Withdrawal From Coparenting Interactions During Early Infancy

Abstract

This study examines early withdrawal in the coparenting system, and the utility of a brief problem-solving discussion about coparenting responsibilities as a means for evaluating such withdrawal. One hundred and fifteen couples were evaluated both prenatally and at 3 months postpartum. During prenatal assessments, parents rated their personalities and completed marital assessments. After the baby arrived, they completed a negotiation task in which they discussed disputes about parenting roles and responsibilities, and interacted together with the baby in a triadic play assessment. Fathers’ but not mothers’ withdrawal during coparenting negotiations was associated with greater disengagement and less warmth during triadic play and with fathers’ feelings that mothers did not respect their parenting. Fathers’ but not mothers’ withdrawal during coparenting negotiations was also forecast by low ego resilience and by an increase in depressive symptomatology during the postpartum. As the negotiation task appeared to be an effective provocateur of withdrawal when confronting coparenting disagreement, it may prove useful for eliciting this aspect of coparental process in work with couples.

While virtually all mental health contacts with families of infants and toddlers engage mothers and occasionally consult fathers or other family caregivers, most routinely fail to assess the emerging coparenting alliance developing between the two parents (Feinberg, 2002; Fivaz-Depeursinge & Corboz-Warnery, 1999; McHale & & Cowan, 1996; McHale, Kuersten-Hogan & Rao, 2004; McHale, Khazan et al., 2002). While much of this neglect is attributable to the dyadic model saturating infant mental health practice (McHale, 2007a), it is also the case that coparenting assessments are hindered by a general lack of agreed-upon strategies for evaluating coparental functioning during early infancy. Self-report indicators of coparenting developed in research settings (Abidin & Brunner, 1995; McHale, 1997) do not offer useful norms or categorical thresholds denoting clinical distress levels, while creative observational assessments such as the Lausanne Trilogue Play (LTP) (Fivaz-Depeursinge & Corboz-Warnery, 1999) are not yet in widespread use.

The absence of coparenting conceptualizations in infant mental health practice cannot be over-stated, exemplified by the Diagnostic Classification of Mental Health and Developmental Disabilities of Infancy and Early Childhood, Revised Edition’s exclusively dyadic relationship focus. Indeed, even when coparenting issues have been targeted by infant-family scholars, withdrawal and disengagement from alliance-building efforts often go unattended. This is concerning, as disengagement is problematic when apparent very early in the burgeoning family group process (Cox, Paley, Burchinal, & Payne, 1999; Paley et al., 2005). Unfortunately, withdrawal is notoriously difficult to pinpoint from self-reports or observations of coparenting process, despite the prominent role it plays in stressful marital interactions (Christensen, 1987; Christensen & Heavey, 1990; Gottman & Krokoff, 1989; Stanley, Markman, & Whitton, 2002).

What this suggested to us, however, is that structured coparenting discussions may provide a useful window into early coparenting withdrawal. If discussions about caregiving differences activate withdrawal among new parents disposed to disengage emotionally, such withdrawal behavior should be linked to other indicators of co-parenting disengagement and discontent, assessed concurrently. Further, withdrawal should occur more often among parents showing certain risk signs for withdrawal, such as poor ego resilience, greater marital distress, or increased depressive symptomatology from the pre- to postpartum. If so networked, a case might be made that coparenting discussions hold value as diagnostic and treatment planning tools in therapeutic contacts with distressed families of young infants.

PAST RESEARCH ON WITHDRAWAL DURING HUSBAND-WIFE MARITAL INTERACTIONS

Emotional withdrawal is an active ingredient in marital dissatisfaction (Christensen, 1987; Christensen & Heavey, 1990; Gottman & Krokoff, 1989; Gottman & Levenson, 1992) and dissolution (Gottman, 1994). While both men and women withdraw from intense marital conflicts, withdrawal is far more common among men (Christensen & Heavey, 1993). Men certainly sometimes do demand (and women withdraw) when an issue is of concern and importance to them (Christensen & Heavey, 1990; Klinetob & Smith, 1996), leading some to offer a social structure explanation. Jacobson (1983, 1989), for example, proposed that partners with greater investment in maintaining the status quo (men, in the case of shared childcare) withdraw in order to avoid change.

Studies of marital withdrawal patterns following new parenthood transitions are few in number. Thorp et al. (2004), in one of the lone relevant studies, found that first-time mothers unhappy with the division of childcare labor at 8 weeks postpartum reported more pronounced demand-withdrawal patterns, with greater withdrawal by fathers following mothers’ demands escalating mothers’ stress levels. Yet we know nothing about which new fathers are most likely to withdraw from mothers’ demands, or about parallel withdrawal processes in mothers. Marchand and Hock (2000) reported that depressive symptoms predict conflict avoidance among both women and men, but did not specifically address withdrawal patterns. Breiding (2004) linked domineering marital behavior by husbands to wives’ depressive symptoms and poorer marital adjustment, but also did not assess withdrawal. Only Babcock et al.’s (1993) study of couples plagued by domestic violence specifically examined wives’ withdrawal, documenting male demand–woman withdraw patterns more frequently in violent than in nonviolent marriages.

