|Home | About | Journals | Submit | Contact Us | Français|
This study reports the psychometric properties of a multi-domain measure of the coparenting relationship in dual-parent families.
152 couples participating in a transition to parenthood study completed the Coparenting Relationship Scale and additional measures during home visits at child age 6 months, 1 year, and 3 years.
Psychometric and construct validity assessments indicated the measure performed satisfactorily. The 35-item measure demonstrated good reliability and strong stability. Subscales measuring theoretically and empirically important aspects of coparenting (coparenting agreement, coparenting closeness, exposure of child to conflict, coparenting support, coparenting undermining, endorsement of partner’s parenting, and division of labor) demonstrated good reliability as well. A 14-item brief overall measure showed very strong associations with the overall measure. Relations of the full scale with a measure of social desirability were weak, and the full scale was positively associated with positive dimensions of the dyadic couple relationship (love, sex/romance, couple efficacy) and inversely associated with negative dimensions (conflict, ineffective arguing)—as expected.
This initial examination of the Coparenting Relationship Scale suggests that it possesses good psychometric properties (reliability, stability, construct validity, and inter-rater agreement), can be flexibly administered in short and long forms, and is positioned to promote further conceptual and methodological progress in the study of coparenting.
Family research, like all science, develops as a dynamic interchange among advances in methods, observed data, and theory. For example, observations of increasing rates of divorce in the United States in the second half of the 20th century prompted family researchers to examine the consequences of divorce on children (Emery, 1982; Hetherington, Cox, & Cox, 1982). Divorce researchers seeking a framework to understand the consequences of divorce utilized notions from family systems theory and clinical practice that parents in a healthy family system form a coherent sub-system or coalition. In the midst of conflict and divorce, this sub-system often broke down and parents pursued their conflict by competing for influence over or affection from the child, or at least failed to maintain a parental coalition that served to provide leadership for the family (Ahrons, 1981; Durst, Wedemeyer, & Zurcher, 1985; Howe, Bishop, Armstrong, & Fein, 1984). This work led to a focus on coparenting conflict in divorcing families as an important influence on children’s post-divorce adjustment, and measurement focused on issues such as triangulation of the child into ongoing parental conflict, loyalty struggles, and coordination of unified parenting policies and practices across households.
This conceptual framework was brought into the study of non-divorced families by researchers such as McHale and Belsky and their colleagues (Belsky, Crnic, & Gable, 1995; McHale, 1995, 1997). These and other authors developed measures of coparenting (Margolin, Gordis, & John, 2001; McHale, Kuersten-Hogan, & Lauretti, 2001) that have included a range of constructs that focused on 2-parent, nuclear families. Incorporating dimensions of conflict and triangulation from the earlier work with divorced families, these measures also often included an assessment of coparental support—broadly considered as the positive endorsement of one’s partner’s parenting competencies, emotional and instrumental support in the face of parenting strains and stress, and recognition for both the partner’s and the couple’s achievements and successes in parenting (Abidin & Brunner, 1995). In addition, other aspects of positive or negative coparenting are sometimes assessed, such as cooperation, promotion of family integrity, satisfaction with the division of childrearing labor (Cowan & Cowan, 1988), and covert undermining of parental authority. A weakness of the available measures is that many seem to capture some aspects of coparenting relations, but there has been a lack of an overarching conceptual specification of the coparenting relationship that has driven the development of the coparenting dimension scales. To varying extents, the constructs in most coparenting measurement instruments appear to have emerged based on a combination of intuition and empiricism (e.g., labeling emergent factors in exploratory factor analysis).
These “first generation” measures of coparenting produced an emerging body of knowledge about the salient dimensions of coparenting, the predictors of coparenting relationship quality, and the consequences of difficulties in coparenting for parents and children. Integrating this body of research, Feinberg (2003) developed a 4-domain model of coparenting and a theoretical account of why coparenting plays a significant role for both parents’ own adjustment, their romantic relationship, their parenting, and children’s well-being. Recent research has supported key aspects of that model, as well as extending and sometimes challenging some elements (Brown, Schoppe-Sullivan, Mangelsdorf, & Neff, 2010; Cook, Schoppe-Sullivan, Buckley, & Davis, 2009; Isacco; McConnell, Futris, & Bartholomae, 2009; Schoppe-Sullivan, Brown, Cannon, Mangelsdorf, & Sokolowski, 2008).
