Detouring in the Family System as an Antecedent of Children’s Adjustment Problems

Abstract

Guided by family systems theory conceptualizations of relational boundaries, this study examined unique links between detouring (e.g., alliance between parents against child) and young children’s psychological functioning after accounting for general family negativity and conflict. Participants in this longitudinal (i.e., two annual waves of data collection), multi-method (i.e., observation, survey, semi-structured interview), multi-informant (i.e., parent, teacher, observer) study included 218 young children (M age = 5.76 years) and their families. Findings from cross-lagged panel analysis indicated that detouring uniquely predicted increases in children’s externalizing problems, hostile attribution bias, and peer rejection and victimization over a 1-year period (i.e., the transition from kindergarten to first grade). Conversely, findings failed to support a child effects model in that child adjustment problems did not predict increases in detouring or family negativity and conflict over time.

Young children who display heightened aggression in conjunction with rejection or victimization by their peers are at high risk for persistent psychological, social, and emotional difficulties that carry into adolescence and adulthood (e.g., Bierman, Kalvin, & Heinrichs, 2015; Coie, Terry, Lenox, Lochman, & Hyman, 1995). In recognizing the importance of understanding the determinants of this high-risk pattern of behaviors in childhood, a growing body of research has focused on elucidating family risk factors (e.g., Bolger & Patterson, 2001; Runions & Keating, 2007; Yaros, Lochman, & Wells, 2016). However, in spite of researchers’ repeated calls for understanding how characteristics of the broader family system (e.g., both caregivers and child) impact early child development (e.g., Cox & Paley, 1997; Kerig, 2016), very few studies have moved beyond examining the influence of dyadic relationships (e.g., parent-child, interparental) on early aggression and peer rejection and victimization. Further, studies that have focused on examining triadic family interactions as a predictor of child adjustment have primarily tested broader constructs of negativity or conflict and warmth or support (e.g., Cummings, Koss, & Davies, 2015; Demby, Riggs, & Kaminski, 2017).

Family systems theory (FST; Cox & Paley, 1997; Minuchin, 1974) has been highly valuable in providing a more nuanced approach to characterizing how different patterns of broader family functioning may have unique developmental implications for children. According to FST, the family can be broken down into subsystems consisting of individual members (e.g., child, mother, partner) or dyads (e.g., parent-child, interparental), but the family as a whole is best understood by the functioning within and across each subsystem. Metaphorical boundaries allow for differentiation between these smaller subsystems within the larger family system and serve as implicit rules for defining and understanding family relationships according to both the amount and quality of contact and the transmission of resources across family subsystems. From this perspective, effective family functioning is defined by clear and flexible boundaries that permit each family member to function within subsystems without the interference of other members but access resources from the larger family unit when needed (Cox & Paley, 1997; Minuchin, 1974). Theorists and prior empirical studies have identified other patterns of functioning in which the boundaries are not as clear or flexible. The two most commonly studied boundary disturbances are disengagement (family is emotionally distant) and enmeshment (family is emotionally entangled), and both have been linked with children’s internalizing and externalizing symptomatology (e.g., Jacobvitz, Hazen, Curran, & Hitchens, 2004; Kerig, 1995; Sturge-Apple, Davies, & Cummings, 2010). However, much less is known about how other boundary disturbances in the form of alliances amongst family members are associated with children’s psychological functioning.

Alliance formations are characterized by coalitions among family members where a third member is excluded or attacked in attempts to either bring the allied members closer together or deflect negativity onto the excluded member (Minuchin, 1974). As a primary type of alliance, detouring patterns of family functioning occur when parents form an alliance against the child (Kerig, 2016). Parents may “detour” around their own conflicts by deflecting attention onto the child and his or her needs or difficulties as a way of attempting to bring themselves closer together (Kerig, 1995). Although this behavioral pattern is thought to often be outside of the parents’ awareness, detouring may become an established pattern of family relatedness through unconscious conditioning processes (e.g., parents are unconsciously reinforced by experiencing marital unity that has been difficult to achieve otherwise). This process may create an artificial sense of closeness and reduce strain in the interparental relationship, but the result for the child is often rejection and exclusion from the family system (Fauber, Forehand, Thomas, & Wierson, 1990). In turn, it is theorized that the targeted rejection experienced by children in a detouring family may serve as a robust predictor of adjustment problems (e.g., Kerig, 1995, 2016; Minuchin, 1974). However, very few studies have systematically explored these potential links between detouring and child problems. Therefore, our primary aim was to examine detouring, above and beyond a general construct of family negativity and conflict, as a unique predictor of young children’s psychological and social adjustment problems (i.e., externalizing problems, hostile attribution bias, peer rejection and victimization).

