a single overall assessment of the stressfulness of each parent's job over the two days. The ratings that the coders made of each parent's social behavior during the first hour at home on the two weekdays were also averaged.
Armed with these two sets of ratings—the stressfulness of the parents’ jobs and the parents’ social behavior during their first hour at home after work—Wang tested correlations between them. The husbands’ and wives’ data were analyzed separately because of evidence from the research literature that the two sexes may respond differently to stressors. At first glance, there were no unambiguous and convincing patterns suggesting that parents with more stressful jobs behaved any differently when they got home from work than did the parents who described wonderful relationships with supervisors or said they worked at a relaxed pace. But then Wang dug a little deeper. Some earlier studies, including publications coming out of our own group (though not based on the sample of CELF families), suggested that patterns of negative emotion spillover or social withdrawal may differ according to individual differences in personality or differences in the quality of family relationships. In one of those earlier studies, both patterns—withdrawal and negative spillover—were more likely to be observed in families with higher levels of conflict and more expressions of anger.8 In another previous study, individuals who experienced symptoms of anxiety and depression were more prone to carry their stress into the home; both withdrawal and negative spillover were observed in the portion of the sample that reported more symptoms of emotional distress.9
Parents in the CELF study had all completed standard measures of marital satisfaction and neuroticism, which is the tendency to experience negative emotional states. The neuroticism scale asks about feeling “worthless,” “angry,” “depressed,” and “tense or jittery.” Individuals who score high on neuroticism are thought to have trouble regulating negative emotional states like these and are perceived to be less emotionally stable. The scale we used has shown good internal consistency and has been well validated in other studies.10 To assess marital satisfaction, couples told us about their feelings of closeness and happiness and agreements and disagreements in a variety of areas, such as friends, in-laws, and the handling of finances. This scale has been shown to discriminate well between satisfied and dissatisfied couples and demonstrates good internal consistency.11
Working from the clues in the earlier studies, Wang tested whether men and women in happier marriages, or those with lower neuroticism scores, responded to job stress differently than did the men and women who described their marriages as less satisfying, or those who scored at the higher end of the neuroticism scale. Bingo! Multiple regression statistical analyses revealed a very clear and consistent pattern in the men's data. Among these husbands, we could see evidence of both withdrawal and negative emotion spillover. The impact of job stress was there but depended on the husbands’ levels of marital satisfaction and neuroticism.12
On the one hand, among the husbands in happier marriages and among the men who scored lower on the neuroticism scale, job stress was associated with something that looks very much like social withdrawal at home. Among these men, as the scores on the indicators of job stress got higher we found that they spoke less to their wives and they expressed less negative emotion with both wives and children when they returned home. For example, for the men who described themselves as happily married, there was a strong negative correlation between their experience of stress at work and their expressions of negative emotion, both with their wives (r = -.71) and with their children (r = -.67). Our analysis was based only on occasions when the men in the CELF families were with other family members. The focus here is the quality of the interactions that fathers had with their wives and children during their first hour home as a function of stressors at work. We learned that for the more happily married men and men who reported little emotional distress involvement in those interactions seemed to decrease as stress at work increased.
On the other hand, among the husbands who were not as happy in their marriages and among the men who described more emotional distress, a more stressful job was linked with signs of negative emotion spillover after work. Within this group, as ratings of stress at work increased, so did the indicators of vigorous social behavior. During the first hour at home, they did more talking with wives and their social responses to wives were more intense and elaborated. Perhaps most important, they also expressed more negative emotion, with wives as well as with children. In contrast to the correlations presented above for the happily married men, among the men who described lower levels of marital satisfaction, there was a moderate positive correlation between their experience of stress at work and their expressions of negative emotion, both with their wives (r = +.30) and with their children (r = +.34). In short, the men who said that they were less happily married and who experienced more emotional distress showed evidence of negative emotion spillover; they appeared to be both more active and involved in social interaction—and more negative—as stress at work increased.