The conceptual framework of coparenting developed by Feinberg included 4 overlapping domains (Feinberg, 2003): childrearing agreement, coparental support/undermining, division of labor, and joint management of family dynamics. The domain of childrearing agreement refers not to interparental dynamics, but rather to whether parents’ views of how to rear a child are similar or not. If parents do not agree with each other about how to parent, the opportunity for conflict will be greater and coparenting supportively is likely to require substantial and ongoing negotiation and compromise. Childrearing disagreement has been linked to child behavior problems in the preschool and kindergarten period (Block, Block, & Morrison, 1981; Deal, Halverson, & Wampler, 1989), during adolescence (Feinberg, Kan, & Hetherington, 2007), as well as longitudinally across these periods (Vaughn, Block, & Block, 1988).
Conceptually, coparental support includes affirmation of the other parent’s competency as a parent, acknowledging and respecting the other’s contributions, and upholding the other’s parenting decisions and authority (Belsky, Woodworth, & Crnic, 1996; McHale, 1995; Weissman & Cohen, 1985). The negative counterpart of support is expressed through undermining the other parent with criticism, disparagement, and blame. Some parents adopt a competitive approach in which a gain for one in authority or warmth with the child is a loss for the other (Ihinger-Tallman, Pasley, & Beuhler, 1995). Although considered as part of the same domain, there is good reason to expect these dimensions to be correlated but not identical.
Our intervention work has focused largely on coparental support/undermining, because research indicates that these dynamics are linked to parenting and child outcomes. For example, coparental support and/or undermining are associated with parental self-efficacy, stress, and depression; parenting quality; conscience development; and adjustment and behavior problems from childhood through adolescence (Abidin & Brunner, 1995; Bronte-Tinkew, Horowitz, & Carrano, 2010; Dorsey, Forehand, & Brody, 2007; Feldman, Masalha, & Derdikman-Eiron; Floyd & Zmich, 1991; Jones, Forehand, Brody, & Armistead, 2002). Several studies have found that support/undermining accounts for additional variance in, for example, effortful control, inhibition, and classroom adjustment, after controlling for parenting (Belsky, Putnam, & Crnic, 1996; Karreman, van Tuijl, van Aken, & Dekovic, 2008; Stright & Neitzel, 2003).
The issue of how childrearing labor is divided between men and women has been a focus of considerable research. This literature has been influenced by a broader interest in gendered power relations, their instantiation in daily life, and implications for individual and family well-being (Cowan & Cowan, 1988; Goldberg & Perry-Jenkins, 2004; Milkie, Bianchi, Mattingly, & Robinson, 2002). The division of childrearing labor is woven into the fabric of how coparents view their responsibilities, whether they feel supported in the parental role, and their levels of parental stress and depression. Some research has found that parents’ satisfaction with the way childrearing responsibilities are divided and shared is more important than the actual burden on each (Belsky & Hsieh, 1998; Hackel & Ruble, 1992; Kalmuss, 1992).
The fourth domain of coparenting is parents’ joint management of family relations. As the guiding forces in the family, parents either explicitly or implicitly set standards for how family members treat each other; guide the degree of structure and cohesiveness in family relations; allow or avoid parent-child coalitions; and determine the degree of balance of parents’ interactions (e.g., mother vs. father) with children. An important aspect of this joint responsibility concerns the way that parents expose children to their own conflicts. A great deal of research has linked exposure of children to interparental conflict with negative outcomes in children and parents (Grych & Fincham, 2001; Jouriles, Norwood, McDonald, Vincent, & Mahoney, 1996; O'Leary & Jouriles, 1994). Although couples differ in the degree of negative and hostile conflict, all couples actively or passively decide the extent to which they expose children to their conflicts.
Qualitative interviews with new parents (Feinberg, 2002) drew our attention to an additional, potentially significant aspect of coparenting: sharing the joys of parenthood. This dimension of coparenting, which we termed “parenting-based closeness”, is conceptually related but distinct from coparental support. Whereas coparental support relates to respecting and upholding the other parent’s decisions, parenting-based closeness relates to the shared celebration of the child’s attainment of developmental milestones, the experience of working together as a team, and witnessing one’s partner develop as a parent. Thus, we have included measurement of this construct along with the 4 domains of coparenting described above.
Feinberg (2003) also presented a conceptual model outlining how coparenting may play a central role in family life, providing essential security and support for parents and children. The coparenting relationship was viewed as influenced by individual parent characteristics, the quality of the existing couple or marital relationship, and extra-familial stressors and supports. Coparenting relations were seen to be important influences on the adjustment of parents (e.g., parental stress, efficacy, depression), parenting quality, and ultimately child outcomes. Reciprocal relations were hypothesized throughout, such that parent and child adjustment would in turn influence coparenting quality, and coparenting quality would influence marital relations.
