Marital Problems, Maternal Gatekeeping Attitudes, and Father-Child Relationships in Adolescence

Abstract

We evaluated maternal gatekeeping attitudes as a mediator of the relation between marital problems and father-child relationships in three waves when children were in 7^th through 10^th grade. We assessed each parent’s contribution to the marital problems experienced by the couple. Findings from mediational and cross-lagged structural equation models revealed that increased marital problem behaviors on the part of mothers at wave 1 predicted increased maternal gatekeeping attitudes at wave 2 which in turn predicted decreased amounts of father-adolescent interaction at wave 3. Decreased amounts of interaction with either parent were associated within each wave with adolescents’ perceptions that they mattered less to that parent. Amount of interaction with fathers at wave 2 positively predicted changes in boys’ perceptions of how much they mattered to their fathers at wave 3, and amount of interaction with mothers at wave 2 positively predicted changes in girls’ perceptions of how much they mattered to their mothers at wave 3. The findings did not differ for European-American versus Mexican-American families, or for biological fathers versus step-fathers.

Family systems theorists have long recognized that marital relationship quality affects parenting and the parent-child relationship (Cox, Paley & Harter, 2001), but only recently have researchers examined how it affects fathering. One aspect of marital quality is marital conflict, and there is some evidence that conflict influences fathering more than mothering (Cummings, Goeke-Morey, & Raymond, 2004; Cummings, Merrilees, & Ward George, 2010; Krishnakumar & Beuhler, 2000), although other evidence suggests no difference (Erel & Burman, 1995). Marital conflict undermines the quality of fathering as evidenced by increased negative father-child interactions (Jouriles & Farris, 1992; Lindsey, Caldera & Tankersley, 2009), increased paternal control and insensitivity (Davies, Sturge-Apple, Woitach, & Cummings, 2009), decreased father-child attachment security (Frosh, Mangelsdorf & McHale, 2000), and decreased paternal warmth and monitoring (Schofield, Conger, Martin, Stockdale, Conger, & Widaman, 2009). Marital conflict also erodes the quantity of father-child interactions. Fathers are more likely than mothers to withdraw and disengage from their children in response to marital conflict (Christensen & Heavey, 1990). However, there is limited understanding of the underlying processes that account for the vulnerability of fathering to marital disharmony.

The standard explanation for how marital conflict leads to increased negativity, conflict and emotional unavailability in the parent-child subsystem is the spillover hypothesis, which holds that parents transmit negative emotions from their relationship with each other to their relationships with the child and become emotionally drained, less sensitive, and less attentive in their parenting (Erel & Burman, 1995; Krishnakumar & Buehler, 2000). The fathering vulnerability hypothesis (Cummings, Goeke-Morey, & Raymond, 2004) holds that spillover impacts fathering more than mothering because of the greater salience of the parental role for mothers and because mothers poses a greater ability to compartmentalize their roles as spouse and parent. Davies and colleagues (2009) theorized that marital conflict must cause stable internal changes in parents in order to mediate long term affective spillover effects on parenting. The authors examined parents’ depression and emotional security in the interparental relationship as two potential mediators that would link marital conflict to parenting. Cross-lagged structural equation models suggested that marital conflict decreased the security of the interparental relationship for both mothers and fathers, yet decreased emotional security predicted harsh and controlling parenting only for fathers. Thus fathers’ emotional security in the interparental relationship emerged as an underlying explanatory mechanism for the spillover of marital conflict to harsh and controlling fathering (Davies et al., 2009).

While the spillover and father vulnerability hypotheses help explain the relation between interparental conflict and affective types of parenting problems such as harsh discipline, additional explanatory mechanisms seem needed to link the time fathers spent with their children to marital quality. The purpose of this study was to investigate maternal gatekeeping attitudes as an explanatory mechanism linking marital problems and the amount of fathers’ interaction with their children. Maternal gatekeeping has been conceptualized as maternal attitudes, beliefs and behaviors that restrict, limit, or exclude fathers from involvement with children (Allen & Hawkins, 1999; Fagan & Barnett, 2003; De Luccie, 1995). Davies et al. (2009) also speculated that maternal gatekeeping in response to marital conflict might be another mechanism to explain the relation of marital disharmony to harsh and controlling fathering, by reducing opportunities for fathers to hone parenting skills.

Post-divorce conflict between ex-spouses has been associated with increased maternal gatekeeping (Kelly, 2000), but it is unclear how interparental disharmony impacts maternal gatekeeping among cohabitating parents. Some studies have shown that low marital satisfaction and frequent negative marital interactions are associated with increased maternal gatekeeping (De Luccie, 1995; Cannon, Schoppe-Sullivan, Mangelsdorf, Brown, Szewczyk & Sokolowski, 2008). Consequently we assessed the parents’ perceptions of whether they had problems in their relationship during the past year because either or both of them got angry easily, were critical, moody, domineering, jealous, non-communicative, or had sexual relationships outside the marriage (Johnston, White, Edwards, & Booth, 1986). This yielded a measure of the different types of marital problem behaviors the couple experienced rather than the frequency, intensity, or nature of the interparental conflict. We chose marital problems in particular because Booth & Amato (1994) found that marital problem behaviors were inversely related to positive parent-child relationships.