PAST RESEARCH ON DISENGAGEMENT FROM EARLY FAMILY INTERACTIONS

Withdrawal is a term historically used by marital researchers to describe marital interaction processes (Christensen & Heavey, 1990; Gottman, 1994). Disengagement is a related term that family therapists use to describe families in which one or more family members has become emotionally detached from family commerce (Minuchin, 1974; Minuchin, Montalvo, Guerney, Rosman, & Schumer, 1967). Disengagement is also a concern of infant-family professionals helping new parents (Byng-Hall, 1995); clinically depressed mothers, for example, often exhibit detachment and emotional unavailability (Field, 1994; Murray & Cooper, 1996). More normatively, new fathers often feel excluded by the developing mother-infant bond. Some men respond to feelings of disenfranchisement by pressing their wives for more relationship time, while others channel their energies into caring for the baby (which Hackel and Ruble, 1992, found unsettling to certain mothers who did not welcome fathers’ intrusions into the early parenting role). Still other fathers respond by progressively distancing from the mother-baby dyad.

Who is most likely to disengage? This is a difficult question to answer empirically, in part because clinically disengaged and excluding families are usually under-represented in community research samples (McHale, Lauretti, Talbot, and Pouquette, 2002). Disengagement from the coparenting alliance toward the end of the baby’s first year does appear more common among couples encountering marital difficulties. In an observational study, McHale (1995) found that maritally distressed mothers and fathers exhibited more skewed patterns of engagement with their first-born infants (signifying gravitation toward disengagement by one partner and/or toward over-involvement by the other) than did nondistressed parents. The link between marital distress and skewed engagement obtained principally in families with daughters; in families with sons, maritally distressed parents were more prone to stay mutually engaged, albeit antagonistically (McHale, 1995). Moreover, skewed engagement during infancy predicted greater child anxiety in the preschool years (McHale & Rasmussen, 1998). Such data are important in showing that marital and coparenting systems function as related—but conceptually distinct—forces in families (Cohen & Weiss-man, 1984; Floyd & Zmich, 1991; Frank et al., 1991; Gable, Crnic, & Belsky, 1994).

Disengagement can be a very early emerging, and enduring, characteristic of the family’s core dynamic. Fivaz-Depeursinge and colleagues studied mothers’, fathers’, and infants’ engagement during face-to-face interaction in the LTP (Fivaz-Depeursinge & Corboz-Warnery, 1999; Fivaz-Depeursinge & Favez, 2006). They rated patterns of participation, organization, focal attention, and affective contact, documenting such signals as positioning of family members’ pelvises and torsos and gaze patterns. Carefully analyzing these features, they derived a categorical system specifying whether families maintained cooperative (coordinated on all fronts), stressed (coordinated for participation and organization but compromised elsewhere), collusive (coordinated for participation but noncoordinated for organization), or disordered alliances (the last akin to a disengaged family dynamic, with poor coordination along both participation and organization dimensions). Remarkably, forms of adjustment family members assumed during early infancy (3–4 months) were stable through the early toddler years (Favez, Frascarolo, & Fivaz-Depeursinge, 2006; Fivaz-Depeursinge & Corboz-Warnery, 1999). Moreover, recent work from Fivaz’s laboratory indicates that early family alliances can be predicted from couples’ interaction patterns while enacting a virtual first encounter with their baby during the pregnancy (Carneiro, Corboz-Warnery, & Fivaz-Depeursinge, 2006).

Clinically, risk for withdrawal during the early postpartum may be especially likely among parents lacking personal resilience and/or battling depression; parents overwhelmed by stress often succumb by closing down in defense, while many clinically depressed parents detach emotionally from partners and babies (Field, 1994). Yet while upsurges in depressive symptoms across the transition to new parenthood are tied both to parents’ marital satisfaction and to perceived support from partners (Simpson, Rholes, & Campbell, 2003), few studies have linked increases in depressive symptoms to marital or coparenting behavior. There is also little evidence that personal resilience is linked to marital or coparenting behavior, despite conventional wisdom that resilient individuals are better equipped to weather relationship challenges (Belsky & Hseih, 1998; Bradbury & Karney, 2004).

The trait most closely aligned with resilience studied by marital researchers is neuroticism, proneness toward emotional instability that prompts ineffective coping under duress (Costa & McCrae, 1992). Surprisingly, however, Karney and Bradbury (1997) found no evidence linking marital behavior to neuroticism. Belsky and Hseih (1998) did show that greater declines in self-reported marital satisfaction from 10 to 60 months postpartum were predicted by higher neuroticism, but did not report their data concerning the interrelationship between early marital and coparenting behavior and neuroticism. Neither study directly targeted withdrawal dynamics.

Only two studies directly examine personal resilience and coparenting. Van Egeren (2003) found only meager evidence that stronger ego development in parents of 3-month-olds helped offset coparenting problems. Although men whose wives reported more mature ego development evaluated early coparenting more positively, this was the only linkage Van Egeren found between ego development and perceived coparenting. She did not study coparenting behavior. Talbot and McHale’s (2004) report is the only one to link observed coparenting behavior to personal resilience (operationalized as flexibility), showing that father flexibility attenuates relations between marital and coparenting distress. Mothers’ flexibility was not protective, however, actually heightening the extent to which coparenting harmony declined in the face of lower marital quality. Once again, neither of these studies specifically targeted withdrawal behavior.