As noted above, evidence has continued to accumulate supporting, extending, and sometimes challenging aspects of this model. For example, research has continued to document the existence of a link between coparenting relations and parenting quality: Morrill and colleagues found evidence of such a link for both mothers and fathers of children under 18 (average child age of 10 years; Morrill, Hines, Mahmood, & Cardova, 2010). This study also closely examined alternative statistical models that incorporated marital quality, and found that 2 models fit the data well: In the first statistical model, coparenting mediated the influence of martial quality on parenting (supporting the pathways described by Feinberg’s conceptual model). In the second statistical model, coparenting simultaneously influenced both marital quality and parenting (which is consistent with an understanding of a bidirectional influence between marital quality and coparenting, including a view of coparenting as a key influence on martial quality). However, the data were not completely promiscuous in the sense that they did not support all models: for example, neither marital quality nor parenting served as well as coparenting as the simultaneous predictor of the other 2 constructs.
Greater understanding of the role of coparenting in diverse family contexts has also continued to emerge. For example, a study of low-income, “fragile families” with 1-year-old children reported significant relations between residential fathers’ stress and aggravation in parenting and supportive coparenting, even controlling for a number of father and mother variables (Bronte-Tinkew, Horowitz, & Carrano, 2010). The evidence from this study confirms a link between coparenting relations and parent adjustment among a low-income population.
Another large study of low-income families focused on Mexican American parents, documenting a link between coparenting conflict and both mother-child interaction and father engagement (Cabrera, Shannon, & La Taillade, 2009). Father acculturation moderated the link between coparenting conflict and father engagement, with conflict linked to greater engagement in childrearing for more acculturated fathers but less engagement among less acculturated fathers. This finding reminds us of the importance of culture in shaping reactions and solutions to problematic situations. Cabrera and colleagues noted, however, that their measure of coparenting was limited to the conflict dimension and did not include other dimensions described previously (Feinberg, 2003).
Although valuable and informative, these studies would have been richer if they had included a multi-dimensional measure of coparenting, allowing scholars to understand more precisely how different dimensions of the coparenting relationship are linked to marital/couple relationships, parent adjustment, parenting, and child outcomes. An understanding of such relations at the level of dimensions of coparenting would carry substantial benefit for the development of supportive, preventive, and treatment approaches for families.
To assess these coparenting domains, we developed a comprehensive self-report measure, the Coparenting Relationships Scale (CRS). The measure yields an overall score for coparenting quality as well as subscale scores representing each domain. In addition, to facilitate research into coparenting when participant burden is an acute issue, we created a brief version of the measure. Measurement development and assessment of validity occurred in the context of a transition to parenthood prevention trial focused on coparenting among couples expecting a first child (Feinberg, Jones, Kan, & Goslin, 2010; Feinberg & Kan, 2008; Feinberg, Kan, & Goslin, 2009). Data presented here come from 3 waves of data collected between 6 months and 3 years after the birth of the child.
We first present data on the psychometric characteristics of the overall score of coparenting yielded by the full scale, the subscale scores, and the abbreviated version of the measure. We examine both internal coherence of the full scale and subscales and stability of coparenting over time. We next examine external validity by assessing whether the measure of coparenting is associated with related constructs in a manner consistent with theory and prior research. For example, as a subset of the couple relationship, coparenting quality is conceptually and empirically related to overall couple relationship quality. We examine cross-sectional relations across the first 3 years of parenthood between coparenting and parents’ reports of positive relationship factors (love, sex/romance, couple efficacy) and negative dimensions (conflict, chronic arguing, possibility of divorce).
Participants were 169 co-resident, heterosexual couples that, at the time of recruitment, were expecting their first child. Eighty-two percent of couples were married and the majority of participants (91% of mothers and 90% of fathers) were European American. Median annual family income was $65,000 (SD = $34,372), with a range of $2,500 to $162,500. Average educational attainment was 15.06 years for mothers (SD = 1.82) and 14.51 years for fathers (SD = 2.19); 14.4% of mothers and 29.3% of fathers did not complete any post-secondary school education. All expectant parents were 18 years of age or over. Mean ages were 28.33 (SD = 4.93) years for mothers and 29.76 (SD = 5.58) years for fathers. The sample is generally representative of the background of families from the rural areas, towns, and small cities in of Pennsylvania where the data were collected.
Couples were primarily (81%) recruited into a transition to parenthood intervention study (Feinberg & Kan, 2008; Feinberg et al., 2009) from childbirth education programs at 2 hospitals as well through healthcare providers (8%), advertisements/flyers (7%), word of mouth (3%), or unknown means (1%). Further details about recruitment can be found in Feinberg and Kan (2008). After prenatal baseline data collection (wave 1), couples were randomly assigned to the Family Foundations (FF) intervention (n = 89) or no-treatment control conditions (n = 80). The FF program involved 8 classes, half before birth and half after birth, and focused on promoting positive coparenting relationships (including communication and problem solving, providing coparental support, aligning expectations for childrearing tasks, and coordinating positive parenting).