Detouring as a Predictor of Child Adjustment Problems

The targeted rejection and exclusion experienced in a detouring family, as opposed to negativity and conflict that is generalized across the family, may be especially harmful to children. Detouring may specifically give rise to a pattern of cognitions and behaviors consistent with a high-risk profile of aggression, victimization and rejection, and dysregulation. Social information processing (SIP) theory asserts that when children frequently experience social interactions in which they are the targets of negativity, they are more likely to over-attribute malevolent intent to others, which contributes to the development of a hostile attribution bias that stabilizes over time (Choe, Lane, Grabell, Olson, 2013; Dodge, 2006). In a detouring family, the child is frequently the target of parental rejection or scapegoating. These experiences of exclusion or derision may in turn serve as blueprints for how the child interprets future social situations. For example, he or she may be more likely to perceive a peer as being aggressive (e.g., did it on purpose to harm or exclude the child) even when the situation involves an ambiguous intent (e.g., child not being asked to join in playing a game). Although some studies have shown links between disruptions in the parent-child relationship (e.g., physical abuse, authoritative parenting, parental aggression) and children’s heightened hostile attribution bias (Dodge, Pettit, Bates, & Valente, 1995; Nelson & Coyne, 2009; Runions & Keating, 2007; Yaros et al., 2016), no studies have expanded the focus to triadic family functioning to test whether detouring serves as a unique predictor.

Additionally, the child’s anger and frustration of experiencing ostracism and rejection by his or her parents may engender increases in externalizing problems characterized by relational aggression, overt hostility, and conduct problems (Kerig, 1995; Vuchinich, Emery, & Cassidy, 1988). Although very few studies have tested links between detouring and externalizing problems, some preliminary support for this assertion exists. As one of the only published studies to empirically examine these associations, Kerig (1995) found concurrent links between family members’ ratings of detouring (assessed via Family Cohesion Index) and parents’ reports of child externalizing problems in a sample of 6- to 10-year-old children. Similarly, Yahav and Sharlin (2002) found that in a sample of 10- to 17-year-old children, those with elevated levels of externalizing symptoms reported greater perceptions of being blamed by family members for various family conflicts. However, to our knowledge, no studies have tested these links longitudinally or in early childhood.

In addition to its implications for how children process social information and regulate their behavior, detouring may also have important implications for how children fare socially with peers. Script theory asserts that within the context of early family relationships, children develop internal schemas and expectations regarding themselves and others, which are then applied to subsequent social interactions (Johnston, Roseby, & Kuehnle, 2009; Sroufe & Fleeson, 1986). When these early relationships are neglectful, volatile, or insensitive, children may develop negative expectations regarding future relationships with others as well as views of themselves as incompetent or worthy of blame (Cicchetti, Lynch, Shonk, & Manly, 1992; Cicchetti & Toth, 1998). These internalizations may lead them to select and structure later interactions in such a way that familiar relationship patterns reoccur and are validated (Sroufe & Fleeson, 1986). For example, several studies have shown that experiences of maltreatment, parental criticism, and parent-child conflict are associated with children being the victims of bullying in middle childhood and adolescence (e.g., Bolger & Patterson, 2001; Shields & Cicchetti, 2001; Georgiou & Stavrinides, 2013). Because children in detouring families are often rejected, excluded, or ganged up on by their parents, this may translate into them also being bullied or rejected by other children when they transition into school settings. However, no prior studies have explicitly tested whether detouring predicts peer rejection and victimization.

Early Childhood as a Sensitive Period

We specifically examined detouring as a predictor of child problems during the transition into the early school years based on a number of important developmental considerations. First, early childhood is regarded as a period of heightened risk for children experiencing adversity in their rearing environments (Shonkoff & Phillips, 2000). During this time, children are forming expectancies about the availability and consistency of their care, salient figures in their lives, and broader concerns about the stability and health of relationships in the family (Cicchetti, Cummings, Greenberg, & Marvin, 1990). Therefore, disruptions in family relationships (e.g., detouring, negativity and conflict) during this time are thought to be particularly harmful in relation to children’s psychological and social adjustment. Second, family systems theorists emphasize that examining family processes within the context of major transitions provides unique opportunities for understanding how family functioning impacts child development (e.g., Cowan, 1991; Sturge-Apple et al., 2010). Children’s transition into the early school years represents a key stage-salient task that carries with it many novel challenges (e.g., peer interactions, structured routines, greater time apart from family; Ladd & Burgess, 2001). Children may therefore be more likely during this period to draw on prior experiences with family relationships in guiding their behavior and how they process social information (McHale, Kuersten, & Lauretti, 1996). Third, early childhood is a critical period for the development of stable patterns of social information processing (Dodge, 2006). Although attributions and social processing become more complex and differentiated by situation as children get older, theorists posit that development of hostile attribution bias in early childhood provides the foundation for lifelong social information processing (Dodge, 2006). It is therefore critical to examine these processes and its precursors during this important developmental period. Finally, research has consistently shown that externalizing behaviors (e.g., Loeber, Green, Lahey, Christ, & Frick, 1992) and peer rejection and victimization (e.g., Buhs & Ladd, 2001) in early childhood are robust predictors of adjustment problems across the lifespan.