None of these patterns were observed in the wives’ responses. Whether or not they were happy in their marriages and whether or not they reported symptoms of emotional distress, there was simply no connection between what a wife and mother's life at work was like and how she responded to family members after work. But the husbands’ data did tell a story. At least on the two weekdays that we studied them, the men in the CELF families behaved differently depending on two factors: how much stress they described at work and how “distressed” they said they were. In our analysis, “distress” could mean either how much anxiety and depression they reported generally experiencing or how satisfied they were in their marriages.
STRESS HORMONES AND THE MARITAL ENVIRONMENT
Although the husbands’ social behavior seemed to “tell a story” about how their workdays were carried home with them, it was a different piece of the CELF puzzle that yielded a series of significant patterns for wives. Cortisol is a stress hormone that is released into saliva and that shows a regular daily pattern (peaking in the early morning and decreasing across the day). CELF participants were given portable vials so they could sample their saliva at home and at work, and they turned in three days of samples, taken at regular intervals in the morning, late morning, afternoon, and evening.
CELF researcher Darby Saxbe mapped each spouse's average cortisol trajectory, using all twelve samples across three days to get a sense of each person's daily decline in cortisol, something known as a diurnal slope. Earlier studies had established a relationship between this diurnal slope and other measures of health and well-being; for example, several studies of women with cancer found that women with flatter cortisol slopes, in which cortisol starts out lower in the morning and decreases less over the day, reported more fatigue and even had earlier mortality than women with steeper slopes.13 Other studies linked flatter cortisol slopes with depression, burnout, and chronic stress, and one study found that working mothers who reported more relationship problems had flatter slopes of cortisol as well.14 Given this body of evidence, we were curious whether the marital satisfaction of the couples in the CELF study would correlate with their diurnal patterns of cortisol. As it turned out, it did—but only for the wives. Women who reported greater marital dissatisfaction had flatter slopes of cortisol, the same pattern linked with more health problems and life stressors in other studies. For husbands, cortisol and marital satisfaction did not appear to be linked.15
Wives’ marital satisfaction not only predicted their cortisol slope but also seemed to affect their pattern of recovery after a busy day at work. To explore the question of how the workday affected evening cortisol levels, we used the same measures of workload and negative social interactions at work mentioned above. Whereas Wang's analysis matched job stress ratings to only two weekdays of videotaping, the cortisol analysis could capitalize on all three days of job stress ratings provided by the CELF couples because cortisol samples were collected on each of these days. On the highest workload days, both husbands and wives actually showed lower than usual cortisol levels in the evening—suggesting an exaggerated overcorrection by the structures in the brain and body that regulate cortisol. For women, however, marital satisfaction affected the trajectory of this recovery, such that the “overcorrection” appeared only among happily married women. Women in more unhappy relationships not only showed flatter cortisol during the day, but weaker recovery after busy days at work.16 While it may seem odd that the cortisol results emerged only for women, it actually dovetails with previous findings in the public health and epidemiology literature. It has long been known that marriage carries an overall health benefit for men: married men get a boost in longevity and well-being across the board, when contrasted with single men. In contrast, women appear to be more sensitive to the quality of the marital environment. For example, studies that have contrasted happily married, unhappily married, and single women have found that happily married women appear healthiest, while unhappily married women tend to fare worse than single women.17 The fact that maritally dissatisfied women in the CELF study had flatter cortisol slopes and impaired recovery from work suggests one physiological mechanism through which poor marital quality may specifically affect women's health.
Why was marital satisfaction so important for wives’ cortisol patterns but not for husbands'? Was there something about the quality of the marriage that might affect women's everyday stress more than that of men? Perhaps women in unhappy marriages were doing a greater share of the work around the house. In other words, measures indicating dissatisfaction with marriage might be a proxy for an unfair domestic arrangement. To test this question, Saxbe used the scan sampling of activities data to see what spouses were actually doing at home. As described in the appendix, a researcher recorded the locations and primary activities of every family member at regular ten-minute intervals. For example, a tracking entry tells us that at 7:20 A.M. on Day 1, Macy was located in the kitchen and that she was engaged in the activity of packing lunches for her kids. At the same time her husband, Andre, was in the bathroom, taking a shower. Saxbe worked with a research assistant to sort more than five thousand activity entries into categories like “housework,” “leisure,” “childcare,” and “personal care.” She compared husbands’ and wives’ activities over four days of scan sampling and found striking differences. Within the average couple, compared to wives, husbands spent about twice as much of their time at home engaged in leisure, such as watching television, reading the newspaper, or listening to music. And they spent only about two-thirds of the time doing housework as did their wives.