The Coparenting Relationship Scale was first administered to parents at wave 2 as a written questionnaire when their first child was an average of 6.5 months old (SD = 1.6), again at wave 3 when the child was an average of 13.7 months old (SD = 1.3), and a third time at wave 4 when the child was 36.8 months old (SD = 1.47). Couples had known each other for an average of 7.4 years when the CRS was first administered and had been living together for an average of 4.4 years. Six couples had a second child at the third wave of data collection, and 63 had a second child at the fourth wave of data collection. Wave 2 data was collected via a mail-in questionnaire, and waves 3 and 4 were collected during home interviews between 2004 and 2008. Mothers and fathers separately completed questionnaires. 90% (n = 152) of mothers and 88% (n = 149) of fathers who participated at baseline (during pregnancy) completed the first administration of the CRS. 92% (n = 156) of mothers and 88% (n = 148) of fathers participated in the second administration of the CRS. During the third CRS administration, 82% (n = 139) of mothers and 76% (n = 129) of fathers participated. To test for differences between parents who did and did not participate in data collection, we used multilevel logistic regression models predicting participation status with baseline depression, antisocial behavior, income, educational attainment, social desirability, couple love, and couple conflict. All predictors were non-significant at waves 2, 3, and 4—with the exception of education at wave 4 (i.e., 1 of 21 tests was significant, a level equivalent to chance with alpha = .05). At wave 4, education significantly predicted greater participation in data collection. Aside from study attrition, non-response to CRS items was minimal, with no items being skipped by more than 4 respondents in a particular wave (less than 1% missing overall).
Our validation of the CRS started with an initial 47-item measure developed for the intervention study of Family Foundations. We chose and adapted items from a number of previous measures of the parenting alliance and coparenting (Abidin & Brunner, 1995; Cordova, 2001; Frank, 1988; Margolin, 1992; McHale, 1997), and we created additional items. Items were selected based on their face validity with reference to the 7 proposed subscales based on theory and prior research (Feinberg, 2003). The 7 subscales were based on the 4 domains of coparenting introduced previously (childrearing agreement, support/undermining, satisfaction with the division of labor, and family management). The domain of support/undermining was assessed by 3 subscales. To these 6 subscales, we added an additional indicator of parent-based closeness. The first author then reduced the pool of items for each subscale by eliminating items that were essentially the same with slight wording differences and items that were somewhat tangential to the subscale construct or seemed to overlap with other domains of coparenting. The first author further made item selection decisions in an effort to capture the core aspects of each construct based on prior research and clinical judgment, thus arriving at a 47-item measure. The original source of each item retained after psychometric analyses in a final 35-item measure (see below) is found in the Appendix.
We represented the domain of coparenting agreement with a subscale of the same name. The coparenting support/undermining domain was represented by 3 subscales – Coparenting Support (4 items), Endorsement of Partner’s Parenting (7 items), and Coparenting Undermining (6 items). Coparenting Support assessed one’s perception of coparental support from a partner, whereas Endorsement of Partner’s Parenting assessed one’s own positive attitude towards a partner’s parenting. Because coparenting support and undermining may exist as distinct (but likely negatively correlated) dimensions, we also included a negative subscale, Coparenting Undermining. The management of family relationships domain was assessed with a 7-item subscale focused on whether the parents exposed the child to their conflicts: Exposure to Conflict. The division of childrearing work domain was captured by a subscale called Division of Labor. The 2-item subscale measured satisfaction with partner’s contributions and involvement in childrearing. Finally, we created a 5-item subscale measuring the degree to which coparenting enhanced intimacy and strengthened the couple’s relationship: Coparenting Closeness.
We measured 5 constructs related to the CRS, which enable criterion validity analyses. Couple relationship quality was measured using the Relationships Questionnaire, which consists of 14 items comprising 2 subscales, Couple Love and Couple Conflict (Braiker & Kelley, 1979). The questionnaire used a 9-point scale. The love subscale consisted of 9 items that inquire about the extent to which respondents have a loving, giving, committed, intimate, and cohesive relationship with their partner. Across the 3 time points, alpha ranged from .89 to .94 for mothers and .84 to .93 for fathers. The conflict subscale consists of 5 items which ask respondents to report on the degree to which they experience different aspects of conflict in their relationship. Alpha ranged from .76 to .81 for mothers and .68 to .73 for fathers.
Chronic, ineffective arguing was captured by the Ineffective Arguing Inventory (Kurdek, 1994), which asks respondents to describe their ability to handle disagreements and solve problems in their relationship with their partner. The measure consists of 8 items with responses on a 5-point scale ranging from strongly disagree to strongly agree. We administered the scale only at wave 3, with alphas = .89 for mothers and .90 for fathers. Whereas the Conflict measure captures the amount of conflict couples experience, the Ineffective Arguing measure focuses on couple inability to solve interpersonal problems.