Present Study

To our knowledge, the present study represents the first longitudinal test of detouring as a predictor of early childhood adjustment problems. As a more rigorous test of this novel aim, we specifically examine the unique predictive value of detouring above and beyond a general construct of family negativity and conflict. Innovation in this study is rooted in its goal of testing novel theoretically guided hypotheses regarding links between detouring and behaviors characteristic of a high-risk pattern of aggression, victimization and rejection, and dysregulation. Consistent with FST, we specifically hypothesized that detouring would predict increases in child externalizing problems, hostile attribution bias, and peer rejection and victimization over a 1-year period. The few previous examinations of detouring as a predictor of child behavior problems have been somewhat limited in that they predominantly use cross-sectional data and self-report survey measures of detouring (e.g., Family Cohesion Index; Kerig, 1995). Use of cross-sectional designs do not meet quantitative criteria for testing directionality (e.g., Cole & Maxwell, 2003), and they cannot rule out potential child-driven effects on family functioning. For example, consistent with FST’s proposition of reciprocity between children and the larger family system (Cox & Paley, 1997; Minuchin, 1974), it is possible that child misbehavior may increase parental blame and hostility toward the child. Therefore, concurrent links between detouring and child externalizing problems (e.g., Kerig, 1995; Yahav & Sharlin, 2002) could reflect multiple directions of effect. In the current study, we specified a cross-lagged, autoregressive model to achieve a more definitive understanding of the nature of associations between detouring and child adjustment outcomes (i.e., also explicitly account for child-driven effects on detouring and family negativity and conflict).

Furthermore, a heavy reliance on self-report measures contradicts the theoretical premise that boundary patterns are implicit and, as a result, commonly outside the awareness of family members (Minuchin, 1974). This lack of awareness causes family members to be inaccurate sources, evidenced by low correlations in reports across family informants (e.g., Kerig, 1995). Existing observational assessments of detouring also have weaknesses in that they often conflate detouring with other forms of family alliance (e.g., parent-child coalition; Lindahl & Malik, 2000). Therefore, our study is particularly innovative in that it represents the first rigorous, longitudinal test of detouring as a predictor of child behavior problems and adopts a novel observational assessment of detouring. To reduce the operation of common method and informant variance in the analyses, we utilized a multi-method (i.e., observation, survey, semi-structured interview), multi-informant (i.e., observer, parent, teacher) approach to assessing key constructs. We also included child sex and total household income per capita as covariates.

Methods

Participants

Participants were drawn from a larger longitudinal project designed to examine the impact of interparental conflict and family relationships on young children. Participants in the larger project included 243 families (mother, partner, and preschool child) residing in a moderate-sized metropolitan area in the Northeastern United States. To obtain a demographically diverse sample, participants were recruited through preschools, Head Start agencies, internet advertisements, family-friendly locations (e.g., farmers’ markets, libraries), and public and private daycares. Given the timing of the family interaction task used to assess detouring (described below), the current study only includes data from families at Waves 2 and 3 of the larger project (i.e., when the child was in kindergarten and first grade). Of the 243 families who participated at Wave 1 of the study, 235 returned for Wave 2, but 17 of the 235 mothers no longer had a partner that served as another caregiver for the child. Because the primary predictor (i.e., detouring) in the current study is conceptualized as the interactional pattern between two caregivers and the child, we only include data from the 218 families that had two caregivers. For clarity, throughout the rest of the paper we refer to the kindergarten assessment as Wave 1 and the first grade assessment as Wave 2. Retention across the two waves of data collection was 93%. The average age of child participants was 5.76 years (SD = .47) at Wave 1 and 6.77 years (SD = .48) at Wave 2, and 57% were girls. The sample was racially diverse with 45% of family members identifying as Black or African American, 46% White, 7% multi-racial, and 3% another race. Approximately 15% of family members identified as Latinx. Median household income of families was $36,100 per year (range = $0 to $162,000), with most families (65%) receiving some form of public assistance. Median education for parents consisted of a GED or high school diploma (range = no high school diploma to Master’s or PhD degree). Of adult participants, 99% of mothers and 70% of their partners were the biological parents of the child; 26% of partners were mothers’ romantic partners but not the biological parent, 1% were the child’s grandparent, and 3% were another family member; 90% lived together (M years lived together = 7.03, SD = 4.64); 49% were married.

Procedures and Measures

Parents and children visited our research center laboratory for two annual waves of data collection. Children’s classroom teachers also completed and mailed questionnaires about children’s behavior in kindergarten and first grade. The Research Subjects Review Board at the University of Rochester approved all scientific procedures prior to conducting the study, and parents gave written consent for both themselves and their child to participate. Parents and teachers were compensated monetarily for participation, and children received small toys at each visit.

Detouring.

At Waves 1and 2, families (mother, partner, and child) participated in a family interaction task (FIT) that was adapted from previous studies (e.g., McHale, Kuersten-Hogan, & Lauretti, 2001). Families were instructed that they had 6 minutes to work together to build a tower out of blocks. Experimenters told families that a vertical marker affixed to the table represented the height of the “highest tower a family has ever built,” and shared that they would beat the “record” if they built their tower higher than the marker. If the tower fell before time was up, they were instructed to rebuild it. To maximize the ability to capture families’ natural boundary patterns, additional instructions were limited to the request that the family work together to build the tower. Once an experimenter gave the instructions, the family was left alone to complete the activity. The task was video-recorded for subsequent coding.