There is more evidence to support the association between maternal gatekeeping and decreased levels of father involvement. Maternal gatekeeping behaviors have been associated with decreased father involvement with children (Gaunt, 2008; Meteyer & Perry-Jenkins, 2010; Fagan & Barnett, 2003). Maternal gatekeeping attitudes also have been associated with decreased father involvement with children ranging in age from 2 to 19 (De Luccie, 1995; McBride, Brown, Bost, Shin, Vaughn, & Korth, 2005). However, many of the findings were limited by study designs that utilized a single wave of data, and thus were unable to establish temporal precidence. In the current study we utilized a three-wave longitudinal design in order to investigate directionality of effects between marital problems, maternal gatekeeping, and parent-child relationships.

There is a need for a better understanding of the role of parent gender in processes of marital conflict and disharmony (Cummings et al., 2010). Marital disharmony is usually treated as a dyadic construct and the roles and contributions of the two spouses are seldom distinguished conceptually and behaviorally. In this study, we distinguished the types of marital problem behaviors exhibited by each spouse to gauge the unique impact of each parent’s contribution to marital disharmony on maternal gatekeeping. If more types of marital problem behaviors on the part of fathers are associated with increased maternal gatekeeping, it might indicate that mothers increase gatekeeping to protect children from hostile, angry or otherwise emotionally unavailable fathers. Fagan & Barnett (2003) found that mothers increased their gatekeeping when they perceived that fathers had less parenting competence, which raises the possibility that mothers might also instigate gatekeeping in response to marital problem behaviors on the part of fathers. Alternatively, if more types of marital problem behaviors on the part of mothers are associated with increased maternal gatekeeping it would suggest that gatekeeping is a spillover of mothers’ negative behaviors toward the father from the interparental relationship system to the father-child relationship system.

Theorists (Cummings et al., 2010) have noted the need to include diverse family contexts in order to better understand the causes and effects of conflict-related processes. We assessed family ethnicity (Mexican and European American), family status (intact and step-father), and child sex as moderators of the link between marital problems and maternal gatekeeping. Evidence suggests that maternal gatekeeping is related to stronger maternal identity roles (Gaunt, 2008). Combined with evidence of more traditional marital roles and division of labor among Mexican-American than European-American families (McLoyd, Cauce, Takeuchi, & Wilson, 2000), it is possible that maternal gatekeeping in response to marital problems is more likely among Mexican-American families. Child gender was investigated as a moderator because research on marital conflict and child gender is not consistent: there is some evidence to suggest that father-daughter relationships may be more impacted by marital conflict (Davies & Lindsay, 2001; Krisnakumar & Beuhler, 2000), however others have found no difference (Davies et al. 2009).

Regarding the consequences of maternal gatekeeping for fathering, we assessed the recent frequency of father-child interaction in leisure and recreational activities. Maternal gatekeeping might be more likely to impact discretionary activities than routine activities (e.g., having dinner together, taking the child to school). Discretionary activities might hold meaning for the child regarding the father’s desire to be with the child and, by extension, the importance of the child to the father. Theorists (Sameroff, 2010) have recognized the need to focus on the meanings that children assign to parental behaviors to make progress in understanding how different parental behaviors affect the parent-child relationship. Attention to meaning is additionally valuable when we study parenting in diverse cultural and family structure contexts, because the same behaviors could carry different meanings across multiple family contexts and as a result have context-specific effects.

The meaning we assessed was how much adolescents felt they mattered to their mothers and fathers (Marshall, 2001; Rosenberg & McCullough, 1981; Schenck, Braver, Wolchik, Saenz, Cookston, & Fabricius, 2009). Mattering is conceived as a broader construct than parental acceptance, rejection, consistent discipline, or monitoring. Children could derive a similar message that they matter less to their parents from harsh disciplinary strategies, lack of parental monitoring, or relationships with parents that are not particularly warm or close (Marshall, 2001; Rosenberg & McCullough, 1981). Schenck et al. (2009) found that among 7^th graders in stepfather families, perceived mattering to the resident stepfather and to the non-resident biological father were both associated with internalizing and externalizing symptoms. In the present study we hypothesized that because leisure and recreational activities are discretionary, adolescents might perceive that a reduction over time in the amount of father-child interaction in those activities means that they do not matter as much to their fathers as they once did.

We used three waves of longitudinal data to establish the temporal precedence necessary for mediation and to assess the directionality of effects. In wave 1 the children were in 7^th grade, in wave 2 half were in 8^th and half were in 9^th grade, and in wave 3 all were in 10^th grade. In an initial mediation analysis we included marital problem behaviors on the part of both mothers and father at wave 1, maternal gatekeeping attitudes at wave 2, and father-child interaction, mother-child interaction, and mattering to each parent at wave 3. Mother-child interaction and mattering to mothers served as controls to isolate the impact of maternal gatekeeping attitudes on father-child interaction and mattering to fathers. The three prospective waves provided a rigorous test of mediation by temporally separating the risk factor (marital problems), the mediator (gatekeeping), and the outcomes (father-child interaction and mattering to father) (Cole & Maxwell, 2003). Based on the findings of the initial mediation analysis which indicated that only mothers’ marital problem behaviors predicted gatekeeping, we then utilized a cross-lagged panel design to test the direction of effects between mothers’ marital problem behaviors, gatekeeping and father-child interaction (Gollob & Reichardt, 1987). Finally we tested whether the mediational paths between marital problems, maternal gatekeeping, and fathering were moderated by each of three potential factors: child gender, family ethnicity (Mexican-American vs. European-American) and family type (intact vs. families with a step-father residing with the mother).