The Divorce Proneness measure was developed for the Marital Instability over the Life Course Study (Booth, Johnson, & Edwards, 1983; Edwards, Johnson, & Booth, 1987). This measure has been used widely as a predictor of relationship dissolution. We adapted the short version of the scale (5 items) by adding 3 items and by making the wording appropriate for married and cohabitating couples (alphas = .94 for mothers and .81 for fathers at wave 3; alphas = .95 for mothers and .91 for fathers at wave 4).
The Couple Efficacy measure is a 7-item scale that asks the respondent to rate the degree of perceived control they feel they have in terms of managing conflict in their relationship with their partner (Fincham & Bradbury, 1987). There is a 7-point Likert response scale ranging from Strongly Disagree to Strongly Agree. Higher scores indicate greater efficacy. Data on this measure were collected only at wave 3, with alphas = .92 for mothers and .91 for fathers.
A measure of sex and romance satisfaction was developed for this study and administered at waves 3 and 4. The measure consists of 4 items answered on a 9-point Likert scale. Domains of satisfaction with sex and romance covered include: (1) Sex, (2) Cuddling and touching, (3) Romance, and (4) Passion and excitement. Alphas at wave 3 were .78 for mothers and .87 for fathers. At wave 4, alphas = .85 for mothers and .90 for fathers.
We used the 33-item Social Desirability measure to examine the extent to which CRS subscale scores relate to individual tendency to respond to items in a favorable manner. We administered the social desirability scale at wave 1 only, which included such statements as “I am always courteous, even to people who are disagreeable” (Crowne & Marlow, 1964).
First, exploratory and confirmatory factor analyses were used to examine psychometrics of the CRS subscales based on the a priori assignment of items. These analyses identified items that were redundant or did not cohere with other items, reducing the number of items to 35. All items have a 7-point response scale from not true of us (0) to very true of us (6), except for the Exposure to Conflict subscale, where response categories range from never (0) to very often (6). We calculated all subscale scores by taking the mean of the items comprising each subscale. We created a total score by averaging all items (negative items reverse scored, see Appendix for details).
We also developed a brief 14-item measure of coparenting using 2 items from each of the 7 subscales. We selected items for use in the brief coparenting measure (see Appendix for details) that (1) demonstrated strong correlations with their respective subscale score and (2) conceptually captured the core meaning of each subscale. Higher values of the total score, brief measure score, and separate subscale scores—with the exception of Exposure to Conflict and Coparenting Undermining—indicate more positive coparenting. Table 1 reports the means and standard deviations of the total score, brief score, and all subscales, separately by gender and data collection time point. Table 2 reports the number of items in each subscale, along with data on internal reliability for the total score, brief score, and all subscales.
We calculated the reliability of the overall Coparenting Relationship Scale and its separate subscales by wave and gender using Cronbach’s alpha. To further analyze the factor structure and stability of the CRS we conducted a longitudinal confirmatory factor analysis using Mplus version 5.1 (Muthen & Muthen, 2007). Models were estimated using full information maximum likelihood, which provides optimal estimation of missing data (Wothke, 2000). To account for the nesting of parents within couples, we used aggregate analysis, which appropriately adjusts parameter standard errors and the goodness of fit statistics in response to the nested data structure (Muthen & Satorra, 1995). The model, which is illustrated in Figure 1, contains a separate latent variable for each of the 3 waves of coparenting data collection. To assess the stability of the CRS, we regressed each subsequent wave of coparenting data on the previous wave. Each latent variable is measured by the 7 theoretically derived CRS subscales, which are computed as the mean of the items in each subscale (Undermining and Exposure to Conflict are reversed).
To assess convergent validity, we examined concurrent correlations between the CRS and related constructs, including the previously described measures of Couple Love, Couple Conflict, Divorce Proneness, Sex & Romance, Ineffective Arguing, and Couple Efficacy. Multilevel regression models were used (parents nested within couple) in follow-up analyses to test for relational consistency across treatment and control groups, examining whether intervention status moderated relations between the CRS and the previously listed constructs.