Because very few studies have assessed detouring using observational methods, we used prior family interaction coding systems (Jacobvitz et al., 2004; Lindahl & Malik, 2000) as a guide in developing a novel observational assessment of detouring during the FIT. Specifically, the Detouring scale assesses the degree to which the parents form an alliance against the child in a way that functions to create distance between themselves and the child and increase closeness in their own relationship. Using a 9-point scale ranging from 1 (not at all characteristic) to 9 (highly characteristic), a family rated highly on detouring is characterized by parents who engage in active and purposeful behaviors to exclude the child. Examples include purposefully ignoring the child and his or her appropriate bids for attention (e.g., not responding to child’s constructive requests for help with the blocks or appeals to contribute), belittling the child or the child’s contributions to the task (e.g., “Why would you put that there? You’re messing it up!”), making explicit comments to create distance from and exclude the child (e.g., “Just let Daddy and I do it. We don’t think you know how.”), or using the child as a persistent distraction to focus on (e.g., ganging up on the child when he or she makes a mistake; discussing concerns about the child’s performance or behavior as if the child is not present). More mildly, this is reflected in situations where closeness between the parents occurs at the expense of closeness with the child (e.g., the child withdraws from the task or is not engaged with by parents). A key consideration when coding detouring is that the parents must be actively supporting each other in an alliance against the child. Therefore, detouring was not coded when parents were rejecting to the child in independent ways or if one parent was rejecting the child while the other was simply passive (e.g., mother but not partner engaged in harsh or rejecting parenting behavior in the task). Two trained coders independently rated more than 20% of the videos at each wave to assess interrater reliability. Intraclass correlation coefficients were .83 at Wave 1 and .75 at Wave 2.

Although this code is being used for the first time in the present study, findings from supplementary analyses support its validity as an assessment of detouring. Specifically, we examined associations between our detouring measure and coparenting relationship dimensions in the tower-building task that were coded by a different team of independent raters. First, supporting convergent validity and consistent with theoretical conceptualizations of detouring (e.g., Kerig, 2016), higher ratings of detouring were significantly correlated with observational ratings of coparental collusion (i.e., the degree to which the coparenting relationship is characterized by prioritizing the importance of the partner relationship at the expense of the parents’ relationship with the child; Parry & Davies, 2017) (r = .58, p < .001). Second, supporting discriminant validity of our detouring code, detouring was unrelated to what should be distinct coparenting relationship constructs, including coparental warmth, competition, conflict, or balanced participation in caregiving roles (all ps ≥ .25).

Family Negativity and Conflict.

At Waves 1 and 2, the video-recorded FITs were also coded using the Negativity and Conflict scale of the System for Coding Interactions and Family Functioning (SCIFF; Lindahl & Malik, 2000). The Negativity and Conflict scale assesses the overall negative tone or level of tension in the family and was coded on a 5-point scale ranging from 1 (very low) to 5 (high). To be rated highly on Negativity and Conflict, displays of negative affect (e.g., tension, anger, irritation) must be present throughout most (e.g., more than half) of the interaction and fairly intense in nature (e.g., raised voices, comments and behaviors that convey annoyance, frustration, or hostility). More mildly, this code involves moments of tension, frustration, or anger that are partially resolved, even if some lingering tension may occur. The same two trained coders independently rated over 20% of the videos at each Wave to assess interrater reliability. Intraclass correlation coefficients were .88 at Waves 1 and 2.

Child Hostile Attribution Bias.

To assess children’s hostile attribution bias at Waves 1 and 2, a trained experimenter administered a structured interview with each child after the presentation of each of six peer problem stories adapted from the Social Problem Solving (SPS) vignettes (Dodge et al., 1995). Experimenters encouraged each child to pretend that they were the protagonist in the stories (e.g., “Pretend that you are playing outside at recess and some other kids are playing a game. You would really like to play the game with them, but they haven’t asked you.”). Of the six vignettes, two depicted challenges entering a group of peers with ambiguous intent (e.g., the child wants to play a game with other kids, but they have not invited him/her to play), two depicted provocation with indication of negative peer intent (e.g., another child takes the child’s place in line), and two depicted transgressions with ambiguous peer intent (e.g., the child lets peers play with his/her favorite toy while he/she leaves the room for a few minutes, and the toy is broken when he/she returns).

Children’s responses to three questions from the structured interview were used to assess their hostile attribution bias in each story. Specifically, children were first asked “How come the boy/girl [description of the specific peer event in the story]?” Children’s open-ended responses for each of the six stories were recorded verbatim for subsequent coding. Children also responded with “yes” or “no” to the following two questions for each story: (1) “Are the kids being mean to you?” and (2) “Did the other kids do it on purpose?” Children’s transcribed responses were coded by two trained raters for hostile attributions using a 5-point scale ranging from 1 (very benign) to 5 (very hostile) that incorporated the responses to all three questions. Ratings of 1 were reserved for responses that uniformly and definitively conveyed that peer motives in the story were completely benign (e.g., [why did they do it?] “it was an accident”, [are they being mean to you?] “no”, [was it on purpose?] “no”). Conversely, ratings of 5 reflected responses that consistently conveyed hostile or malevolent intent on behalf of the peers (e.g., [why did they do it?] “they hate me”, [are they being mean to you?] “yes”, [was it on purpose?] “yes”). The two trained coders overlapped on 100% of the children’s responses to assess interrater reliability. Intraclass correlation coefficients for each story were acceptable: mean ICC = .92 (range .90 to .95) at Wave 1 and mean ICC = .90 (range = .82 to .94) at Wave 2. Ratings for the six stories were averaged into a single composite of hostile attribution bias at each wave (αs = .73 at Wave 1 and .69 at Wave 2).