Methods

Participants

Participants came from a longitudinal investigation of the role of fathers in adolescent and emerging adult development, collected in Phoenix, AZ and Riverside, CA. Data for the present study included the first three waves, which encompassed grades 7 (Wave 1), 8 or 9 (Wave 2; 2 cohorts) and 10 (Wave 3). Approximately equal numbers of participants came from each location and included 188 (48%) boys and 204 (52%) girls, European American (EA; n=199) and Mexican American (MA; n=193) families. All three family members were the same, self-identified ethnicity. Approximately half (n = 217) were intact families, the rest (n = 175) were step-father families. Step-father families were defined as families in which the target child’s biological mother had been living for at least the past year with a man who was not the child’s biological father, and in which the target child lived with the mother more than half time. All participants were interviewed in their language of preference (Spanish or English).

Recruitment strategies varied between sites due to differing laws and school district policies. In Arizona, adolescents were recruited from eight ethnically diverse schools in the Phoenix metropolitan area. Teachers administered a short survey to all 7th graders asking about the students’ ethnic background and family composition, in return for a small donation of equipment (scanner, fax machines) to the school. A total of 2,459 families appeared eligible. Families were telephoned according to a random selection scheme to ascertain eligibility, explain the project, and ask for consent to have research staff call the family. A total of 640 families were contacted. Research staff then called families to explain the details of the project, offer a monetary reward for participation, and obtain consent as per university Institutional Review Board procedures. In Arizona, 204 (32%) families were both eligible and initially agreed to participate. In California, families were recruited from two school districts. School staff used emergency contact cards and enrollment data to determine families that appeared eligible. They then contacted these families to explain the project and screen for eligibility. If the families agreed to participate and met eligibility requirements, research staff called families to explain the details of the project, offer a monetary reward for participation, and obtain consent as per university Institutional Review Board procedures. In California, a total of 540 families were contacted and 192 (36%) were both eligible and initially agreed to participate.

With respect to attrition, Wave 1 had 392 families that participated. 312 (80%) families entered the study married and 80 (20%) were cohabitating. Interviews were obtained from at least one family member for 365 families at Wave 2 and 325 at Wave 3, resulting in 17% attrition from Wave 1 to Wave 3. Families who dropped had lower adjusted incomes (M=54,888 vs. M=70,138) than families who were retained but did not differ on any study variables.

Procedures

In Arizona, all three family members were interviewed in different rooms (to ensure privacy), in California, the family traveled to the research site and all members were similarly interviewed in separate rooms. At Wave 2 all families were interviewed via phone rather than in person. Wave 3 procedures for interviews mirrored Wave 1. As part of the larger battery, interviewers presented all the measures analyzed here verbally to family members.

Measures

Marital Problem Behaviors

We selected eight of the original thirteen items of the Marital Problems scale from Johnston et al. (1986) which captured behaviors likely to cause interparental conflict. The items asked whether the couple had marital problems in the past year “because one or both of you” got angry easily, was moody, had feelings that were easily hurt, was domineering, would not talk to the other, was jealous, critical, or had a sexual relationship with someone else. A “yes” response to each item was followed by asking whether “you, your spouse, or both” exhibited the behavior. If the respondent selected “you” or “both” we scored this a 1 for marital problem behaviors exhibited by that individual for that item. If the respondent selected “your spouse” or “both”, we scored this a 1 for marital problem behaviors exhibited by the other individual each item. Thus selections of “both” loaded equally on both partners, but when a respondent nominated one person that person’s score alone would be elevated. Items were summed to form a total marital problem behaviors score (range 0–8) with respect to each reporter and individual (i.e. fathers reported on their own marital problem behaviors and on mother’s marital problem behaviors; mothers reported both as well). Higher scores indicated more distinct types of marital problem behaviors exhibited in the past year. Alphas for mother and father scales ranged from .64 to .77 across all waves. Both parents reported more types of marital problem behaviors exhibited by mothers in each wave (means ranged from 1.63 to 2.03) than by fathers (1.49 to 1.85). There was moderate agreement between mothers and fathers when reporting on fathers’ marital problem behaviors (W1 r=.46, W2 r= .38, W3 r= .38) as well as when reporting on mothers’ marital problems behaviors (W1 r= .38, W2 r= .37; W3 r= .37; all p values were below .001). The 4 total marital problem behavior scores were standardized and we calculated the average Z-score across reporter for mothers’ marital problem behaviors and for fathers’ marital problem behaviors at each Wave. This measure has not been validated with non-white populations. In this sample there were no mean ethnicity differences except for fathers marital problems W2 (t(340.70) = −2.72 p<.01) and W3 (t(273.39) = −2.81, p <.01), with MA families reporting higher fathers’ marital problem behaviors at W2 and W3 than EA families.

Maternal Gatekeeping Attitudes

We targeted maternal gatekeeping attitudes rather than behaviors because mothers might be less aware of their more subtle gatekeeping behaviors or less willing to report on gatekeeping behaviors when self-reporting. We specifically targeted mothers’ attitudes about standards for household responsibility and child care because this construct has been most strongly associated component of maternal gatekeeping to father involvement (Gaunt, 2008). We modified Allen & Hawkins (1999) Standards and Responsibility subscale. First we dropped the two items that referred to mothers’ perceptions of family members’ and husbands’ household skill levels in order to maintain a more explicit focus on mothers setting standards and assuming responsibilities. Second we added an item (#3 below). Third we changed the wording from “domestic responsibilities” to “household tasks or caring for the children” in item #4. Finally we asked about the past 3 months as a timeframe. Final items were: 1. In the past three months, you frequently re-did some household tasks that your husband/partner had not done well. 2. You had higher standards than your husband/partner did for how well cared-for the house should be. 3. You had higher standards than your husband/partner did for how well cared-for the children should be. 4. You liked being in charge when it came to household tasks or caring for the children. Response choices ranged from 1 (very false) to 4 (very true), with higher scores indicating more gatekeeping. Reliability was good (W1 α=.69; W2 α=.70, W3 α=.74), and the average of the four items was used for the maternal gatekeeping attitudes score.

