Externalizing and Internalizing Problems : Contributions of Attachment and Parental Practices

The relation between attachment and parental practices with externalizing (aggression and delinquency) and internalizing (social withdrawal and anxiety/depression) behavioral problems were investigated in this study. Participants were 289 children (from 9 to 12 years old) and 205 caregivers who answered distinct questionnaires: the formers on attachment and the later on parental practices. Results indicated that poor maternal attachment relationships, high levels of parental rejection and being a boy predicted aggression. Moreover, poor paternal attachment and father’s low education predicted social withdrawal. Finally, parental rejection was marginally associated with anxiety/depression. The results reinforce, partially, the existing literature and help to understand the complex relationship between parenting and behavioral problems.

Behavioural problems that precede or are concomitant with most common mental disorders in childhood and adolescence have been the focus of interest of Developmental Psychopathology (Cicchetti, 1984;Sroufer & Rutter, 1984).Within the theoretical framework of Developmental Psychopathology, children who face diffi culties in their relations with peers can be grouped into two broad categories: those with "externalizing problems" -little control (undercontrol) over their emotions, thoughts and behaviours -and those with "internalizing problems" -excessively control (overcontrol) of these processes (Achenbach & Edelbrock, 1978;Achenbach, Howell, Quay, & Conners, predictors of internalizing and externalizing behaviours, in childhood and adolescence, have been the focus of great attention from researchers, in different cultures.Among the many factors involved in the genesis and course of these problems, those related to the quality of parent-child relationships have gained attention in the fi eld of Developmental Psychopathology (Masten et al., 2005).
The parent-child attachment, understood as an innate behavioural system, whose basic function is to search for parental closeness and security, promotes the establishment of an early and persistent bond between the baby and his/ her primary caregiver.Even in the most advanced stage of childhood mothers and fathers tend to be cited as the main source of social support on children's lives (Furman & Buhrmester, 1992).Even though the involvement of both parents in raising their children constitute a fundamental factor in promoting children's adaptative behaviours and, in inhibiting risk behaviours as well, several researchers, in different cultures, have suggested that mothers and fathers play different roles on their children's development (Boller et al., 2006;Bandeira, Goetz, Pontes, & Vieira, 2005).Therefore to investigate children's behavioural problems, not just is necessary to examine the child-mother attachment, but also examining the child's relation with father becomes paramount.
Most studies focusing on attachment relationships and their implications for development has directed its interest to the early relationship between mothers and infants, or mothers and preschoolers.However, few studies have focused on the features and quality of attachment in middle childhood and preadolescence (Dwyer, 2005).In the second half of this decade, some important research on this subject were published (Booth-Laforce et al., 2006;Diener, Russell, Behunin, & Wong, 2008), but they were mostly restricted to the U.S.A. and Europe.Yet, a small number of these studies included, besides the mother, the father in their analysis (Booth-Laforce et al., 2006).It has been shown that the attachment relationship between children and their fathers would be more salient in predicting children's social-emotional functioning, especially, regarding internalizing problems (Brumariu & Kerns, 2010;Desjardins & Leadbeater, 2011;Roelofs, Meesters, ter Huurne, Bamelis, & Muris, 2006).
In investigating the quality of child's attachment with both parents, it is essential to characterize what is meant by attachment in middle and late childhood.According to Kerns, Keplac and Cole (1996), the quality of this relationship, at this stage of development, derives from the child´s perception of: (a) unconditional parental availability and responsiveness, (b) an ease and open communication, particularly in time of more intense emotional states, and (c) trust and emotional support that can be received in times of stress.
Besides the quality of the attachment, the educational practices adopted by parents toward their children tend to interfere in the occurrence of behavioral problems in the course of their development.Educational practices refer to a set of behaviors or specifi c strategies that parents use in everyday situations in order to raise their children to socialize them according to culturally accepted standards (Darling & Steinberg, 1993).
Regarding parental sociodemographic variables, for some researchers, the experience gained by parents through life course tend to favour the use of positive parental practices such as behavioural control and to restrain the use of negative practices such as rejection and psychological control (Bornstein & Putnick, 2007;Marin & Levandowski, 2008).Other studies point to an association between low parental warmth and family and community low socioeconomic status (having low educational level as a leading indicator; Klebanov, Brooks-Gunn, & Duncan, 1994).
The goal of this study was to examine the contributions of: (a) the quality of child attachment (to the mother and/ or to the father), and, (b) the parental practices (rejection and/or behavioural and/or psychological control) in the occurrence of externalizing problems (aggression and delinquency) and internalizing diffi culties (social withdrawal and anxiety/depression).Since these aspects confi gure two distinct yet interconnected parental systems, it is expected that both, attachment quality and the parental practices would contributes uniquely for variability in internalizing and externalizing behavioral problems.