Looking at the activities data in relation to evening cortisol levels, even more interesting results emerged. Unsurprisingly, people who spent more of their time at home doing housework showed less “recovery” at the end of the day—that is, their cortisol levels didn't drop as much from the afternoon to the evening—and parents who devoted more of their time to leisure activities showed greater recovery. What seemed to matter most, however, was not parents’ absolute allotment of housework or leisure time but the time they were spending relative to their spouse. In other words, in couples where women were spending a lot more time doing housework than their husbands and a lot less time in leisure, those wives were more likely to show diminished recovery at the end of the day.18 This result confirmed our initial hunch that women's cortisol patterns might reflect their feelings about the division of the labor within their households. It came as a surprise that, while there was evidence of job stress being carried back into the family by both husbands and wives, the findings for fathers related to their behavior at home and the findings for mothers related to their stress hormone levels. If we had focused on only one “system"—their overt social behavior, for example, but not cortisol—we might not have captured different patterns of results for men and women.
PUTTING IT ALL TOGETHER
How were social withdrawal, negative emotion spillover, and flattened cortisol slopes manifested in the CELF families? From having viewed the videotapes many times we knew that the CELF fathers didn't go around yelling or screaming at their children and wives, and they didn't completely ignore them either. What exactly was being reflected in higher or lower response intensity scores for those thirty-second slices of video, or higher or lower levels of cortisol, at different times of day? To understand these patterns better, it helps to look closely at a couple of scenes from family life. Focusing exclusively on the CELF families in which fathers reported high levels of stress at work, we identified one who reported a low level of distress on one indicator (marital dissatisfaction) and one who reported a high level of distress on both indicators (marital dissatisfaction and neuroticism). The early part of their evenings at home after work should help us to understand what social withdrawal and negative emotion spillover look like in the lives of middle class families.
ANDRE: A QUIET PRESENCE
Andre illustrates social withdrawal as a response to job stress. His ratings of both workload and the quality of his social relationships at work indicate above-average levels of job stress, and his marital satisfaction score was higher than any other father's in our study. Here's a man who is very happy in his marriage and stressed by his job. Let's see what happens when he returns home from work. On the first weekday of video recording the family followed their common routine of going out to dinner. Because the videographers weren't prepared to film in a restaurant—this family was one of the first to participate in the CELF study and this scenario had not been anticipated—their dinner was not captured on tape. Once the family returns home from the restaurant, Andre immediately tends to his aquariums, a favored hobby of his. During this time, Andre occasionally responds to comments made by his wife, Macy, and children, Peter (age ten) and Nancy (age eight), who are nearby, but doesn't initiate conversation. He seems very calm and relaxed as he works with the aquariums.
When those tasks are completed, Andre asks Peter if he needs help with homework, but Peter wants to work on origami with his dad. For the next forty minutes, while sitting on the floor of Peter's bedroom, father and son patiently work together to try to create a complex shooting star design by folding paper. They are following instructions from a book, and both are very involved and focused on the project. At times Andre reads and patiently guides Peter's folding; at other points Andre takes the origami and silently works to figure it out himself. Peter occasionally reads to his father from the origami book; Andre hears and answers “mmhmm” but doesn't look up. There is little talking but plenty of collaboration, watching, and experimenting with different approaches. There is a quiet connection between father and son, as they both appear to be absorbed in the task and cooperate with each other to reach the goal. As they work together, their mood remains positive, with occasional smiles and giggles, and neither gets frustrated or upset.