Child Externalizing Problems.

Maternal report on the Externalizing scale of the MacArthur Health and Behavior Questionnaire (HBQ; Ablow et al., 1999) was used to assess child externalizing problems at Waves 1 and 2. The Externalizing scale consisted of the average of three subscales on the HBQ: (a) Relational Aggression (mean of 6 items; e.g., “tells others not to play with or be a peer’s friend”), (b) Overt Hostility (mean of 4 items; e.g., “kicks, bites, or hits other children”), and (c) Conduct Problems (mean of 12 items; e.g., “destroys things belonging to his/her family or other children”). Response alternatives for each item were: 0 (Never or not true), 1 (Sometimes or somewhat true), and 2 (Often or very true). Internal consistencies of the resulting scale were acceptable at both time-points (internal consistency of three subscales within scale: αs = .80 at Wave 1 and .89 at Wave 2).

Child Peer Rejection and Victimization.

Assessments of peer rejection and victimization at Waves 1 and 2 were derived from teacher reports on the Peer Relations scale of the HBQ. The scale consisted of the average of two subscales: Bullied by Peers (mean of 3 items; e.g., “is picked on by other children”) and Peer Acceptance/Rejection (mean of 8 items; e.g., “actively disliked by other children who reject him/her from their play”). Each item was rated along four-point scales ranging from 1 (Not at all like child) to 4 (Very much like child). Internal consistencies of the resulting scale were acceptable at both Waves (internal consistency of two subscales within scale: αs = .70 at Wave 1 and .65 at Wave 2).

Covariates.

Two demographic covariates, derived from a maternal interview at Wave 1 included children’s sex (0 = females; 1 = males) and total household income per capita, calculated by dividing total annual household income by the number of individuals living in the home.

Results

Descriptive Analyses

Table 1 provides the means, standard deviations, and intercorrelations for the variables used in the primary analyses.

Table 1.

Means, standard deviations, and intercorrelations of the main variables.

	M	SD	1	2	3	4	5	6	7	8	9	10	11
1. Child Sex	.43	.50	--
2. Household Income Per Capita	9.97	6.81	.09	--
Wave 1 (Kindergarten)
3. Detouring	2.77	2.02	−.02	−.21*	--
4. Negativity & Conflict	2.71	1.36	.03	−.26*	.31*	--
5. Hostile Attribution Bias	3.82	.86	.03	−.02	−.04	-.06	--
6. Externalizing Problems	.22	.23	-.00	−.18*	.11	.12	.02	--
7. Peer Rejection/Victimization	1.40	.48	−.03	−.23*	.10	.13	−.01	.30*	--
Wave 2 (First Grade)
8. Detouring	1.99	1.39	−.06	−.11	.32*	.15	−.05	.16*	.17	--
9. Negativity & Conflict	2.69	1.07	.09	−.22*	.24*	.45*	−.12	.11	.11	.09	--
10. Hostile Attribution Bias	3.64	.76	−.02	−.05	.18*	.11	.17*	.05	.01	.02	.11	--
11. Externalizing Problems	.22	.27	−.01	−.14*	.20*	.17*	−.02	.77*	.29*	.29*	.09	−.04	--
12. Peer Rejection/Victimization	1.42	.47	−.09	−.17*	.20*	.02	−.04	.26*	.47*	.28*	.04	.03	.26*

Note. Child sex: 0 = females, 1 = males; income in thousands.

^*p < .05.

Analytic Strategy for Primary Analyses

Prior to conducting our analyses, we examined whether rates of missingness in the dataset were associated with any of the 12 study variables and covariates. Only three analyses were significant: higher rates of missingness were associated with peer rejection and victimization at Wave 2 (r = .28, p < .001) and family negativity and conflict at Waves 1 (r = .17, p = .02) and 2 (r = .17, p = .04). Hypotheses were tested using autoregressive path analysis within a structural equation modeling (SEM) framework, specified with Mplus Version 7.4 statistical software (Muthen & Muthen, 1998–2012). To maximize the size of the sample, full-information maximum likelihood estimation (FIML; Enders, 2001) was used to accommodate missing data (12.39% of values). The FIML method is regarded as the method of choice for estimating missing data because it minimizes bias in regression and standard error estimates for all types of missing data (i.e., MCAR, MAR, NMAR) when the amount of missing data does not exceed 20% (Schlomer, Bauman, & Card, 2010).