Participants
The study included 289 children (48.2% girls, 51.8% boys) from two schools in each of two cities in southern area of Brazil, and 205 caregivers (181 mothers, 18 fathers and 6 other caregivers), accounting for a total of 494 subjects.The children had an average of 10.5 years (SD = .77),mothers, 37.28 years (SD = 7.06) and fathers 41.13 years (SD = 7.92).Most mothers (84.4%) and fathers (82.4%) had primary or secondary level of education.

Measures
The quality of the attachment was accessed by the Security Scale (Kerns et al., 1996), which consists of 15 items for mothers (α =.74) and the same 15 items for fathers (α =.81).Example, "Some kids fi nd it easy to rely on his/her mother for help, BUT other kids fi nd it hard to tell his/her mother when they need it".Children were instructed to fi rstly choose, between the two, the statement that best described them and, next, to indicate the degree of agreement (1 = partly true and 2 = true) with the statement.The internal consistency reported for these and other scales employed were derived from the sample data analyzed in this study.
The dimensions of parenting practices were accessed through the Child Rearing Practices-Report Questionnaire (CRPR-Q) reviewed by Rickel and Biasatti (1982).The CRPR-Q is composed of 42 items, with a six-point scale (with 1 = strongly disagree and 6 = strongly agree) that access the basic dimensions of parenting: control (restrictiveness) and affection (nurturance).Exploratory Factor Analyses identifi ed two dimensions: 1. Rejection, composed of three items (α =.69) denoting disappointment, hostility and confl ictual interaction between parent and child (e.g "I feel a little disappointed with my child "). 2. Control, which in subsequent analyses was divided into two factors: (a): behavioral control, with 14 items (α =.70) that accesses how parents supervise their children's behavior, through means of plausible requirement of maturity, and through monitoring (Ex: "I keep informed about where my child is and what he/she is doing") and (b) psychological control, with 13 items (α =.69), marked by parental intrusiveness aiming to control not only the child´s behavior but also his/her emotions, and done by means of affection withdrawal threatening, guilt induction, and other forms of emotional manipulation (Ex: "I believe that my child should be aware of how much I sacrifi ce myself for him/her").The internalizing and externalizing problems were accessed separately through four subscales of the Child Behavior Checklist 4-18 years (Achenbach, 1991).For externalizing problems we employed two subscales: (a) aggression (α = .90),which contains 21 items that describe openly aggressive behaviors, and (b) delinquency (α = .83),with 15 items covering behaviors that violate legal or moral rules.For internalizing problems we employed the following subscales: (a) social withdrawal (α = .75),composed by 10 items that describe behaviors of social isolation, shyness and social diffi culties with peers, and (b) anxiety/depression (α = .82),which contains 18 items that access symptoms of depressive behavior and thinking and generalized anxiety.

Data Collection Procedure
This study was approved by the Ethics Committee in Human Research of the Universidade do Vale do Rio dos Sinos (UNISINOS), process number 038/2009, on May 29th 2009.Once obtained the approval of the Ethics Committee, the research proposal was submitted to and approved by the Principal of the schools.All the children received an informed consent to be read and signed by their parents/caregivers.The sample was composed only by the children whose parents/caregivers signed the informed consent form.Children completed the questionnaire that accesses attachment quality towards both parents, during regular class hours reserved for the study.The caregivers completed two questionnaires, one regarding their parenting practices and the other regarding their children´s behavior problems at school or at another prearranged locations (as in community centers or in their homes) in individual or group situation.

Data Analysis
Data analyses were performed using descriptive (frequency, mean and standard deviation) and inferential statistics (correlation analysis and multiple regression analysis).The Pearson r was employed as the correlation coeffi cient in this study, since the data were parametric, with a signifi cance level of .05 or lower.
The hierarchical multiple regression analyses were used to investigate the unique and combined contribution of the explanatory variables (quality of attachment and parenting practices) on the criteria variables (externalizing and internalizing behavior problems).In hierarchical regression analysis, the variables were included in the regression equation in different blocks, which allowed to control the prior variables, allowing also to make an integrative analysis, in which the effects of all explanatory variables on externalizing and internalizing problems (criteria variables) were taken into account.