When Nancy enters the room and sings a silly song that she learned at Girl Scouts, she could be perceived as annoying and disruptive. Neither Andre nor Peter, however, seems bothered by the song that Nancy keeps repeating; they remain focused on the task. Not long after, father and son decide to stop and continue the project tomorrow. There are twenty-eight steps to produce the shooting star, and they've gotten about halfway through. Andre, however, continues to work on it for a few more minutes. They then carefully store the not yet completed paper star for the next day.
As the children transition to their bedtime routines, Andre and Macy quietly go through paperwork at the table. Most of it comes from school, and they occasionally consult about decisions (e.g., choices of after-school activities for the children), but most of the time they work side-by-side in silence. During that time, Macy tells Andre about a problem with a planned school field trip in Peter's class; there isn't enough space on the bus for all the children. Andre volunteers to help out by going on the field trip, and Macy fills out the volunteer form for him. For the most part, Macy and Andre are focused on reading, filling out forms, and other paperwork and don't say much to one another. Even when Nancy comes into the room and joins her parents at the table, again singing a silly song, Andre ignores her and stays on task.19
It's easy to see why, when the CELF coders checked in every ten minutes, Andre earned low scores for response intensity, talking, and negative emotion. We don't know if he “felt too tired to talk,” as the participants in previous studies have reported, but he did seem to prefer to focus his attention on tasks—such as tending to the aquariums, origami, and paperwork—over conversation. It's interesting to observe this very happily married man jump into family activities—making origami with Peter, looking at paperwork, arranging to attend a school field trip. He doesn't go off to watch television or read a newspaper alone. He's physically near his family, often collaborating with them, and there is no sign of negative mood or annoyance. At the same time, he doesn't choose to capitalize on the many opportunities for conversation with wife and children that are presented. Those are the kinds of opportunities that the father in our next family is quick to grab.
STEVE: AN IRRITATED TENSENESS
Steve presents an interesting contrast to Andre. His ratings of workload and negative social interactions at work were also above average, in fact even a bit higher than Andre's, indicating a high level of perceived job stress. But his marital satisfaction score was below the average score of other CELF husbands and far below Andre's rating. Moreover, his score on the neuroticism scale was almost twice the average score. According to the findings described earlier, his behavior when he returns home from work might tell us something about what constitutes negative emotion spillover in the everyday lives of middle-class fathers.20
Like Andre, this father also jumps into family activities as soon as he gets home. Steve checks in with everyone and their activities, asking questions about school, the children's day, and meal preparation. He reminds his wife to take her vitamins, takes the children on an errand with him, and gets involved in tasks that are of direct relevance to his family. He's proactive in engaging the children and finding opportunities to teach them about the world. At the same time, there are numerous instances in which he complains, criticizes, and expresses irritation, almost all directed toward his wife, Dinah. In fact, within just the first fifteen minutes of his arrival home, we see a number of behaviors in rapid succession that could be a reflection of a negative emotion spillover process. For example, soon after his arrival home, he and Dinah have a brief conversation about the fact that he was not able to find a bill that he was supposed to mail that morning. When Dinah reminds him that items to be dropped in the mail are always left on the table by the front door, he responds in an almost prosecutorial style.
| STEVE: | I told you it wasn't that big of a deal, right? Didn't I? | |
| DINAH: | Um, no, you didn't. | |
| STEVE: | I didn't? | |
| DINAH: | No . . . | |
| STEVE: | I remember tapping the words “slightly consternated.” | |
| DINAH: | Oh. | |
| STEVE: | You don't remember that? | |
| DINAH: | Yeah. | |
| STEVE: | You do remember that? | |
| DINAH: | Mm hmm. | |
| STEVE: | Then I'm correct. | |
| DINAH: | Does that mean ((pause)) that it's no big deal? |
Eight-year-old Gwen enters the kitchen at that moment and asks when she will be going to the home of her mother's cousin, Fran. While Dinah explains that Fran has been sick, Steve responds to his daughter's question by saying, “Not for a long, long time,” and chuckles. After a pause, he adds, “until she gets a house. Her place is too small to hang out at, Mag.” It's hard not to view these comments as a slight against his wife's cousin. A bit later, while sitting on a small balcony outside their bedroom, Dinah talks to the researcher about how important this cousin is to her, and her eager anticipation of a family gathering at which they will hang out together. Dinah explains that Fran's “energy and excitement about life” has always “rejuvenated me” and acted as a source of inspiration. Steve must know all this, so his put-down of a treasured cousin in front of the camera must have stung.