Summary and Prospectus

Early disconnection in the family’s coparenting alliance is important. It does not simply fade with time, and predicts later adjustment difficulties for children. Yet studies of early coparenting withdrawal are uncommon, as there is no field-tested paradigm for evoking this dynamic. As a result, factors prompting such withdrawal are poorly understood. It is not known whether postpartum increases in depressive symptoms or poor personal resilience breed early withdrawal and disengagement; Van Egeren’s (2003) and Talbot and McHale’s (2004) reports suggest only that flexibility, broadly defined, may foster better coparenting adaptation, but even here evidence is equivocal. Virtually nothing is known about families where mothers withdraw from couple interaction in the early postpartum, although Babcock et al.’s (1987) work hints that women’s withdrawal may owe to a partner’s impulsivity, temper, or aggression.

To advance our understanding of early withdrawal in the coparenting alliance, we needed a paradigm to evoke withdrawal behavior during observational assessments. Prior research suggested that problem-solving discussions about caregiving work might serve this purpose and, if so, enable a search for predictors and correlates of such withdrawal. We hence outlined the following hypotheses:

1. If withdrawal from coparenting discussions is problematic, it should hamper parents’ ability to attain satisfying mutual consensus through such discussions.

2. If indicative of a deeper structural family theme, withdrawal from discussions about coparenting work should correlate with disengagement during observed family (mother-father-baby) interaction and with parents’ impressions that their partners do not respect their parenting efforts.

3. Greater ego resilience should protect against withdrawal, with resilient parents better able to endure differences in perspective with their partner—inevitable when coconstructing a new coparenting alliance. Extrapolating from findings that aggression triggers withdrawal, we speculated that withdrawal would also be heightened among parents, especially women, whose partners reported poorer self-control. And based on findings linking withdrawal to marital distress and depression in clinical populations, we hypothesized that withdrawal would surface more often when there was preexisting marital distress or heightened post-partum depressive symptomatology.

We also examined whether men’s and women’s withdrawal had similar correlates in families with male and female infants, given prior research hinting that coparenting withdrawal in distressed families is more likely when the baby is a girl.

METHOD

Participants

Participants were 115 married couples (women: 87% Caucasian, 13% of African, Hispanic, or Asian descent, or of mixed race, men: 89% Caucasian, 11% of African, Hispanic, or Asian descent, or of mixed race), all contributors to a federally funded longitudinal study of the transition to new coparenthood. Couples were recruited from prebirth classes in an urban Northeastern city. Virtually all participants (95% of fathers and 93% of mothers) indicated that the pregnancy had been planned. Couples had been married, on average, 3.7 years (SD=2.29, range=0–11), and together as a couple for 6.9 (SD=3.49, range=1–20). Average age of mothers was 31.7 (SD=5.1, range=22–47 years) and fathers 33.3 (SD=6.0, range=21–49 years). Median family income in 2002 US dollars fell in the 70,000–75,000 range, with a sample range from 30,000–35,000 to over 100,000. Forty-nine percent of the infants born were girls and 51% boys.

Procedure

During the pregnancy’s third trimester, couples visited a university-based Family Study Center. There, they completed two “revealed difference” problem-solving discussions in which they worked for 10 minutes (5 minutes per problem) to make headway on one area the husband indicated was problematic (e.g., handling family finances, demonstrations of affection), and one the wife indicated was problematic (Sagrestano, Christensen, & Heavey, 1998, outline the importance of using one husband-identified and one wife-identified topic in such assessments). Following each discussion, both partners immediately completed a Post-Discussion Questionnaire (PDQ) (cf. Talbot & McHale, 2004) evaluating their own and their partner’s conduct during the discussion. At session’s end, they took survey instruments home to complete and return.

Three months after the babies were born, couples were visited at home (97% retention rate) at times when the babies were typically awake and alert. Partners engaged together in a “Who Does What” (WDW) discussion of coparenting issues, designed to afford insights into the couple’s negotiation of conflict and regulation of negative affect, and interacted together with their baby in the LTP (Fivaz-Depeursinge & Corboz-Warnery, 1999) allowing an evaluation of the family’s capacity for play, sharing of, and regulation of positive affect. They also completed paper-and-pencil inventories.

The WDW discussion adhered to Schoppe-Sullivan, Mangelsdorf, Frosch, and McHale’s (2004) procedure using Cowan and Cowan’s (1992) “WDW” instrument. With their infants alongside them, each parent individually completed the 24-item WDW lists of childcare roles (diapering, feeding, making doctor’s appointments, middle of the night needs). Parents rated items on 1–9 scales (1=she does it all; 5=we share equally; 9=he does it all). After completing WDWs independently, partners were asked to share their responses and reach consensus on each item. Our procedure differed from Schoppe-Sullivan’s in two respects: (a) neither partner was told in advance that they would be divulging their original answers, and so the request caught parents unaware; and (b) discussions were not halted after 5 minutes; couples took, however long needed to complete the task (M=13.2 minutes; range: 6–22 minutes).

The LTP is the lone validated assessment procedure available for evaluating triadic (mother-father-infant) relationship dynamics during earliest infancy. It is divided into four distinct segments, which in this study took approximately 2 minutes each (8 minutes total). The LTP begins with two “2 plus 1s” when first one parent, and then the other, engages with the baby while the second parent is “just present.” These are followed by a three together, during which all three family members engage together in play at the same time, and a “2 plus 1” in which parents engage together as the baby assumes a third-party position. The family’s manner of handling the triangular interactions together is then evaluated.