Results
Before proceeding with the predictive analysis, we performed a series of correlational analyses between the explanatory and the criteria variables, which are presented in Table 1.According to the results presented in Table 1, poor maternal attachment was correlated with aggression, delinquency and social withdrawal in childhood, and showed a tendency to be correlate with depression/anxiety.However, the quality of paternal attachment was not associated with any of externalizing problems.With regard to internalizing problems, poor paternal attachment was associated with social withdrawal and showed a tendency to be associated with child depression/ anxiety.
The investigation of the correlations between the different dimensions of parenting practices and behavior problems indicated that high levels of parental rejection and low behavioral control correlated with aggression and delinquency.In contrast, only parental rejection correlated with anxiety/depression.Social withdrawal did not correlate neither with parental control practices nor with rejection, and, fi nally, psychological control did not correlate with any behavioral problem (externalizing or internalizing types).
Hierarchical regression analyses were performed through stepwise method due to its exploratory nature.The explanatory variables were included in a single block in the regression equation.The socio-demographic variables (children´s sex and age, mother´s and father´s education level and age) were included in a separate block to see if the presence of these variables would affect the predictive value of the explanatory variables and would change the multiple regression coeffi cients.The gender variable was transformed into a dummy variable.The results are reported below in Table 2. Table 2 shows that maternal attachment and parental rejection predict aggression [F (2.156) = 5.50, p ≤ .01].Together the two variables accounted for 7% of the variance on aggressive behavior in children.By adding the socio-demographic variables to the model, the regression coeffi cient increased with the presence of the variable sex of the child.Being a boy offered an additional contribution to the model (r ² changed from .07 to .09).Thus, the best regression model was the one which combined high levels of parental rejection, poor attachment with the mother and the fact of being a boy.Together, these variables explained 9.1% of the variance on child aggression.
Similarly, with respect to delinquency, the results indicated that poor maternal attachment and parental rejection predict delinquency [F (2,170) = 9.09, p ≤ .001].The two variables together explained 10% of the variance on child delinquency.However, unlike the model of aggression, the model that predicts delinquency was not altered with the inclusion of socio-demographic variables.Regarding the model for social withdrawal (see Table 3) that included only the explanatory variables in the regression equation (attachment quality towards both parents, rejection, behavioral and psychological control) we found that only paternal attachment was a good predictor [F(1, 165) = 9.62, p ≤ .01].The addition of socio--demographic variables in the regression model produced a small increase in the regression coeffi cient (from .05 to .073), if one considers the tendency of association of the variable level of paternal education.Thus, specifi cally, poor paternal attachment and lower levels of paternal education offered unique contributions to the model and, combined, explained 7.3% of the variance on child social withdrawal [F(2.164)= 6.43, p ≤ .01].
On the other hand, when we entered only the explanatory variables into the model, only parental rejection [F(1.156)= 4.52, p ≤ .05]was predictive of anxiety/ depression.In fact, parental rejection explained 2.8% of the variance on this dimension.With the entry of socio--demographic into the model a different picture emerged -maternal age, paternal age and educational level were marginally signifi cant in predicting anxiety/depression [F(4.153)= 3.13, p ≤ .05].Thus the fi nal predictive model was the one that grouped unique contributions of parental internalizing problem as it was done here or, in some cases, the studies measure only symptoms of anxiety and depression, not the indicators of social withdrawal (Roelofs et al., 2006;Rohner & Britner, 2002).
Studies on social withdrawal (Rubin, Burgess, & Hastings, 2002;Rubin, Coplan, & Bowker, 2009) indicate that the affective climate that exists between the inhibit child and his parents can be both of acceptance and affection and of little affection as well.That is, little affection does not seem to be, necessarily, a characteristic of parents of children socially inhibited.In fact, for these authors, it is the characteristic of being over solicitous (parents who are controlling, very affectionate, but, paradoxically, little responsive to the real needs of children) which may put children at risk of becoming socially withdrawn.
In relation to externalizing problems, the results suggest that aggressive children with signs of delinquency also tend to have caregivers whose strategies for behaviour control are ineffective, confi rming fi ndings in the literature (Aunola & Nurmi, 2005;Barber, 1996;Laird et al., 2010).However, contrary to the fi ndings of various researchers, psychological control was not correlated with either externalizing problems (Aunola & Nurmi, 2005;Barber, 1996;Finkenauer et al., 2005), nor with internalizing problems (Aunola & Nurmi, 2005;Barber, 1996;Finkenauer et al., 2005;Karreman et al., 2009).At this point, it is necessary to consider the meaning that psychological control would have on our culture.As in the other Latin American cultures, psychological control can be understood as a socially acceptable way of promoting the need for children to be good citizens and of teaching the children to respect (and even feared) parental authority (Halgunseth, Ispa, & Rudy, 2006).Thus, unlike the negative impact that forms of psychological control appear to have on Euro-American children (For a review see Barber, 1996), Brazilian children can be resilient to the effects of maternal emotional manipulation strategies when it is perceived as a way of correcting their behavior.Indeed, in this present study, caregivers had higher scores on items that denoted emotional control and overprotection of children (eg, "I do not let my daughter questioning my decisions" and "I prefer that my child not try things she/he may fail "and on items denoting clearly emotional manipulation strategies, which typically would induce anxiety in children of Western developed countries (e.g,"I believe that my daughter should be aware of how much I sacrifi ce for she/ he","I say to my daughter how much I feel ashamed and disappointed when she/he misbehaves").