To address our primary analytic aim of examining the nature and directionality of associations between family (i.e., detouring, negativity and conflict) and child (i.e., externalizing problems, hostile attribution bias, peer rejection and victimization) variables over time, we specified an autoregressive path model using repeated measures of all primary constructs across the two time-points. All autoregressive paths between contiguous measurement occasions were freely estimated, and structural paths were estimated between: (a) both family-level variables at Wave 1 and the three child variables at Wave 2 and (b) the three child variables at Wave 1 and the two family-level variables at Wave 2. Correlations were estimated among all exogenous (Wave 1) predictors and between residual error terms on all endogenous (Wave 2) outcomes. Given their potential roles as covariates, total household income per capita and child sex were included as covariates in the primary analyses. However, results indicated that child sex was unrelated to any of the other variables in the model and did not alter the pattern of significant results. Therefore, to maximize parsimony, child sex was dropped from the model, and only income was included as a covariate in the primary analyses. The resulting model had eight degrees of freedom (i.e., two degrees of freedom from a priori exclusion of cross-lagged paths between detouring and family conflict and six degrees of freedom from a priori exclusion of cross-lagged paths among the three child outcomes). Model fit was assessed using standard criteria, including chi-square (χ²) tests, root-mean-square error of approximation (RMSEA), the comparative fit index (CFI), and the standardized root mean squared residual (SRMR). Acceptable fit values include a non-significant χ², RMSEA and SRMR of .05 or lower, and CFI of .90 or higher, with .95 and higher preferred (Hu & Bentler, 1999).

Results of Primary Analyses

The path model shown in Figure 1 provided an excellent fit with the data, χ² (8, N = 218) = 4.15, p = .84; RMSEA = .00; CFI = 1.00; SRMR = .02. For clarity, only significant correlations are depicted in Figure 1. Autoregressive paths from Wave 1 to Wave 2 were significant for detouring (β = .27, p = .001), negativity and conflict (β = .43, p < .001), hostile attribution bias (β = .18, p = .01), externalizing problems (β = .75, p < .001), and peer rejection and victimization (β = .45, p < .001). Total household income per capita did not predict changes in any of the child or family variables over time (all ps ≥ .47)

Figure 1.

Autoregressive path model examining associations among detouring, family negativity and conflict, child hostile attribution bias, externalizing problems, and peer rejection and victimization. Parameter estimates for the structural paths are standardized path coefficients. Dashed lines indicate non-significant pathways. For clarity, only significant correlations are shown. W1 = Wave 1 (Kindergarten); W2 = Wave 2 (First Grade). * p < .05. ** p < .01. *** p < .001.

Findings for the structural paths supported our hypotheses that detouring would predict increases in adjustment problems over time for children over and above family negativity and conflict. Specifically, higher ratings of detouring predicted children’s greater hostile attribution bias (β = .16, p = .04), externalizing problems (β = .11, p = .03), and peer rejection and victimization (β = .19, p = .03) at Wave 2 after controlling for Wave 1 autoregressive paths. Conversely, negativity and conflict did not predict increases in any of the child variables over time (all ps ≥ .34), and none of the child variables predicted increases in detouring (all ps ≥ .29) or family negativity and conflict over time (all ps ≥ .17).

Discussion

Detouring in the family system has been theorized to predict a wide variety of negative adjustment outcomes for children (e.g., Kerig, 1995, 2016; Minuchin, 1974), but very few studies have empirically examined these expected associations. To address this significant gap in knowledge, our longitudinal, multi-method, multi-informant study tested whether detouring uniquely predicted increases in children’s hostile attribution bias, externalizing problems, and peer rejection and victimization during the transition from kindergarten to first grade. Findings indicated that detouring predicted increases in all three adjustment outcomes above and beyond a general construct of family negativity and conflict. Furthermore, in contrast to hypotheses proposed by child effects models, none of the child variables predicted increases in either detouring or negativity and conflict over time.

Building on the few prior cross-sectional examinations of associations between detouring and child adjustment problems (e.g., Kerig, 1995; Yahav & Sharlin, 2002), our findings break new ground in establishing detouring as a unique predictor of prospective changes in children’s hostile attribution bias, externalizing problems, and peer rejection and victimization in early childhood. Speaking to its salience as a risk factor, detouring predicted all three adjustment outcomes, while the general construct of family negativity and conflict did not. These findings support FST’s assertion that there is inherent value in identifying the relational function (i.e., purpose or motive based on the context of the interaction) underlying the organization and pattern of behaviors within the family system (Combrinck-Graham, 1985, 1989; Minuchin, 1974). Whereas assessments of negativity and conflict capture physical forms of family behaviors (e.g., yelling, frustration, anger), there could be multiple explanations or functions of these behavioral displays, and they may not necessarily be directed at the child. However, detouring within the family specifically functions to increase closeness in the interparental subsystem at the expense of closeness with the child (Minuchin, 1974). Therefore, the child is actively excluded from a key part of the family system and denied resources provided by the family (Fauber et al., 1990). This differs from parents’ engagement in coordinated coparenting, wherein parents display mutual support for one another as parents, respect one another’s authority, and create a structured, safe climate for the child (Feinberg, 2003). Whereas these behaviors signify a supportive parental relationship and are generally linked with positive developmental outcomes for children (Feinberg, 2003), the targeted rejection and exclusion experienced by the child in a detouring family may predict his or her adjustment problems (Kerig, 1995, 2016; Minuchin, 1974).