Father-Child and Mother-Child Interaction

Adolescents answered 5 items at each wave; taken from a previous study by Coltrane, Parke, & Adams (2004) that asked how often parents and children did activities together over the past 3 months. Items included: 1.In the past three months, how often did you play a videogame, board game, or any other indoor game with your (mother or dad/stepdad) at home? 2. How often did you go shopping together? 3. How often did you play a sport or participate in an outdoor activity together? 4. How often did you bake or cook a meal together? 5. How often did you go to entertainment, movies, or sporting events together? Response choices ranged from 1 (never) to 5 (very often). Items were summed to form one composite of child reported father-child interaction (W1 α=.68, W2 α=.71, W3 α=.70) and one composite of mother-child interaction (W1 α=.68, W2 α=.73; W3 α=.75).

Mattering

Adolescents completed a 7-item scale previously used with this sample (Schenck, et al., 2009), adapted from Rosenberg & McCullough’s (1981) review of correlates of mattering to parents. Items were rated on a five point scale, “1”= “strongly agree,” “5” = “strongly disagree.” Some items were reverse scored so that higher scores reflect higher perceived levels of mattering on each items, and then summed to create an overall mattering score. Items included: 1. My (Step/dad/mother) really cares about me. 2. I believe I really matter to my (Step/dad/mother). 3. I think my (Step/dad/mother) cares about other people more than me. 4. I’m not that important to my (Step/dad/mother). 5. There are a lot of things in my (Step/dad/mother)’s life that matter more to him/her than I do. 6. I know my (Step/dad/mother) loves me. 7. I am one of the most important things in the world to my (Step/dad/mother). For mattering to father/step-father reliability was acceptable (W1 α = .86, W2 α = .89, W3= .92) and reliability for mothers was also acceptable (W1 α = .77, W2 α = .81 W3 α =.82).

Results

Preliminary analyses

With the exception of adolescent reports of mattering to mothers at all waves, all variables were within acceptable ranges of being normally distributed. Mattering to mothers was slightly above acceptable ranges of skew (W1 = −2.34, W2= −2.40, W3 = −2.45) and kurtosis (W1 = 7.49, W2 = 7.00, W3 = 7.19), as adolescents tended to not report low scores on this measure for mothers. We utilized Maximum Likelihood Robust (MLR) estimation to estimate all models that included mattering to mothers. Correlations, means, and standard deviations among all variables are presented in Table 1. Correlations showed relations between both parents’ marital problem behaviors and maternal gatekeeping attitudes at each wave and across waves. Maternal gatekeeping and father-child interaction were negatively correlated within Wave 2 and Wave 3, as well as from Wave 2 gatekeeping to Wave 3 father-child interaction. Father-child interaction was positively correlated with mattering to father at each wave, and across waves. Mother-child interaction was positively correlated with mattering to mother within each wave, and from Wave 1 to Wave 2, and from Wave 2 to Wave 3. Family income was not correlated with any variables and thus was not included as a control variable.

Table 1.

Means, Standard Deviations, and Intercorrelations for All Variables

	1.	2.	3.	4.	5.	6.	7.	8.	9.	10.	11.	12.	13.	14.	15.	16.	17.	18.	19.	20.	21.
Wave 1
1. F Mar Prb	--
2. M Mar Prb	.61**	--
3. M Gate	.14**	14**	--
4. F-C Int	−.04	−.06	.08	--
5. M-C Int	.03	.00	.01	.56**	--
6. F Matter	−.13*	−.13**	−.01	.31**	.17**	--
7. M Matter	−.05	−.06	−.08	.10	.20**	.45**	--
Wave 2
8. F Mar Prb	.67**	.42**	.03	−.03	.01	−.17**	−.07	--
9. M Mar Prb	.34**	.64**	.12*	−.06	−.06	−.19**	−.10	.57**	--
10. M Gate	.11*	.17**	.56**	−.05	−.04	−.12*	−.08	.16**	.23	--
11. F-C Int	−.02	−.01	.00	.47**	.26**	.17**	−.03	−.09	−.06	−.12*	--
12. M-C Int	.04	−.01	.03	.37**	.50**	.13*	.11*	.02	−.06	−.05	.51**	--
13. F Matter	−.18**	−.10*	−10*	.21**	.19**	.55**	.33**	−.21**	−.18**	−.18**	.33**	.19**	--
14. M Matter	−.05	−.05	−.11*	.10	.17**	.37**	.55**	−.006	−.04	−.05	.05	.22**	.47**	--
Wave 3
15. F Mar Prb	.64**	.42**	.17**	−.01	.05	−.12*	.01	.65**	.41**	.16**	.00	.08	−.15**	−.01	--
16. M Mar Prb	.38**	.60**	.21**	−.04	.01	−.15*	−.02	.40**	.69**	.22**	−.02	.01	−.15**	−.04	.61**	--
17. M Gate	.19**	.26**	.58**	−.05	−.04	−.16**	−.06	.20**	.30**	.67**	−.04	−.04	−.19**	−.09	.24**	.27**	--
18. F-C Int	−.08	−.07	−.03	.45**	.29**	.13*	−.13*	−.12*	−.08	−.15**	.54**	.39**	.20**	−.06	−.16**	−.10	−.19**	--
19. M-C Int	.02	−.07	.03	.29**	.40**	.04	−.02	−.00	−.05	.04	.30**	.58**	.16**	.15**	−.05	−.07	.00	.52**	--
20. F Matter	−.22**	−.13*	−.01	.23**	.15**	.46**	.07	−.23**	−.15**	−.11	.25**	.19**	.63**	.14*	−.23**	−.13*	−.15*	.45**	.19**	--
21. M Matter	−.04	−.07	−.07	.08	.08	.21**	.32**	.02	−.04	.10	.03	.17**	.21**	.44**	−.04	−.02	.00	.10	.33**	.30**	--
M	0.01	0.01	2.53	13.62	15.07	31.40	32.91	0.01	−0.01	2.46	13.05	14.54	31.02	32.98	0.03	.01	2.54	11.60	12.89	30.82	33.18
SD	0.86	0.83	0.75	3.94	4.18	4.57	3.21	0.86	0.83	0.77	3.97	4.08	5.21	3.34	0.86	0.84	0.79	3.76	4.01	6.04	3.11