Predictive Models
For the externalizing problems investigated, at least one variable in each parental system, specifi cally, a fragile maternal attachment and high levels of parental rejection, offered unique contributions to explain both delinquency and aggression (although, in this case, the regression coeffi cient of the variable maternal attachment was only rejection, maternal age, paternal age and paternal educational level and together these variables explained 5.1% of the variance on anxiety/depression in childhood.

Discussion
This study was based on the assumption that both the infant attachment system and the parenting practices system contribute uniquely to exacerbate or minimize the signs of behavior problems in childhood, due to its negative or positive quality.At the same time, it was assumed that such contributions would continue to exist, even if the effects of socio-demographic variables that have been shown to be important in explaining each of the behavioral problems in the study were controlled for.Therefore, a series of correlational hypotheses were tested preliminarily, which were partially confi rmed, and the results are discussed below.

Mother´s and Father´s Attachment Quality and Behaviour Problems
Children who, reported they are sure that their mothers are available, responsive and supportive had also low scores in aggression, delinquency, social withdrawal and had a strong tendency to be less anxious and depressed.These fi ndings are in accordance with in the international literature which indicates a negative relationship between secure maternal attachment and externalizing problems (Fearon et al., 2010) and between secure maternal attachment and internalizing problems (Brumariu & Kerns, 2010;Nishikawa, Hägglöf, et al., 2010).
In contrast to the fi ndings of Booth-Laforce et al. ( 2006), paternal attachment was not correlated with externalizing problems in the sample studied.On the other hand, children who reported low confi dence in the availability, responsiveness and emotional support from their fathers tend to demonstrate internalizing problems (which was signifi cant for social withdrawal and with a strong correlational trend to anxiety/depression), confi rming the results found by Brumariu and Kerns (2010), Desjardins and Leadbeater (2011) and Roelofs et al. (2006).