Less than a minute after Steve's comments about Fran, he asks Dinah if she and their five-ear-old son, Miles, had found a library book that was missing. When mom reports that they are looking for it, he replies, “Doesn't look like he's looking for it . . .” In what appears to be an effort to vouch for her son, Dinah tells her husband that Miles has been working hard all afternoon and that they will find it, not to worry. Shortly after that, in response to his wife's question about whether he had gotten a chance to fill out some school forms for Miles, Steve comments that he received “a chilly reception tonight.” Dinah sounds hurt and misunderstood as she replies, “No, I'm sorry.” Her husband then adds, “. . . racing home to get one of those.”
A few minutes later, Dinah seems to make one more attempt to turn things around. She describes how she helped Miles to choose a magazine to order through a program at his school. She says, “It was really cool because I had him write down his favorite choices, and I had him rate them. And then he added the numbers up.” The process resulted in Miles choosing to order a subscription to Highlights Magazine. Despite Dinah's pleased tone, Steve is quick to criticize the choice by saying that they've had that magazine in the past and either the old copies are still around the house or he may have thrown them out because they were an eyesore without any good information in them. Dinah's response, “I think he can read them in the library,” seems like a surrender.
With so much hostility apparently being directed toward Dinah, it seems almost like a relief when, at that point, Steve decides to take the children with him to run an errand. Steve is handy at fixing things, and he needs a part to repair the family washing machine. He and the children take a walk to a local hardware store that stays open late, and during that time he's relaxed and talkative with the children, but as soon as they get back to the house he seems to tense up. Dinah reports that the dinner she prepared is just about ready; some baked potatoes are cooling on the kitchen table. Steve walks into the kitchen and, seeming almost to be talking to himself but loud enough for Dinah to hear, asks, “Has the oven been turned off?” Then he answers his own question, with a loud “No.” After a pause, he adds, “Forget to turn off the oven a lot . . . a pet peeve of mine.”
Steve's behavior appears to be consistent with the negative emotion spillover findings from previous studies in which both members of a couple agreed that after more difficult days at work the stressed partner was impatient or touchy or disagreed with something the spouse said. The CELF coders assigned Steve high scores for response intensity, talking, and negative emotions. We can see that those thirty-second glimpses into his first hour at home accurately characterized what actually went on. That evening, he was an involved, curious, and active family member. It is also the case that at times Steve seemed impatient, was sometimes upset by and disappointed in his wife's responses to his questions, and admitted to being peeved. If we use Steve's family to tell us what negative spillover looks like, we don't find an overtly angry and aggressive man, scolding his children or openly insulting his wife. Instead, we find an active and involved father, whose interactions with his wife, in particular, often have a tense and even combative quality. It is easy to imagine Dinah feeling let down or hurt by his complaints and annoyed reactions during that first hour.
CORTISOL IN ANDRE'S AND STEVE'S FAMILIES
At the same time, what was happening to cortisol levels in our two exemplar families? Here, we focus on wives, not husbands, since results emerged only among them. Macy's marital satisfaction score—like her husband's—is well above the average for CELF wives. Also like her husband, Dinah has lower than average marital satisfaction; in fact, it is even lower than her husband's score. Fittingly, Macy consistently has higher morning cortisol than Dinah. She also tends to have lower evening cortisol than Dinah. In fact, on the evening that we described earlier, Dinah had one of the highest evening cortisol levels in the study. Since high morning cortisol and low evening cortisol are both markers of a steeper diurnal cortisol slope, the pattern that researchers have found to be “healthier” and more adaptive, it looks like both Macy and Dinah fit our expectations for how marital satisfaction might be linked with cortisol.