^*p<.05.

^**p<.01.

F Mar Prb = Father’s marital problem behaviors. M Mar Prb = Mother’s marital problem behaviors. M Gate = Maternal Gatekeeping. F-C Int = Father-Child Interaction. M-C Int = Mother-Child Interaction. F Matter = Mattering to father. M Matter = Mattering to mother.

Analytical Plan

We first tested the hypothesized mediating role of maternal gatekeeping attitudes in the relation between marital problem behaviors, father-child interaction, and perceived mattering to fathers. We included mother-child interaction and mattering to mother as controls to evaluate whether this process was specific to father-child relationships. Second, we utilized more stringent cross-lagged models to establish the directionality of significant longitudinal paths from the initial mediation model. Third, we tested moderation of significant longitudinal cross-lagged paths by child gender, family ethnicity (EA, MA), and family structure (intact, step-father). Due to the multiple cohorts at Wave 2 we also tested longitudinal paths for cohort differences using chi-square difference tests and there were no significant differences across all models.

Statistically, mediation was assessed using path analysis in Mplus 6.1, analogous to multiple regression when all variables are manifest. We tested mediation using product of the coefficients approach with bias-corrected bootstrapping to estimate the confidence intervals (MacKinnon, 2008). We tested moderation of any significant longitudinal paths in the cross-lagged models by gender, ethnicity, and family type using multiple group analysis. Chi-square difference tests were conducted between a model where each individual longitudinal path was constrained to be equal across groups and a model where the path was free to vary across groups. The resulting chi-square difference tests gives a value with 1 degree of freedom, and if significant (values of χ²>3.84 for α=.05), indicates that the path significantly differs across groups. For models that used MLR as an estimator we used Satorra-Bentler adjusted chi-square difference tests or the Strictly Positive Satorra-Bentler chi-square difference test. In the interest of conserving space, we report values only for significant tests of moderation.

Mediation Analysis

Due to lack of data to support a hypothesized relation between individual parent’s marital problem behavior and maternal gatekeeping we initially to determine whether mothers’, fathers’ (or both) marital problem behaviors predicted later maternal gatekeeping. Figure 1 shows the standardized path coefficients obtained for the model specifying that maternal gatekeeping attitudes mediates the relation between marital problem behaviors, father–child interaction and perceived mattering to father, controlling for mother-child interaction and perceived mattering to mother. Model fit was good (χ² (10) = 14.61, p=.15, CFI = .98, RMSEA = .03, SRMR = .03). Wave 2 maternal gatekeeping attitudes significantly mediated the relation between Wave 1 mothers’ marital problem behaviors and Wave 3 father-child interaction (unstandardized ab= −.14, p=.04). The overall mediation from Wave 1 mothers’ marital problem behaviors to Wave 2 maternal gatekeeping to Wave 3 father-child interaction to Wave 3 mattering to father was also significant (sum of unstandardized indirect effects= −.11, p=.05). There was no significant path between fathers’ marital problem behaviors and maternal gatekeeping. Finally, fathers Wave 3 marital problem behaviors were uniquely negatively related to adolescent Wave 3 reports of mattering to fathers (β= −.23, p<.001).

Figure 1.

First mediation model using maternal report of gatekeeping. χ² (10, N=392) = 14.61, p=.15, CFI = .98, RMSEA = .03, SRMR = .03. * = p<.05. ** = p<.01 ***=p<.001. M MP=Mothers’ marital problem behaviors. F MP=Fathers’ marital problem behaviors. M Gate=Maternal gatekeeping. M Int=Mother-adolescent interaction. F Int=Father-adolescent interaction. F Mat=mattering to father M Mat=mattering to mother. W1=Wave 1. W2=Wave 2. W3=Wave 3. All paths are reported in standardized path coefficients. Non-significant paths are omitted for ease of interpretation, however direct paths were included in the models for predictors to outcomes (e.g. C′ path for mediation).