Parental Practices and Behavior Problems
In the present study, as it has been reported in the literature (Nishikawa, Sundbom, et al., 2010;Roelofs et al., 2006;Rohner & Britner, 2002), parental rejection was an important correlate of externalizing problems (aggression and delinquency).It was also found to be associated with anxiety/depression in children, but not with social withdrawal, confi rming in part the results of other studies (Nishikawa, Sundbom, et al., 2010;Roelofs et al., 2006;Rohner & Britner, 2002).This may be due to the fact that the studies that found an association between negative parenting practices (like rejection) and internalizing problems used a global score of internalizing problems (Nishikawa, Sundbom, et al., 2010), and not separate scores for each marginally signifi cant).These variables remain as significant predictors, even when controlling sociodemographic variables (specifically, the child's age, maternal and paternal ages and educational levels).Moreover, being a boy offered an additional contribution to the model which predicts aggressiveness; a result consistent with other studies (Burt, Mikolajewski, & Larson, 2009;Prinzie, Onghena, & Hellinckx, 2006).
With respect to predictive analysis of internalizing problems, only the child´s attachment towards his/her father was able to predict social withdrawal, not diminishing its predictive power in the presence of sociodemographic variables.Thus, the perceived poor attachment to the father, along with low paternal education level explained the increase levels of social withdrawal in childhood and preadolescence.It is possible that in this particular age (late childhood and early teens) fathers assume a specifi c role and relevance in the consolidation of child personality (Lieberman, Doyle, & Markiewicz, 1999;Steele & Steele, 2005).
In the prediction of anxiety/depression, parental rejection showed predictive contribution, but with a marginally signifi cant effect.Therefore, unlike the fi ndings of Doyle and Markiewicz (2005), the hypothesis that both groups of variables (attachment and parenting practices) could provide independent contributions in predicting internalizing problems was not confi rmed in this present study.This indicates the need of more specifi c studies examining cultural values that may explain these variations.Individual characteristics of parents, such as age and educational level, seemed also to have effects on the child´s psychological well-being.However, it is likely that the age of the parents does not exert a direct impact on the emotional problems of the children, but rather, indirect effect, perhaps mediated by their beliefs and values as well as their perception of social networks support in their educational practices (Marin & Levandowski, 2008).Finally, the decline of paternal education level, as well as in social withdrawal, seems to be also associated with increased levels of anxiety and depression in children.This result is consistent with studies showing that low parental education is associated with adverse developmental outcomes of their children, since they affect the way as parents exercise their parenting (Klebanov et al., 1994).Furthermore, the suggestion that parental rejection was the best predictor and correlate of anxiety depression fi nds strong support in the revision work of Rohner and Britner (2002).
Finally, it seems appropriate a comment the contribution of variables exclusively related to father in the prediction of internalizing problems and, on the other hand, the variables exclusively related to mothers in the prediction of externalizing problems.In this sense, researchers have emphasized the protective function that a paternal attachment appears to have against internalizing problems in children from middle childhood (Brumariu & Kerns, 2010;Desjardins & Leadbeater, 2011;Roelofs et al., 2006).Regarding the aggressive behaviors associated with maternal variables, they can be revealing behaviors that Rohner (2002) called "defensive independence" and that Dozier, Stovall, and Albus (1999) called strategy of "minimizing the need for bonding".That means that these fi ndings indicate a need of more studies focused in the specifi cities and the impact of paternal and maternal practices on children´s developmental behavioral trajectory.

Final Considerations
In general, the results obtained with children in Southern of Brazil reinforce the international literature on the relationship between parenting and behavior problems.Specifi cally, the results confi rmed the relevance of parental rejection, behavioral control and poor maternal attachment in the occurrence of externalizing problems.
The fi ndings related to internalizing problems are less clear and are not in complete agreement with the literature.Such problems often are unnoticed by parents, since they include internal emotional states with diffi culty expression or not clearly perceived.Furthermore, the role of psychological control, associated with internalizing problems in studies another countries, have not showed the same infl uence among the Brazilian children in the present sample.The reasons for this difference in results may be due to cultural differences, in Brazil like other Latin cultures, where the acceptance of parental practices which inadvertently employ strategies involving a emotional manipulation of the children (induction of guilt, threats of punishment) are seen as socially acceptable ways to discipline them.This argument, however, needs further investigation.It is recommended that futures studies examine in a more comprehensive way the impact of social and cultural specifi cities associated with the role of psychological control in our culture using different and validated instruments.Furthermore, it is important to notice that the values of the regression coeffi cients in the adjustment of the proposed models were quite low, indicating that the explanatory variables investigated contributed modestly to explain the variability of the internalizing and externalizing behavior problems.Thus, new variables should be included in future predictive studies, to give explicit account of the multifaceted nature of these behavior problems.
Beside the contributions of this study, there are also some limitations.First, the measures on parenting practices were applied to any caregivers (mother or father or any other person responsible for raising the child), making it impossible to carry out comparative analyses between maternal and paternal practices.This study refl ects the general vision of the mothers, who were the majority of caregivers (88.3%).Furthermore, parenting practices were assessed from the parents' perspectives, and would be important to obtain the children perceptions about these practices as well.Assuming the possibility of fi nding divergent views, information obtained from both parents and from children could bring more accurate information to understanding the development of internalizing and externalizing problems.
In studying development, it is also necessary to conduct research including longitudinal data that can be helpful to understanding of the trajectory and consequences of internalizing and externalizing problems in childhood.Finally, in future studies, it is suggested to include information on the quality of other social relationships that children establish (eg, friendship) in the peer group, a sphere of social relationships that, at this stage of development, plays a fundamental role in children´s socialization processes.