This finding seems to make sense when we remember how the husbands in our two families coped with their own workday experiences. Both fathers were under stress at work; this may have taken a toll on their abilities to connect with their families. Andre limited his conversation with family members, which contributed to a quiet, fairly low energy night at home. Steve, on the other hand, showed more tense and irritable behavior. Furthermore, the emotional valence of both husbands’ behavior may also be tied to their own cortisol levels, introducing another piece to the puzzle that considers how stress physiology and social behavior may be linked. In a separate analysis, we found a statistically significant negative correlation between the men's average amount of negative emotion expression in the home and their average diurnal cortisol slopes on the two observed weeknights of the study (r = -38). For husbands, steeper cortisol slopes were associated with less negative emotion expression with family members after work.21 This suggests that for someone like Andre—who did not display overt signs of annoyance or negative mood—taking a more withdrawn and emotionally neutral approach to interacting with family members may over time confer some of the health benefits associated with steeper or more adaptive cortisol slopes. For Steve, on the other hand, our data suggest that displays of criticism and hostility would ultimately be linked with a flatter and less healthy cortisol slope. Thinking about which family environment might be more “stressful” for the other partner, it is likely that Steve's overt touchiness and complaints presented more of a challenge for Dinah than Andre's quiet behavior did for Macy. While it is always difficult to extrapolate from only a few cases, it does seem fitting that Dinah showed a cortisol profile that is consistent with higher stress levels overall.
CONCLUSION
The way families cope with everyday stressors, like the transition from work to home, is not monolithic. Our research illustrates that individual differences in relationship quality (i.e., marital satisfaction) and personality (i.e., neuroticism) help to shape the patterns of behavior that individual working parents show in response to work stress. Negative emotion spillover and social withdrawal represent different ways of responding to a difficult day at work. Which strategy, if any, a person chooses may depend on a host of factors, including those explored in our research. Factors like marital satisfaction and the division of household labor appear to influence physiological stress as well, as our research on the hormone cortisol seems to suggest. While we found different patterns of results for husbands and for wives, the underlying theme is clear: in understanding the interplay between work and home, particularly the influence of job stressors on the family, it is important to consider an individual's gender, personality, and family circumstances. This makes the complexity and richness of the CELF study especially warranted as we explore the life of the family both inside and outside the home.
NOTES
1. E.g., Costigan, Cox, and Cauce 2003.
2. For reviews of the research literature summarized above, see Bumpus, Crouter, and McHale 1999; Perry-Jenkins, Repetti, and Crouter 2000; Repetti 2005.
3. Repetti 1989, 1994; Repetti and Wood 1997; Story and Repetti 2006.
4. More detail about the coding and inter-rater reliability process can be found in Wang, Repetti, and Campos 2011a. Inter-rater reliability is just one of several methodological issues that arise when coding naturalistic family interaction; a more in-depth discussion of these issues from the point of view of psychologists can be found in Repetti, Wang, and Sears 2011.
5. Wang, Repetti, and Campos 2011a.
6. Repetti 1989, 1994.
7. Ibid.; Repetti and Wood 1997.
8. Story and Repetti 2006.
9. Repetti and Wood 1997.
10. Costa and McCrae 1992.
11. Locke and Wallace 1959.
12. Readers are referred to the Wang, Repetti, and Campos (2011a) article for a detailed report of the neuroticism findings.
13. Sephton et al. 2000.
14. Saxbe 2009.
15. Saxbe, Repetti, and Nishina 2008.
16. Interested readers should consult Saxbe, Repetti, and Nishina 2008 for a more detailed description of these patterns in the data.
17. Kiecolt-Glaser and Newton 2001.
18. For more about these findings, see the original published report: Saxbe, Repetti, and Graesch 2011.
19. The next day Andre went on the school field trip with Peter's class. Since his evening interactions did not follow a day at work, we won't focus on them here.
20. Because Steve did not get home from work until very late on the first weekday of video recording and the CELF camera crew had already left, we focus here on the second weekday of film.
21. Wang, Repetti, and Campos 2011b.