Measures

Withdrawal From the Coparenting Alliance at 3 Months Postpartum

Withdrawal during the discussion of WDW child-care

Both men’s and women’s withdrawal behavior was evaluated during the consensus portion of the WDW discussion. Ratings were provided by the first author (D. B. E.) working with an advanced graduate student in clinical psychology. Judges, who remained blind to all other family process data for the study, began the evaluation process by first watching a subset of 20 randomly selected tapes together. Jointly, they determined preliminary ratings from 0 (altogether absent or evident only fleetingly) through 3 (exhibited for prolonged periods of several seconds at a time, on multiple occasions throughout the session) for those cases, actively discussing cases on which their preliminary impressions disagreed. They then took to rating the full cadre of 115 cases independently, although 20% (23 families) were rated jointly. Intraclass correlations (ICCs; see Shroutt & Fleiss, 1979) were monitored to track interrater reliability and help guard against observer drift. ICCs for the WDW ratings, and for all of the study’s other observational measures, are provided in the text below.

For the WDW task, ratings of Withdrawal were generated separately for mothers (M=0.56, SD=0.91, range=0–3, ICC=.79) and for fathers (M=0.52, SD=0.92, range=0–3, ICC=.82). These ratings captured the extent to which each person showed tangible evidence of drawing back from active communication and collaboration. Scores of 0 were assigned when the person was generally collaborative and exhibited no sustained withdrawal or nonparticipation. Nonzero scores were assigned only if parents participated in limited fashion, repeatedly offered little or no reaction when confronted with dissonant views, and/or permitted the partner to unilaterally determine the “consensus” coparenting ratings. As did Fivaz-Depeursinge and Corboz-Warnery (1999) and Julien et al. (1989), coders also sought out certain behavioral indicators to support judgments about whether parents were truly acceding to and distancing from the partner. These included drawing and maintaining a physical distance prohibiting joint review of surveys, bodily posturing away from the partner, and/or repeated diversion of eye contact. Nonzero scores, then, signified palpable evidence that significant withdrawal behavior was being shown, from episodically to throughout.

To help establish whether Withdrawal, as evaluated, was a uniquely disruptive process in the context of the WDW task, raters also estimated three other dimensions of couple process, as well as success in reaching convincing mutual consensus. 0–3 scales were also used for each of these measures:

Positive Affect (rated for the couple as a unit; M=1.92, SD=0.97, range=0–3, ICC=.78), an index of pleasantness and positive task orientation of the discussion. In low-scoring couples, expressions of warmth, affection, and humor were absent, smiles were rare, facial expressions neutral or pained, and discussion atmosphere serious, joyless, and/or somber. In higher scoring couples, there was a congenial and upbeat outlook even as the partners disagreed. They sometimes infused humor into discussions, made eye contact, smiled, laughed and joked episodically, and (on rare occasions) used supportive touch.

Negative Affect (also rated for the couple as a unit; M=0.51, SD=0.84, range=0–3, ICC=.85), an index of partners’ exhibition of overt irritability or annoyance. Low-scoring couples refrained from expressing overt annoyance; although sometimes pensive, they avoided hostile commentary, exaggerated frowns, and scowls. Moderate-scoring couples conveyed some irritability or annoyance facially, verbally, or both. In high-scoring couples, anger was poorly contained and parents occasionally even stood and moved about to regain composure.

Defensiveness (rated separately for mothers; M=0.65, SD=0.89, range=0–3, ICC=.65 and for fathers, M=0.59, SD=0.95, range=0–3, ICC=.70) denoted parental responses to perceived criticisms from the partner by issuing denials, by making preemptive remarks after an initial criticism (“I’ll bet I’m low on all of these, huh?”), and/or by responding to an apparent difference in perspective with a counterattack. Low scores reflected an absence of such behaviors, while high scores signified multiple instances of defensive maneuvering.

Couple Consensus (rated for the couple as a unit; M=2.05, SD=0.86, range=0–3, ICC=.85) captured the success and convincingness of the process of negotiation and resolution of disagreement. Couples receiving low scores exhibited either pervasive inability to reconcile differences in perspective or pseudo-consensus established absent any meaningful discussion. Higher scores were assigned when couples mutually weighed differences and came to joint agreements on most or all issues they viewed differently.

Disengagement during triadic play

A second, blind team of judges (postdoctoral and advanced graduate students in clinical psychology) trained by the second author (J. P. M.) evaluated LTP sessions for evidence of disengagement from the play interaction. In the context of the LTP, disengagement (M=2.18, SD=1.12, range=1–5, ICC=.73) signified sustained “absences” from contact with the baby. Such absences could be exhibited either posturally (maintaining a closed-off posture, or with body askew, rather than openly facing the baby), or through patterns of attending (sitting expressionless and with a vacant look either toward the partner-baby interaction or, more commonly, elsewhere in the room, for a prolonged span of several seconds). In the coding system used to evaluate the interactions, disengagement was assessed for the coparental team as a unit rather than each partner individually. The second indicator relevant to disengagement was overall warmth (M=4.47, SD=1.24, range=2–7, ICC=.70) during the interaction. In high-warmth families, mothers and fathers were in regular and positive affective contact with one another as well as with the baby; lower scoring families exhibited no positive mother-father affective contact and/or muted warmth in parent-child dyadic subsystems.