Our findings that detouring predicted increases in all three of the outcomes (i.e., hostile attributions, externalizing, and peer victimization/rejection) examined in this study are particularly consequential because research has shown that these behaviors cohere to a high-risk pattern of reactive aggression (Card & Little, 2006; Dodge, 2006; Hubbard, McAuliffe, Morow, & Romano, 2010). Aggression that is reactive (e.g., defensive or “hot-blooded”; involves angry, emotionally dysregulated responses to perceived offenses or frustrations) serves as a potent predictor of psychosocial adjustment problems across the lifespan (e.g., delinquency, peer victimization, emotion dysregulation, and ADHD symptoms, internalizing problems, poor social skills; Card & Little, 2006). In young children at high risk for reactive aggression, a strong hostile attribution bias often goes hand in hand with difficulties regulating behaviors and emotions (e.g., externalizing problems; Choe et al., 2013; Shields & Cicchetti, 1998; Teisl & Cicchetti, 2008). Because these children are highly dysregulated, reactive, and aggressive, several studies have shown that they also tend to be bullied or rejected by their peers (e.g., Card, & Little, 2006; Schwartz et al., 1998). Over time, this pattern of behaviors crystallizes into a stable profile of reactive aggression, and links between reactive aggression and adjustment problems become stronger as children get older (Card & Little, 2006; Dodge, 2006; Hubbard et al., 2010). Therefore, because detouring was linked with all three risk factors for reactive aggression, these children may be at especially high risk for a wide array of difficulties across the lifespan.

However, it is also important to note that not all three child outcomes in our study were related to one another. Although externalizing problems and peer rejection and victimization were associated with each other at both time-points (rs = .30 at Wave 1 and .26 at Wave 2), hostile attribution bias was not related to either outcome at either time-point. Consistent with the concept of multifinality in developmental psychopathology (Cicchetti & Rogosch, 1996), it is possible that detouring may lead to different outcomes for different children based on the context in which it occurs and the variability in child, parent, and family characteristics. As one example, although detouring families all share the function of increasing closeness in the interparental relationship at the expense of the relationship with the child, it may be expressed in somewhat different ways. In some cases, parents may actively exclude the child or be openly negative, hostile, and rejecting toward the child (e.g., scapegoating; Buchanan & Waizenhofer, 2001). However, detouring also occurs when parents achieve some degree of intimacy or closeness, albeit in a tenuous and, in many cases, artificial way, through excessive, shared concerns about the child’s well-being (e.g., fragilizing the child; Minuchin, 1974). Therefore, in other cases, it may be expressed as a shared concern for the child and his or her distress (Kerig, 1995). Ultimately, this still creates closeness in the interparental relationship, and they tend to interact more with each other about the child’s behavior to the relative exclusion of the child in the interaction sequence. It could be that children’s increased hostile attribution bias emerges only in cases when detouring is expressed in the former, rather than latter, behavioral pattern. Moreover, individual differences in child temperament (e.g., negative emotionality, inhibitory control) and emotion regulation may also alter the ways in which children are impacted by detouring and how this affects their adjustment outcomes. Alternatively, it is also possible that (a) differences in the methods used to assess hostile attribution bias (i.e., coded responses to vignettes) versus the other two outcomes (i.e., questionnaire) could explain the lack of a correlation, or (b) at this relatively early developmental stage, the forms of adjustment are still in the process of consolidating together and may emerge as a correlated suite of behaviors as children progress through the elementary school years. Further research is needed to identify potential mechanisms and moderating factors that explain the variability in outcomes for children who develop in families characterized by detouring.

In addressing why detouring may predict child adjustment problems in our study, it is possible that children’s appraisals of themselves and their families may be key underlying mechanisms. Script theory asserts that children’s early experiences with family relationships are reflected in their internal working models of themselves and their families (Johnston et al., 2009). Therefore, when these early relationships are disrupted or unsupportive, there may be negative repercussions for the ways children view themselves (e.g., incompetent, worthy of blame, victim) and their families (e.g., uncaring, unsupportive, volatile, threatening). Because children tend to rely more heavily on prior representations as a way to simplify and comprehend challenging and novel interpersonal contexts (Johnston et al., 2009), children from detouring families who are repeatedly targets of parental rejection or exclusion may in turn draw on these internal representations in interpreting others’ actions as being aggressive and purposeful (i.e., hostile attribution bias), struggle with regulating their anger and aggressive behavior (i.e., externalizing problems), and seek out or structure interactions in other settings where they are also excluded, rejected, or targeted (e.g., peer victimization and rejection). This interpretation is consistent with findings from other studies that children’s internal representations of themselves or their families mediate associations between other indices of relationship discord (e.g., family instability, interparental conflict) and adjustment problems (e.g., Coe, Davies, & Sturge-Apple, 2017, 2018).

Research guided by the cognitive-contextual framework and emotional security theory may also point to other potential underlying processes. For example, although not directly examined in relation to detouring, findings from studies examining adolescent triangulation into interparental conflict (e.g., cross-generational alliances) identify children’s appraisals of self-blame for interparental conflict and emotional reactivity in the face of conflict as mediators of associations between triangulation and psychological adjustment problems (e.g., Buehler & Welsh, 2009; Fosco & Grych, 2010). Extrapolated to our findings, it is possible that when children repeatedly experience detouring in family interactions, they may begin to blame themselves for family conflict and be much more reactive to conflict when it occurs. In turn, these cognitions and emotions may manifest in subsequent adjustment problems. Additionally, as a complementary interpretation of our results, coercive family process models posit that if aggressive behavior is effective in defusing interparental conflict by deflecting attention onto the child, parents may inadvertently reinforce this behavior by engaging in detouring rather than arguing with one another (Davis, Hops, Alpert, & Sheeber, 1998). In turn, child behavior problems continue and proliferate into other contexts (e.g., at school, with peers). Because our study did not explicitly test intermediary processes in the predictive pathways, it will be important for future research to further explore what underlying mechanisms account for associations between detouring and child adjustment problems.