Directionality of Mediation Effects and Tests of Moderation

Next, we utilized cross-lagged models to test the direction of significant mediational paths when controlling for previous time points with autoregressive paths. Figure 2 shows the cross-lagged model from Wave 1 mothers’ marital problem behaviors to Wave 2 maternal gatekeeping attitudes, and from Wave 2 maternal gatekeeping to Wave 3 father-child interaction. Although fathers’ marital problem behaviors were correlated with maternal gatekeeping, we did not include fathers’ marital problem behaviors or mother-child interaction because neither was related to the proposed mediator (gatekeeping attitudes) in the first SEM mediation model. Model fit was good (χ² (6, N=392) = 8.41, p=.0.21, CFI=.998, RMSEA=.03, SRMR=.01). Results replicated the findings in the first mediation model with the respective cross-lagged paths and enabled us to specify directionality of effects. Mediation of the relation between mothers’ Wave 1 marital problem behaviors and Wave 3 father-child interaction remained significant for maternal gatekeeping attitudes (unstandardized ab= −.044, 95% C.I. [−.127, −.003]). Mothers’ marital problem behaviors at Wave 1 predicted changes in gatekeeping from Wave 1 to Wave 2 and mothers marital problem behaviors at Wave 2 predicted changes in gatekeeping from Wave 2 to Wave 3. Gatekeeping at Wave 2 predicted changes in father-child interaction from Wave 2 to Wave 3. Earlier gatekeeping attitudes did not predict later changes in mothers’ marital problem behaviors and earlier father-child interaction did not predict later mothers’ marital problem behaviors or gatekeeping attitudes. Moderation of these cross-lagged paths was not significant for child gender, family structure, or family ethnicity.

Figure 2.

Cross-lagged model for mothers’ marital problem behaviors, maternal gatekeeping and father-child interaction. χ² (6, N=392) = 8.41, p=.0.21, CFI=.998, RMSEA=.03, SRMR=.01. * = p<.05. ** = p<.01 ***=p<.001. M MP=Mothers’ marital problem behaviors. M Gate=Mother’s report of maternal gatekeeping. F Int=Father-Child Interaction. W1=Wave 1. W2=Wave 2. W3=Wave 3. Non-significant paths are omitted for ease of interpretation.

Second we utilized a cross-lagged model to test the direction of effects from parent-child interaction to mattering to the parent across all three waves. Results for fathers are presented in Figure 3. Model fit was good (χ² (2, N=392) = 0.92, p=.0.63, CFI=1.0, RMSEA=.00, SRMR=.01), and a significant cross-lagged path indicated that Wave 2 father-child interaction positively predicted changes from Wave 2 to Wave 3 mattering to father (β= .09, p<.05). Moderation of this path was significant by child gender ( χ²_diff =7.16, df_diff = 1, p=.01). The path was significant for males (β= .19, p<.001) but not for females (β= −.001, p= .99). No path was found to lead from mattering to father to later father-child interaction at any wave.

Figure 3.

Cross-lagged model for father-child interaction and mattering to father. (χ² (2, N=392) = 0.92, p=.0.63, CFI=1.0, RMSEA=.00, SRMR=.01). * = p<.05. ** = p<.01 ***=p<.001. F Int=Father-child interaction. F Mat=Mattering to father. W1=Wave 1. W2=Wave 2. W3=Wave 3. Non-significant paths omitted. Italic=Females. Bold=Males.

Results for mothers are presented in Figure 4. Model fit was good (χ² (3, N=392) = 17.62, p=.001, CFI=.96, RMSEA=.11, SRMR=.03) with the exception of RMSEA, which had values higher than needed for good fit. Given acceptable values for CFI and SRMR, and to compare to fathers, we retained this model. Wave 2 mother-child interaction positively predicted changes from Wave 2 to Wave 3 mattering to mother (β= .11, p<.05), which matched the results for fathers. Also similar to fathers, moderation of this path was significant by child gender (χ²_diff =5.73, df_diff = 1, p=.02). The path was significant for females (β= .21, p< .001) but not for males (β= .02, p= .82). No path was found to lead from mattering to mother to mother-child interaction at any wave.

Figure 4.

Cross-lag model for mother-child interaction and mattering to mother. (χ² (2, N=392) = 4.38, p=.0.11, CFI=.99, RMSEA=.06, SRMR=.02). * = p<.05. ** = p<.01 ***=p<.001. M Int=Mother-child interaction. M Mat=Mattering to mother. W1=Wave 1. W2=Wave 2. W3=Wave 3. Italic=Males. Bold=Females. Non-significant paths omitted.

Finally we utilized a cross-lagged model to test the direction of effects from W1 marital problem behaviors on the part of fathers to W3 mattering to fathers (Figure 5). Maternal gatekeeping did not mediate the relation between W1 father’s marital problem behaviors and W3 mattering to fathers and was thus not included as a mediator here. Model fit was good (χ² (2, N=392) = 1.14, p=0.56, CFI=1.0, RMSEA=.00, SRMR=.01). The path from Wave 1 fathers’ marital problem behaviors to Wave 2 mattering to father was positive and significant (β= −.11, p<.01) and the same held for Wave 2 fathers’ marital problem behaviors to Wave 3 mattering to fathers (β= −.13, p<.001). These results indicate that when fathers exhibited more marital problem behaviors, adolescents perceived they mattered less to their father. There was also a significant and positive path from Wave 1 mattering to father to Wave 2 fathers’ marital problem behaviors (β= −.10, p<.01). Moderation of this path was significant by family type (χ²_diff =4.59 df_diff = 1, p=.03). The path was significant for step-families (β= −.14, p= .02) but not for intact families (β= .05, p=.36).