Perceived respect as a parent by the coparenting partner

Parents completed Abidin and Brunner’s (1995) Parenting Alliance Measure (PAM) to indicate the degree to which they felt validated as a parent by their partner. Validity studies find PAM scales to be correlated in expected directions with measures of parenting stress and family and marital functioning (Konold & Abidin, 2001). The instrument’s internal consistency is .97, with a 4–6 week test-retest reliability of .80. We used the instrument’s Respect scale (M for mothers=13.88, SD=1.52, range=7–15; M for fathers=13.65, SD=13.65, range=8–15), comprised of items including “During pregnancy, my child’s other parent expressed confidence in my ability to be a good parent” and “My child’s other parent believes I am a good parent.”

Hypothesized Predictors of 3-Month Withdrawal

Prenatal ego resilience and self-control

Ego resilience and self-control were assessed via parent self-report on the California Psychological Inventory (CPI), a widely used and well-validated instrument (Gough, 1990). Parents scoring highly on the CPI’s Ego Resilience subscale (Klohnen, 1996) were those reporting open-mindedness, perceptiveness, interest in understanding why others behave as they do, and openness to viewpoints different from their own. Across several samples, α coefficients for this scale have ranged from .81 to .88. ER scores are related to overall adjustment in the anticipated direction on the Index of Adult Adjustment (Klohnen, 1996). Parents scoring highly on the CPI’s Self-Control scale (Gough, 1990) describe themselves as calm, mature, dependable, and with good control of their own feelings, while lower scoring individuals do not attribute these qualities to themselves. In this sample, the mean Ego Resilience score for mothers was 19.2 (SD=3.44, range=11–26) and for fathers 19.2 (SD=3.69, range=7–26). The mean Ego Control score for mothers was 23.9 (SD=5.4, range=2–34), and for fathers 23.12 (SD=5.7, range=10–35).

Prenatal marital distress

Indices of prenatal marital distress were formed separately for mothers and fathers. For each parent, the index was comprised of one conflict, one withdrawal, and one marital satisfaction indicator. Conflict and withdrawal indicators were formed from clinical ratings of the revealed difference discussions augmented by parents’ postdiscussion ratings of their own and their partners’ conduct during the interactions. For the clinical ratings, a third team of blind coders (advanced doctoral students in clinical psychology) rated couple discussions from videotape using a marital interaction coding system developed and reported by Cox, Owen, and Lewis (1989; Paley et al., 2005). Coders had 20 hours of training, reviewing interactions from prior research studies and rating them with the third author (J. A. T.). Raters were blind to all study hypotheses and background information about the couples. Dimensions relevant to this report, all rated on 1 (low) to 9 (high) scales, included:

Egalitarian Power (M=6.42, SD=1.79, range=2–9, ICC=.68), rated for the couple as a unit. This rating captured the extent to which both partners demonstrated equal and unencumbered voice during discussions. In high-scoring couples, each partner convincingly and openly expressed opinions, whether consenting or dissenting. In lower scoring couples, one partner dominated exchanges or there was lack of leadership altogether.

Overt Conflict (M=3.99, SD=1.97, range=1–9, ICC=.89), also rated for the couple as a unit. This variable assessed the degree of openly expressed, irreconcilable conflict. High scores signified high conflict and no movement toward reconciliation; moderate scores overt conflict with some resolution; and low scores either no overt conflict, or conflict aired and dispensed with effectively.

Withdrawal, rated for each partner individually (M=2.78, SD=2.03, range=1–9 for men, ICC=.81; M=2.30, SD=1.74, range=1–9 for women, ICC=.83). This variable assessed the extent to which each partner curtailed engagement in the discussion by breaking off eye contact, orienting away from the partner, increasing/maintaining physical distance, giving up in the discussion, and being unresponsive. High scores signified clear and more frequent evidence of withdrawal from discussions, while low scores indicated minimal or no evidence of such withdrawal.

As the Egalitarian Power and Overt Conflict ratings were highly intercorrelated (r=– .60, p<.001), they were combined to form an overall observed couple Conflict index, while the husband and wife Withdrawal ratings were kept separate on theoretical grounds.

These Conflict and Withdrawal indices were supplemented by each partner’s 1 (low) through 9 (high) PDQ ratings of their own and their partner’s conflict and withdrawal behavior during the interactions. PDQ items significantly associated with the observed Conflict index included both partners’ responses to the items “I (my partner) was demanding during the discussion” and “I (my partner) was critical.” These items were added together, and the composite score was standardized. This standardized score was then added to the (also standardized) Conflict index to create a final summary measure called Marital Conflict. A like process was followed for husband and wife Withdrawal scores. PDQ items associated with the two observed Withdrawal indices were (a) both partners’ responses to the item “I (my partner) was withdrawn” and (b) wives’ responses to the items “I (my partner) was avoidant.” PDQ items associated with husbands’ observed Withdrawal were summed, standardized, and then added to men’s (also standardized) observed Withdrawal score to create Husband’s Withdrawal; the same process was followed for PDQ items associated with women’s Withdrawal in forming a Wife’s Withdrawal score.