The overall Coparenting Relationship Scale demonstrated excellent internal consistency, with Cronbach’s alphas ranging from .91 to .94 across gender and data collection time points. The brief Coparenting Relationship Scale also maintained good internal consistency, with alphas ranging from .81 to .89. Alphas for the 7 coparenting subscales were more varied, but generally strong. Coparenting Closeness, Exposure to Conflict, Coparenting Support, and Coparenting Undermining all had good internal consistency across gender and data collection time points, with alphas ranging from .75 to .90. The internal consistency of Endorsement of Partner Parenting was strong for women (alphas ranged from .83 to .88), but weaker for men (alphas ranged from .61 to .71). The internal consistency of the Coparenting Agreement subscale was also weaker but still in the acceptable range and consistent across gender, with alphas ranging from .66 to .74. The Division of Labor subscale contains only 2 items, thus alpha could not be computed. The correlations between the 2 items were moderate, ranging from .33 to .59. Table 2 provides a more detailed description of the internal consistency results, reporting alphas for all subscales separately across gender and wave.
Table 3 presents correlations between the full CRS scale, the brief CRS scale, the 7 CRS subscales, and a measure of social desirability. Results indicate that the brief CRS scale was an excellent approximation of the full CRS scale, with a correlation of .97 for mothers and .94 for fathers. Furthermore, the separate CRS subscales were strongly related to the total CRS score for both men and women, with correlations ranging in magnitude from .54 to .85. The 1 exception is the Division of Labor subscale for men, which maintained a more moderate correlation of .37.
Figure 1 presents the longitudinal confirmatory factor analysis structural equation model. Model fit was acceptable but not outstanding with RMSEA = .06 and CFI = .93. The chi square model fit index was significant, indicating that the model does not precisely replicate the covariance matrix. Nevertheless, all CRS subscales maintained high loadings on the CRS latent factor for all 3 time points. The lowest factors loadings were for Division of Labor at wave 3 and Exposure to Conflict at wave 4, both of which were still moderately strong at .46. The model also indicates that coparenting scores were stable across time, with regression coefficients of .74 from wave 2 to wave 3 and .71 from wave 3 to wave 4.
Table 4 presents criterion validity data, detailing the concurrent correlations between the CRS scale and several related measures across the 3 data collection time points. The CRS scale maintained a strong positive association with Couple Love (rs ranged from .60 to .76), Couple Efficacy (r ranged from .61 to .65), and Quality of Marriage (rs ranged from .64 to .71). The CRS scale also maintained a moderate to strong negative association with Couple Conflict (rs ranged from −.34 to −.61), Ineffective Arguing (rs ranged from −.53 to −.62), and Divorce Proneness (rs ranged from −.38 to −.72). Correlations were generally stronger for women than for men, especially with regard to Divorce Proneness, and the Sex & Romance scale. For Divorce Proneness at wave 3, r = −.38 for men and r = −.61 for women; at wave 4, r = −.41 for men and r = −.72 for women. A Fisher’s z test for equality of correlations across males and females found the correlation magnitudes to be significantly different at wave 3 (z = 2.71, p < .01) and wave 4 (z = 3.41, p < .001). Gender differences in correlations for Sex & Romance are similar, with r = .43 for women and .20 for men at wave 3. At wave 4, r = .39 for women and .25 for men. The Fisher’s z test for equality of correlations across gender was significant at wave 3 (z = −2.33, p < .05) but not wave 4 (z = −1.27, p = .21), indicating the correlation magnitudes are significantly different at wave 3 but not wave 4. Table 3 presents discriminant validity results, which indicate the CRS and its subscales maintained a relatively small association with social desirability. Correlations ranged in magnitude from .04 to .26 and were consistently larger for men but the differences in correlation magnitude between men and women were not significant.
Follow-up analyses examining the moderating influence of intervention status on the relation between the CRS and the measures listed in Table 4 largely resulted in null findings. However, significant intervention status by CRS interactions indicated concurrent relations between the CRS and (a) Divorce Proneness at waves 3 and 4 and (b) Couple Conflict at wave 4 were stronger in the treatment group as compared to the control group. Differences between the intervention and control groups in 2 of the 3 instances are of a small magnitude: A 1 SD unit change in the CRS predicted a −.37 SD unit change in wave 3 Divorce Proneness for the control group and a −.45 SD unit change in the treatment group. Similarly, the standardized relations between the CRS and Divorce Proneness at wave 4 were −.38 for the control group and −.49 for the treatment group. For Wave 4 Couple Conflict, the magnitude of the difference was larger: the standardized relation is −.32 in the control group and −.67 in the treatment group. (All relations are significant for both groups.)
This study presents initial psychometric data on a new measure of coparenting, based on prior research and theory in this growing area of investigation. Coparenting has been demonstrated to be an important factor linked to parent adjustment, parenting quality, and child outcomes (Abidin & Brunner, 1995; Belsky, Putnam et al., 1996; Dorsey et al., 2007; Floyd & Zmich, 1991; Jones et al., 2002; Karreman et al., 2008; Stright & Neitzel, 2003). In some studies, these links have been demonstrated (1) in longitudinal designs, (2) controlling for parent characteristics and parenting, and/or (3) as stronger compared to overall couple relationship quality. To facilitate more precise investigations of the predictors and consequences of coparenting, we developed the Coparenting Relationship Scale, comprised of several subscales suggested by prior quantitative and qualitative research and theory.