In contrast to our primary hypothesis that detouring would predict increases in child adjustment problems over time, we did not find any empirical support for propositions of child-driven effects models that child behavior problems would serve as precursors of detouring. Our use of a multi-method, multi-informant measurement battery may partially explain the null results. Specifically, researchers have posited that parents’ perceptions of their children’s behavior, rather than the behavior itself, predict how they interact with their children (Webster-Stratton, 1990). Therefore, we may not have identified any child-driven effects on detouring or negativity and conflict because two of the child outcomes (hostile attribution bias, peer rejection and victimization) were assessed using teacher reports and coder ratings of child responses to vignettes. Furthermore, it could be that bidirectional relations between children’s behavior and detouring happen on a moment-to-moment basis. Therefore, global assessments of adjustment problems may not fully capture how parents and children respond to one another in the moment to perpetuate a coercive transactional process between child misbehavior and detouring in the family. Micro-level coding systems designed to capture child behavior and reciprocal relations between parent and child behavior in the context of the task may reveal transactions not evident in the data used in the current study (e.g., Elam & DiLalla, 2018). Alternatively, it is possible that bidirectional relations become more pronounced when children get older. For example, many studies demonstrating support for child-driven effects on family functioning have been with samples of children in middle childhood and adolescence (e.g., Cui et al., 2007; Fite et al., 2006; Jenkins et al., 2005; Kerr & Stattin, 2003). However, studies that test these types of models with younger children have not consistently found such associations (e.g., Coe, Davies, Hentges, & Sturge-Apple, 2019; Eisenberg et al., 2005). Further research with more waves of data and alternative methods is needed to more thoroughly examine potential bidirectional associations between detouring and child functioning.

Findings must also be interpreted in the context of our study’s limitations. First, although families in our study were from a variety of racial, ethnic, and socioeconomic backgrounds, caution should be exercised in generalizing our results to families experiencing higher risk, high affluence, or those with children in different developmental periods (e.g., middle childhood, adolescence). Second, effect sizes of predictive pathways were relatively modest in magnitude, which points to high variability in the experiences of families and children who experience detouring. It will be critical for future research to examine whether other parent (e.g., psychopathology), child (e.g., temperament), and family (e.g., conflict) factors moderate these associations. Third, the novel observational assessment of detouring utilized in our study may be regarded as a strength, but it will be important for future work to replicate and further expand on our findings. Fourth, exploring characteristics of the child (e.g., self-blame, cognitive appraisals, internal representations, emotional reactivity) and parent (e.g., parenting difficulties, childrelated cognitions) as explanatory mechanisms for the effects of detouring on child outcomes is an important future direction for research. Fifth, because our study was limited to only two waves, future research should more fully address how detouring in the family system may impact the potentially dynamic and reciprocal relationship between hostile attribution bias, externalizing problems, and peer rejection and victimization over time.

In addition to potential limitations of generalizability and conceptual considerations, it is also important to acknowledge the potential limitations of our measures and analytic approach. First, in the full analytic model, there was a negative correlation between the residual error variance of Wave 2 hostile attribution bias and externalizing problems (r = −.15). This could be the result of a method effect or another variable outside of the model having a negative effect on one outcome and a positive effect on the other. However, because (a) the variables are not related at a bivariate level, (b) there is not a theoretically or empirically guided explanation, and (c) the correlation is modest in magnitude, it is likely this inverse correlation is spurious. Second, although our study utilized measures that have been demonstrated to have strong psychometric properties in prior research, internal consistency of our measures of hostile attribution bias and peer rejection and victimization in our sample were on the lower end of acceptable. It will therefore be critical for future research to replicate our findings in other samples of families and children. Similarly, because all variables included in analytic models were manifest (i.e., we did not use any latent variables), our constructs still contain measurement error and can result in biased estimates. Future work with larger samples and more expansive measurement batteries for each construct will be important to afford analyses using latent variables with multiple indicators. Finally, although our use of autoregressive cross-lagged modeling with repeated measures of all family and child constructs allowed for examination of possible child-driven effects on family functioning, it will be critical for future research to more definitively test whether bidirectional relations exist using multiple waves of data (i.e., three or more time-points) and other analytic approaches (e.g., Berry & Willoughby, 2017; Hamaker, Kuiper, & Grasman, 2015).

In summary, findings from our multi-method, multi-informant, longitudinal study indicated that detouring in the family system uniquely predicted increases in young children’s hostile attribution bias, externalizing problems, and peer rejection and victimization, over and above a general construct of family negativity and conflict. Results support the developmental utility of FST boundary conceptualizations in understanding how characteristics of the broader family system impact child development. Applying FST boundary constructs from family therapy to developmental research designs has the potential to better inform interventions (e.g., Cowan & Cowan, 2006) and lead to a more in depth understanding of the developmental implications of different family dynamics.