Besides the Marital Conflict and Husband/Wife Withdrawal measures, the third indicator used to form the final husband and wife Marital Distress scores was reported marital satisfaction on Locke and Wallace’s (1959) Marital Adjustment Test (MAT). The MAT is a widely used instrument associated in expected directions with both reported and observed marital distress (Gottman, Markman, & Notarius, 1977). Higher scores on the MAT signify greater satisfaction, while scores below 100 denote marital distress. In our sample, the average prenatal MAT score for mothers was 124.1 and for fathers, 123.6. Five percent of mothers and 7% of fathers scored in the clinically distressed range during the pregnancy.

To form the final Marital Distress index for each parent, MAT scores were adjusted to reflect dissatisfaction (multiplying them by −1) and converted to z scores. Marital Conflict and Husband and Wife Withdrawal were also converted to z scores. A Marital Distress score for women was then formed by summing Marital Conflict, Wife’s Withdrawal, and wife’s MAT dissatisfaction. The parallel score for men included Marital Conflict, Husband’s Withdrawal, and husband’s dissatisfaction. Internal consistencies for both Marital Distress indicators were good (α=.83 for husbands and .87 for wives). Increases in depressive symptomatology between pregnancy and 3 months. We operationalized increases in depressive symptoms as a discrepancy score between parents’ 3-month and earlier prenatal reports on the 20-item Center for Epidemiological Studies Depression Scale (CES-D; Radloff, 1977). The widely used CES-D shows excellent internal consistency (α>.85) and test-retest stability (r>.5; Radloff, 1977). Respondents report how often during the past week they experienced various symptomatic events, from “rarely or never” (less than once daily) to “most or all of the time” (5–7 days). Scores of 16 or higher signify symptoms of clinical concern. Prenatally, 32% of mothers scored 16 or above (M=12.26, SD=8.79, range=0–50); by 3 months, this percentage fell to 20%. Indeed, mothers showed on average an almost 5-point decline (M=4.75) from pregnancy to the 3-month follow-up. However, 20% of mothers showed upsurges in symptoms following the transition to parenthood, ranging from 1 to 19 points. Among fathers (M CES-D score during pregnancy=9.16, SD=7.16, range=0–44), 24% showed similar increases, ranging from 1 to 15 points.

RESULTS

Below, we outline evidence that the WDW paradigm effectively elicited couple distress, particularly withdrawal. We then address correlates and predictors of such withdrawal.

The Significance of Withdrawal During the WDW Discussions

The WDW procedure was effective in prompting distress behavior. Seventy percent of all couples received nonzero ratings on at least one 0–3 point distress scale (Negative Affect, his/her Defensiveness, his/her Withdrawal). More than a quarter of the sample (32 of the 115 couples, or 28%) showed distress on three, four, or all five of these indicators.

As anticipated, greater Withdrawal by both mothers and fathers interfered with effective consensus building. To establish whether Withdrawal was a uniquely important process in the context of the WDW, we conducted multiple regression analyses first entering Positive Affect, Negative Affect, and mother’s/father’s Defensiveness as a block in predicting Couple Consensus scores. As a predictor set, these couple process variables explained 43% of the variance in the couple’s ability to reach convincing consensus during the WDW discussions, R²=.43, F(4, 110)=24.45, p< .001. Mother/ father Withdrawal scores, then entered on a second step, explained an additional, statistically significant 17% of the variance in consensus scores (R²Δ=.17, FΔ=26.35, p<.001), with both men’s Withdrawal (β=−.25, p<.001) and mother’s Withdrawal (β=−.36, p<.001) making a significant contribution in the prediction of consensus building.

Hence, those partners unable to reach convincing mutual consensus were indeed more likely to withdraw from WDW discussions. Noteworthy Withdrawal (scores of 1, 2, or 3 for one or both partners) was seen in half (58 of 115) of all families; in 10% (12 of the 115 families), both parents showed withdrawal. Overall, 31% of all study participants (35 of 115 fathers and 35 of 115 mothers) exhibited noteworthy signs of withdrawal during the task.

Having established the significance of withdrawal in the context of the WDW assessment, we next turned our attention to linkages between WDW withdrawal and the concurrently measured indicators of coparenting adjustment. Before completing the relevant analyses, given prior research that coparenting withdrawal in distressed families may be more common when the baby is a girl, we first examined whether child gender influenced the overall patterning of relationships between WDW withdrawal, the time couples invested in completing the WDW task, and the indicators of self-reported and observed coparenting. Of interest was the statistic Box’s M, summarizing the degree of similarity in variance-covariance matrices for families of boys and families of girls (Tabachnik & Fidell, 1996). A statistically significant Box’s M suggests that the variance-covariance matrices are different and that correlations should be examined separately for the two groups. However, findings fell short of statistical significance (Box’s M=12.28, ns), indicating that data for the two groups could be examined together.

Associations Between WDW Withdrawal and Disengagement From Family Play

As predicted, men’s Withdrawal from WDW discussions was indeed associated with both greater disengagement and lower warmth during LTP play interactions. Table 1 summarizes these and other significant correlations presented in this and subsequent sections. Although analyses indicated that WDW-LTP associations for boys and girls could be examined together, we did note a conceptually interesting gendered pattern—the WDW-LTP linkages found for fathers were statistically significant only in families with daughters (r=.34, p<.01 for the high WDW withdrawal-high LTP disengagement connection; r=−.26, p<.05 for the high WDW withdrawal-low LTP warmth connection). Father Withdrawal during the WDW was not significantly associated with either greater LTP disengagement (r=.18, ns) or lower LTP warmth (r=−.19, ns) in families with sons.¹ This theoretically interesting pattern noted, differences in magnitudes of association for families with boys and families with girls were not statistically significant (Z_diff=.99, ns, for the high withdrawal-high disengagement boy-girl comparison, and Z_diff=.40, ns, for the high withdrawal-low warmth boy-girl comparison).