Figure 5.

Cross-lag model for fathers’ marital problem behaviors and mattering to fathers. (χ² (2, N=392) = 1.14, p=0.56, CFI=1.0, RMSEA=.00, SRMR=.01). * = p<.05. ** = p<.01 ***=p<.001. M Int=Mother-child interaction. M Mat=Mattering to mother. W1=Wave 1. W2=Wave 2. W3=Wave 3. Italics=Intact families. Bold=Step families. Non-significant paths omitted.

Discussion

There has been considerable evidence of spillover from marital conflict onto parenting (Krishnakumar & Buehler, 2000), particularly onto fathering and father-child relationships (Cummings, Goeke-Morey, & Raymond, 2004; Cummings et al., 2010), yet our knowledge of the causal mechanisms underlying long-term spillover process over the course of years remains limited. The only other study that investigated mechanisms underlying long-term spillover process speculated that maternal gatekeeping might be an important mechanism that reduces father involvement following marital conflict (Davies et al., 2009). We found that from 7^th to 10^th grade, marital problems, specifically more marital problem behaviors on the part of mothers, predicted increased maternal gatekeeping attitudes, which subsequently predicted decreased amounts of father-child interaction, which was associated with adolescents perceiving that they mattered less to their fathers. These findings confirm a role for maternal gatekeeping in spillover processes, suggest modifications to the father vulnerability hypothesis, and also suggest that future research might profitably investigate transactional processes involving the child in spillover and father vulnerability effects. We elaborate each of these points below.

Although there is evidence for the relation of maternal gatekeeping to reduced father involvement (De Luccie, 1995; Cannon et al., 2008), the antecedents of gatekeeping remain unclear. We measured separately mothers’ and fathers’ marital problem behaviors to distinguish each partner’s unique contribution to maternal gatekeeping attitudes. If marital problems generally led to gatekeeping, then both mothers’ and fathers’ marital problem behaviors should be related to maternal gatekeeping attitudes. However, we found that only mothers’ marital problem behaviors were associated with later gatekeeping attitudes, and that across all three waves the direction of effects was from increased mothers’ marital problem behaviors to increased maternal gatekeeping attitudes. Additionally, mother’s marital problem behaviors, not mother’s perceptions of father involvement, predicted later maternal gatekeeping.

In general, our findings provide novel evidence for spillover effects specific to mothers and suggest possible modifications to the fathering vulnerability hypothesis (Cummings, Goeke-Morey, & Raymond, 2004). According to the fathering vulnerability hypothesis, mothers are better able than fathers to compartmentalize their affect between family subsystems, and thus prevent marital disharmony from spilling over to mother-child relationships. Consistent with this hypothesis, we found that mothers’ marital problem behaviors did not spill over directly to the mother-child relationship, whereas fathers’ marital problem behaviors did. Yet mothers did not compartmentalize and prevent their marital problem behaviors from spilling over into the co-parenting system, as revealed by the association between mothers’ marital problem behaviors and gatekeeping attitudes. Mothers might play a more active role in father vulnerability processes than previously recognized, and because mothers’ gatekeeping attitudes were not related to fathers’ marital problem behaviors the current data suggest that mothers’ gatekeeping attitudes were not simply reflections of attempts to shield children from hostile, conflictual fathers. Indeed, the current results suggest mothers may intentionally or unintentionally sabotage father-child relationships in response to marital problems.

We found that maternal gatekeeping attitudes had meaningful prospective associations with father-adolescent relationships. First, gatekeeping attitudes when children were in 8^th and 9^th grades mediated the relation between mothers’ marital problem behaviors at 7^th grade and the amount of time fathers spent with children in quotidian activities at 10^th grade, and predicted reductions from 8^th and 9^th to 10^th grade in the amount of time fathers spent with children. Maternal gatekeeping attitudes appear to have implications for adolescent outcomes, particularly since earlier work (Schenck et al., 2009) found that perceived mattering to fathers was linked with internalizing and externalizing symptoms. Finally, the amount of same-gender parent-adolescent interaction when children were in 8^th and 9^th grades predicted changes in adolescents’ perceived mattering to that parent from 8^th and 9^th to 10^th grade. The amount of time mothers spent with their daughters in 8^th and 9^th grade appeared to send a message laden with meaning about how much the daughter mattered to the mother, because it predicted whether the daughter would become more or less secure in her perception over the next year or two, independent of the amount of time they were actually spending together in 10^th grade. The same was true of fathers and sons. The gender specificity in mattering process suggests child attribution processes are at work. For example, by 8^th and 9th grade boys might prefer action movies over the children’s movies they enjoyed earlier. If fathers don’t go to shared-interest movies with them, adolescents could take that to mean that their fathers don’t want to spent time with them, undermining their sense importance to their fathers. However if mothers don’t go to action movies with male adolescents, they could easily attribute that to mothers’ lack of interest in those types of movies. The reverse attributions about mothers’ versus fathers’ discretionary involvement could be made by girls. Other studies have also found differential effects based on same-gender parent-child dyads (Jouriles & Farris, 1992; Butler & Shalit-Naggar, 2008).