The current study indicates that an overall multi-dimensional index of coparenting quality can be reliably assessed and is fairly stable during the early stages of family formation in a sample of families who entered family formation as 2-parent families. The dimensions of coparenting suggested by a conceptual framework (Feinberg, 2003) and prior formative research with new parents (Feinberg, 2002) are also supported by 7 internally consistent subscales that are moderately associated with one another and largely strongly correlated with the overall score. These findings allow for more precise measurement of coparenting dimensions and further research efforts to understand whether there are specific linkages between these dimensions and parent, child, and family functioning.
The brief 14-item version of this measure --retaining 2 items from each of the conceptually and empirically distinct constituent subscales-- is an excellent representation of the overall subscale scores. This suggests that coparenting quality can be adequately assessed in clinical or other settings where time is limited. Given the strong associations between coparenting and other dimensions of family life and child outcomes described above, it is noteworthy that the brief measure can be incorporated into future studies of child and family issues without sacrificing other measures or data quality.
We found moderate agreement between mothers and fathers on overall coparenting quality, measured by the total score or brief scale. In fact, this level of agreement is towards the higher end of what is typically demonstrated for intra-family agreement on salient features of family relationships (De Los Reyes & Kazdin, 2005; Feinberg, Howe, Reiss, & Hetherington, 2000; Margolin, Hattem, John, & Yost, 1985; O'Leary, 2006). Nonetheless, the correlations between parents on the subscale scores were smaller, suggesting that mothers and fathers may perceive their coparenting relationship somewhat differently. It will be important for future research to investigate discrepancies between mothers’ and fathers’ assessments of their coparenting relationship. It is possible that such discrepancies are associated with aspects of prenatal expectations and/or postnatal coparenting quality, relationship closeness, or couple communication.
In addition, we note that the subscale assessing endorsement of partner’s parenting is less internally consistent for fathers than for mothers, and the division of labor score is less strongly linked to the overall score for fathers. It is possible that these dimensions are less cohesive or salient in the minds of fathers, and may warrant additional development work. Furthermore, the division of labor subscale is only comprised of 2 items; we recommend that those desiring a greater focus on this dimension explore measures specific to this construct (Atkinson & Huston, 1984; Barnett & Baruch, 1987). We note, too, that parents’ scores correlate on this measure at about the same level as most other scales, which one would not expect if the reliability of the subscale were problematic. Moreover, correlations between mothers’ reports on the Division of Labor subscale and other subscales are generally stronger that those for fathers. This pattern is consistent with research indicating that mother’s experience of dissatisfaction with the division of labor is particularly troublesome.
Finally, the overall coparenting score is moderately to strongly associated with a range of dyadic relationship variables in expected directions, ranging from love and sex/romance to conflict and ineffective arguing. These findings contribute to the criterion validity of the measure and accord with prior research linking coparenting to the general quality of the couple relationship (McHale, 1995; Schoppe-Sullivan, Mangelsdorf, Frosch, & McHale, 2004). That the associations with divorce proneness and sex/romance appear stronger for mothers than for fathers stands in contrast to research on links between parenting and couple relationships, which have been shown to be stronger for fathers (Belsky, Youngblade, Rovine, & Volling, 1991). It is possible that fathers view the coparenting relationship as a less salient element contributing to the likelihood of relationship dissolution, whereas mothers view coparenting as more central to relationship stability. Fathers may also be able to compartmentalize the issue of sex and romance from coordination of parenting more than mothers can, especially early in the child’s life when the burden of childcare responsibilities may limit—especially for mothers, who may be staying home and/or breastfeeding—interest in and opportunities for sex and romance. Such differences between parents require further investigation and demonstrate the importance of collecting information about coparenting from both parents to capture a complete picture of family functioning.
In most cases, the relations of CRS scores and other variables were not significantly different across intervention and control groups. Significant differences between groups were found in 3 cases, with the relation being stronger for the intervention group for all 3—however, two of these instances involved only small differences between groups. One possibility for these results is that the focus on aspects of coparenting in the program enhances the salience of coparenting among parents, and thus improves the accuracy of their self-reports regarding coparenting relations.