TABLE 1.

Associations Among Main Study Variables

Variables	1	2	3	4	5	6	7	8	9	10	11	12	13	14
3-month outcome measures
1.WDW withdrawal —men	—
2. WDW withdrawal —women	−.03	—
3. Disengagement during triadic play	.31***	.04	—
4. Warmth during triadic play	−.29**	−.18*	−.53***	—
5. Respected as parent by partner—men	−.18**	.12	−.08	.09	—
6. Respected as parent by partner—women	−.07	−.05	−.05	.05	.25**
Prenatal predictors
7. Ego Resilience —men	−.25**	−.17	−.09	.15	.24**	.23**	—
8. Ego Resilience —women	.01	−.04	−.14	.05	.03	.20**	.14	—
9. Ego Control—men	−.03	−.17*	−.09	.12	.00	.00	.28***	.11	—
10. Ego Control —women	.00	−.14	−.11	−.03	.01	.08	.13	.17*	.16	—
11. Marital Distress —men	.19*	−.04	.24**	−.17*	−.32***	−.44***	− .24**	−.21**	−.25**	−.09	—
12. Marital Distress —women	.03	−.01	.16	−.12	−.37***	−.28***	−.20**	−.35***	−.33***	−.18*	−.83***	—
13. Pre-Post increase in Depression —men	.25**	−.11	.00	.01	.13	.04	.09	.05	.14	.02	−.17	−.10	—
14. Pre-Post increase in Depression —women	.11	.01	.11	−.08	−.13	−.10	.02	.13	.04	.15	.02	−.07	−.04	—

^*p<.10

^**p<.05

^***p<.01.

These patterns of associations were not nearly as pronounced for women. Among women, withdrawal during the WDW was only marginally related to observed warmth during the LTP interactions (r=−.18, p<.10) and unrelated to LTP disengagement (Table 1). No gendered distinctions qualified the effect.

Linkages Between WDW Withdrawal and Sense of Being Respected as a Parent

As in the WDW-LTP analyses, we found the predicted significant negative relationship between WDW withdrawal and reported feelings of respect from the partner. Again, however, the finding obtained for fathers only. Withdrawing fathers were less likely to believe that mothers respected them as parents (r=−.18, p=.05). The same was not true of withdrawing mothers, among whom WDW withdrawal was unrelated to beliefs about fathers’ respect for their parenting competencies (r=−.05, ns).

Prenatal Predictors of Withdrawal During the WDW Interaction

As with the concurrent analyses, we conducted preliminary analyses to determine whether data for families of boys and girls could be interpreted together, examining child gender effects on the pattern of relationships between pre- and pre-to-post natal predictors (Marital Distress, Ego Resilience, Ego Control, and upsurge in depressive symptoms), time to complete the WDW task, and withdrawal behavior during the WDW. Findings again indicated that there was justification in considering boy and girl data together (Box’s M=14.10, ns).

As anticipated, men’s Ego Resilience predicted later father Withdrawal during the WDW coparenting discussions, with more resilient men significantly less likely to withdraw (Table 1). Father Withdrawal was also predicted marginally by men’s prenatal Marital Distress scores (r=.19, p<.10), but not by men’s own or by women’s Self-Control (Table 1). Mothers’ Withdrawal during the WDW was not, however, forecast by any of these factors save for a marginally significant association with lower Self-Control by men, in the hypothesized direction (r=−.17, p<.10). Mother Withdrawal was not predicted by women’s prenatal Ego Resilience or women’s prenatal Marital Distress (Table 1).

Increases in Depressive Symptomatology and WDW Withdrawal

As anticipated, increased depressive symptomatology was significantly correlated with more pronounced withdrawal behavior by men during the WDW but, contrary to expectations, not with greater WDW withdrawal for women (Table 1). The absence of a connection between increases in depression and WDW withdrawal for women remained even after controlling for the baseline level of women’s depressive symptoms during the pregnancy (pr=.07).

DISCUSSION

Our aim in this study was to examine whether meaningful patterns of withdrawal can be elicited from a focused discussion of early coparenting and, if so, what such patterns might tell us about the individual and family adjustment of the withdrawing partners. We found that a WDW discussion was sufficiently evocative to prompt couple dynamics—withdrawal, in particular—that can be difficult to discern from family play interactions. Fathers who withdrew from WDW discussions at 3 months post-partum were less ego-resilient, and tended to be partnered in marriages already showing distress signs before babies arrived. They reported an upsurge of depressive symptoms during the early postpartum months, felt less respected by their wives as parents, and cocreated with their wives and babies triadic relationship patterns characterized by greater disengagement and lower warmth—especially when coparenting baby daughters. These WDW withdrawal-triadic process linkages were not statistically significant in families with sons, echoing McHale (1995). In short, fathers who withdrew from coparenting discussions at 3 months postpartum felt themselves to be, and looked to be, in less supportive coparenting alliances. Further, withdrawing fathers’ self-characterization of not rolling well with the punches squares well with Talbot and McHale’s (2004) finding that men are more ego-resilient in families where marital distress does not destabilize coparenting alliances.