Fathers’ marital problem behaviors in 7^th grade were directly associated with adolescents’ perceived mattering to fathers in 10^th grade and the direction of effects across all three waves was generally from fathers’ marital problem behaviors to perceived mattering. This perceived threat process was not mediated by decreased father-child interaction following fathers marital problem behaviors. There was one transactional path involving fathers’ marital problem behaviors and perceived mattering, and this path was moderated by family type. For step-families, when 7^th graders perceived they mattered less to their step-father, step-fathers increased their marital problem behaviors in the next year or two. It is possible that when adolescents feel that they are unimportant to their step-fathers they engage in more acting out or displays of negative affect and conflict with their step-fathers, causing step-fathers to display “bottom up” spillover of conflict from the parent-child relationship to the marital relationship.

The present study focused on the quantity of father-child interaction and implicated maternal gatekeeping attitudes as an explanatory mechanism of spillover effects. Davies et al. (2009) focused on the quality of fathering and implicated fathers’ interparental relationship security as a mechanism of spillover effects. In what follows we discuss how these two sets of findings might be incorporated into a more transactional model of spillover processes and father vulnerability. The indications from both studies are that marital disharmony perpetuates not only decreased inter-parental attachment security for mothers and fathers but also increases maternal gatekeeping attitudes. Maternal gatekeeping has a direct effect on the quantity of father-child interaction and plays an active role in fathering vulnerability. However, effects on the quality of fathers’ parenting (e.g., harsh discipline) are likely to involve transactional processes that include the child as an active participant (Sameroff, 2010). Because reduced father-child interaction as a result of maternal gatekeeping leads children to feel increasingly that they don’t matter to their fathers, that experienced rejection should lead to increased child anger at and acting out with their fathers. Because fathers’ inter-parental attachment security has been eroded, they in turn should be more vulnerable to attribute their children’s acting out as rejection, which should lead to increased anger at or withdrawal from their children, and more harsh and controlling parenting behavior. As fathers engage in more negative parenting children would feel increasingly rejected, and this process would become self-sustaining over time. The child as an active participant has been missing in process models of spillover and has instead been granted the role of passive recipient of spillover effects from parenting behaviors, and in turn the parent-child relationship. Correspondingly, the locus of spillover processes has been sought in stable internal dispositions of parents such as the ability to compartmentalize, and evolutionarily-determined dissociation between attachment and caregiving systems (Davies, et al., 2009). Understanding children’s attributions, reactions, and contributions to family spillover processes will lead us to a more situational and transactional understanding of dynamic family spillover processes (Sameroff, 2010), and that understanding should also yield important implications for how to intervene.

Finally, we note that the time that parents spent with adolescents was related to adolescents’ perceived mattering at each wave for each parent, even after controlling for associations at the prior waves (Figures 3 and 4). The relation was strongest for both parents at 10^th grade. This is inconsistent with popular notions that adolescents are too old to care about spending much time with their parents. On the contrary, time spent with parents appears to become more important as adolescents age and carries indications about how important they are to their parents. It is noteworthy that the relation between time spent together and perceived mattering was stronger at each wave for fathers than for mothers. These findings in combination with earlier evidence that perceived mattering to both resident step-fathers and non-resident biological fathers was associated with internalizing and externalizing symptoms (Schenck et al., 2009), have implications for family interventions, and also for public policy regarding child custody after divorce, where parenting time allotted to fathers has been disproportionately small.

The present study has some limitations. We relied on mothers as reporters of gatekeeping attitudes and combined mothers’ reports with those of fathers for marital problems, so there was some degree of overlapping method variance between marital problem behaviors and maternal gatekeeping attitudes. Although controlling for prior reports of marital problem behaviors and gatekeeping attitudes in a cross-lagged model helps to reduce shared method variance, independent reports of all constructs would help to reduce reporter bias and shared method variance. In addition, adolescents reported on parent-child interaction and mattering, so there was overlapping method variance with those constructs as well. Lastly, emotional availability of the parents was not measured and it is unclear whether adolescents’ perceptions of mattering to parents would be more proximally influenced by time together or parental emotional availability.

Some have highlighted the possibility that fathers can also engage in gatekeeping (Schoppe-Sullivan, Cannon, Brown, Mangelsdorf, & Sokolowski, 2008; Trinder, 2008), and theoretical work suggests that there are likely to be bi-directional influences across spouses with respect to gatekeeping attitudes and behaviors (Adamsons, 2010). We were unable to address paternal gatekeeping hypotheses with the present data. However, this is a crucial future step for empirical examination if in fact both parents extend the marital battlefield into vying for control of children through increased maternal and paternal gatekeeping following marital conflict. In addition, gates can open as well as close, and it is important to recognize that mothers and fathers can also facilitate each other’s levels of domestic involvement (Parke, in press).

Conclusion

The present findings indicate that mothers play an active role in fathering vulnerability to marital problems by revealing maternal gatekeeping attitudes as an explanatory mechanism for effects of increased mothers’ marital problem behaviors on decreased father-child interaction. The direction of effects from 7^th to 10^th grade was from mothers’ marital problem behaviors to mothers’ gatekeeping attitudes to father-child interaction, and these relations were not moderated by child gender, family structure, or family ethnicity; suggesting that maternal gatekeeping has substantial generality in its negative impact on father-child relationships. In contrast to popular notions that adolescents care increasingly less about spending time with their parents in quotidian activities as they age, the present findings also reveal that adolescents imbue the time that they do spend with their parents, especially their same-sex parents, with much meaning about how much they matter to their parents. Finally, there was direct spillover from fathers’ marital problem behaviors to adolescents’ decreased perceptions of how much they mattered to their fathers.