We acknowledge that our self-report measure of coparenting is inherently subjective and may be limited in capturing some dimensions of coparenting dynamics. For instance, observational coding schemes have been developed to assess the ways in which parents balance interactions when they are together with their child (McHale et al., 2001). However, it would be difficult to observe parents’ satisfaction with the division of childcare responsibilities or their perceptions that their partner supports their parenting. It seems that self-report and observational measures can complement one another in assessing coparenting and should be used in concert when possible. Further understanding of the relations and meaning of observational and self-report measures is a major task for the field of family studies and development in general.
This study was limited by its eligibility criteria for participation (i.e., cohabiting heterosexual couples) and by its use of a largely European American and married sample (although there was a wide range of income, and the sample was generally representative of the semi-rural region). Furthermore, although study attrition was modest and could not be predicted at waves 2 and 3, it was predicted by educational attainment at wave 4.
We also focused on coparenting when it first emerged at the transition to parenthood, but the salience of coparenting dimensions may change as children grow older and additional children are added to the family. For instance, triangulation of children in coparenting conflict may become more likely as children grow older, whereas balancing interactions during triadic play may become less salient. Moreover, the level of parent agreement about childrearing issues may change during development as different issues become salient and parents make new types of decisions and rules for their children.
One may wonder what the added value is of one more couple-relationship construct, especially given that coparenting is relatively strongly linked to other existing measures of the dyadic couple relationship. The strongest association, between love and coparenting, ranges from correlations of .60 to .76. Although strong by the standards of most social science research, these associations account for between about a third and half of the variance in either measure—leaving considerable distinct variance. Moreover, as described above, prior research has shown coparenting quality to be linked to parenting and child adjustment over and above such dyadic couple factors. Assessing coparenting as distinct from the couple relationship may also yield insights into how and when couples are able to coparent effectively even when the couple relationship is troubled. Equally important, coparenting is amenable to measurement in a wide array of family contexts in which assessing dyadic love, for example, is not relevant—including contexts in which mothers and fathers are no longer in romantic relationships, or in which a mother and grandmother or other kin coparent together (Dorsey et al., 2007; Jones, Shaffer, Forehand, Brody, & Armistead, 2003).
The measure described in this paper offers some advantages for researchers and practitioners compared to existing measures of coparenting. First, the measure covers the core dimensions of coparenting identified in the literature as salient for parent, child, and family well-being. Second, the measure includes a novel dimension, the extent to which parenting brings a couple closer together, which is a positive, growth-promoting, and even celebratory aspect of the coparenting experience. Third, psychometric analyses yield generally favorable results: the subscales and overall measure are generally reliable; parents demonstrate reasonable levels of agreement on overall coparenting relationship quality; there is an expected degree of stability in the overall score across early childhood; and correlations with other measures indicate a degree of construct validity. Fourth, the measure has an abbreviated form which corresponds to the longer measure well. The abbreviated measure will allow researchers to briefly assess coparenting, while retaining the ability to relate scores obtained with the brief version to means and variances for samples receiving the full version. The advantages of this measure will also facilitate the assessment of the domains of coparenting in clinical practice, allowing intervention to capitalize on areas of strength and focus on improving areas of difficulty.
We also note that the wording of this version could be modified to refer to the coparenting of multiple children. However, we believe that coparenting relations may differ from one child to the next, and thus recommend that some subscales (e.g., exposure to conflict) be administered regarding each child separately. Moreover, we have created a version of this measure that assesses coparenting between a single parent and her own mother (i.e., the child’s grandparent). For more information, please contact the first author.
We hope that this measure, or future refinements of it, will be useful for examining family relationships across various contexts. We look forward to further inquiries examining the reliability and validity of this measure in an array of families, with diverse sociodemographic backgrounds, levels of risk, and stages of family development.
We are grateful to the families who participated in this study. We thank Jesse Boring, Megan Goslin, Carmen Hamilton, Richard Puddy, Carolyn Ransford, and Samuel Sturgeon for their assistance. This study was funded by grants from the National Institute of Child Health and Development (1 K23 HD042575) and the National Institute of Mental Health (R21 MH064125-01), Mark E. Feinberg, principal investigator.
For each item, select the response that best describes the way you and your partner work together as parents:
|Not true of us||A little bit true of us||Somewhat true of us||Very true of us|
These questions ask you to describe things you do when both you and your partner are physically present together with your child (i.e. in the same room, in the car, on outings).
Count only timeswhen all three of you are actually within the company of one another (even if this is just a few hours per week).
|Never||Sometimes (once or twice a week)||Often (once a day)||Very Often (several times a day)|
How often in a typical week, when all 3 of you are together, do you:
(R) = Reverse score the item
Brief Measure of Coparenting:
Mark E. Feinberg, Prevention Research Center, Pennsylvania State University, Marion Suite 402, University Park, PA 16801. Email: ude.usp@11fem.
Louis D. Brown, University of Texas School of Public Health.
Marni L. Kan, RTI International.