Withdrawal by mothers, our data suggested, did not owe to any of these factors, save for a marginally significant association linking women’s WDW withdrawal to poorer self-control as reported by their husbands. We had hypothesized that less resilient, increasingly depressed, and/or maritally distressed women might too withdraw more from WDW discussions. But our data indicated that they do not. Moreover, women’s withdrawal from WDW discussions did not portend greater disengagement during play with the baby to the same extent as did men’s withdrawal. Hence, we made only little headway in identifying the significance of withdrawal from coparenting discussions by mothers, although we believe our data provide a rather firm handle on its meaning for men. These things said, we advance these findings and interpretations with tempered enthusiasm, for several reasons.

First, we underscore that ours was a community sample with relatively few families experiencing pronounced clinical levels of distress. The absence of any connection between increases in postpartum depressive symptoms and mothers’ withdrawal during the WDW may have owed to the absence of clinically depressed participants in this study, as we might anticipate that many clinically depressed mothers would manifest withdrawal both from marital engagement and from babies. We also note, although, that MacEwen, Barling, and Kelloway (1992) found self-reported symptoms of depression to predict only anger in, but not withdrawal from, marital interactions. Second, the relatively small proportion of couples experiencing substantial marital distress may have accounted for the relatively weak cross-time association we found between marital distress and adjustment for men, and absence of such a connection for women. More compelling links, and perhaps differently patterned findings, may surface in a sample comprised of clinically distressed families. It is also possible that challenges coders encountered in evaluating egalitarian power in some expectant couples may have weakened the marital index developed for this study, although the internal consistencies of the overall marital composites used were themselves quite good.

Third, the proportion of families (just over one half) in which convincing patterns of WDW withdrawal were shown by one or both partners may seem low, given the position of some writers (e.g., Stanley et al., 2002) that features of demand-withdraw patterns can be detected in nearly all couples to some extent. The system we developed did not scrutinize couples for every gaze aversion, uncomfortable posture shift, or slowness to respond as do some researchers using microanalytic criteria, but instead sought to establish whether parents showed a repeated pattern of such behavior either at a low, moderate, or convincingly high level. As such, we admittedly may have overlooked more subtle withdrawal patterns in many families, although we believe that the level of scrutiny we brought to bear mirrors that of clinicians in the field. For the reasons enumerated above, we conclude only cautiously that men’s withdrawal from early coparenting discussions may have different sources and signify something different than women’s withdrawal in the early emerging family dynamic.

Finally, we also underscore additional limitations of a sample dominated by middle class white couples. Working class men follow women’s parenting leads more closely than do middle class men (Entwisle & Doering, 1981), leaving it unclear whether predictors and correlates of coparenting withdrawal would be similar in lower socioeconomic families.

Implications for Practice

Family clinicians have long appreciated the dangers of partner withdrawal, but few are in a position to formally evaluate early family interactions with the scrutiny of the laboratory studies that have provided guideposts for measuring couple (Clements, Stanley, & Markman, 2004; Gottman, 1994; Gottman & Krokoff, 1989; Markman & Hahlweg, 1993) and triadic (Fivaz-Depeursinge & Corboz-Warnery, 1999) processes. This study augments prior laboratory research by providing a pointed, useful means for helping establish whether disengagement has begun seeping into early family process. Discussions of shared childcare work provide a focused entry point in evaluations of early coparenting alliances and their dynamics. They afford concrete, in vivo examples of the very patterns triggered when parents engage under duress. Withdrawal during WDW discussions by men can be traced to greater disengagement during family interactions, perhaps especially so when the couple’s infant is a girl. And withdrawing fathers are less ego-resilient than other men and view their partners as unsupportive or invalidating of their early parenting efforts, considerations important for clinical interventions when father withdrawal is at question.

For mothers, by contrast, WDW withdrawal does not have the same connotations. Data from this study only suggest that clinicians who note maternal withdrawal from coparenting discussions might be on the lookout for a tendency to disengage rather than engage with a partner who exhibits poorer self-regulation when upset, although even here this finding was significant only at a trend level. Given this caveat and the sampling limitations noted above, these propositions are advanced only as general guidelines that may be irrelevant for any given couple, but that do provide creative new hypotheses for clinicians working with new parents struggling to establish a collaborative coparental alliance.

In using WDW discussions during clinical evaluations of coparental distress, clinicians might consider use of videotaped playback of conversations to highlight themes, alone or in combination with interaction guidance (e.g., McDonough, 1995). Couples’ reactions to the exercise could also simply serve as points of entry for targeting the dynamic of withdrawal in their coparental alliance, before such withdrawal becomes an ingrained component of the family process. Field testing of the WDW procedure is needed both in studies targeting its practical value as a clinical instrument and in the context of initiatives giving more comprehensive attention to the value of explicitly assessing coparenting dynamics even during the earliest postpartum months (see Fivaz-Depeursinge & Favez, 2006; McHale, 2007b). In particular, future work with this paradigm and assessment scheme is needed to establish longitudinal correlates of coparental withdrawal for both children and families. Given mounting evidence for the marked stability of coparental solidarity through time, and for the disruptive effect that lack of solidarity has on child adjustment, devoted attention to early coparenting and family group dynamics is a particularly important direction for family research and practice.