Introduction

Anxiety disorders are among the most prevalent mental disorders in childhood and adolescence (Albano et al. 2003). These disorders have negative implications for different areas of children’s functioning and, if left untreated, can have a profound impact on the emotional and social development of young people (Gosch et al. 2006). Longitudinal studies have also shown that childhood anxiety disorders often precede anxiety disorders in adulthood and may play a causal role in the development of comorbid conditions, such as depression (Cole et al. 1998; Woodward and Fergusson 2001).

Anxiety disorders tend to run in families. Children of parents with anxiety disorders are more likely to develop anxiety problems than are children of parents without clinical levels of anxiety (Beidel and Turner 1997). In addition, studies have also demonstrated that parents—particularly, mothers—of children diagnosed with anxiety disorders suffer from a higher prevalence of anxiety disorders compared to parents of non-anxious children (Cooper et al. 2006).

The transmission of anxiety from parents to children can be explained by a variety of factors, including biological vulnerability, exposure to adverse life events and chronic stress, and different types of learning processes (McLeod et al. 2007). A number of etiological models of childhood anxiety (Chorpita and Barlow 1998; Creswell et al. 2010; Hudson and Rapee 2004) assume that parental behavior and parenting styles play a key role in this process, and such models are supported by growing empirical evidence (McLeod et al. 2007; Rapee 1997; van der Bruggen et al. 2008; Wood et al. 2003).

The majority of studies on the relationship between parenting and childhood anxiety have focused on the role of parenting styles. Two main broad dimensions are systematically examined in these studies: emotional warmth and control. Empirical findings have generally indicated that parenting styles characterized by low warmth and high control are associated with higher levels of childhood anxiety problems (e.g., Siqueland et al. 1996). However, these links are not particularly unique because research has demonstrated that these negative parenting styles are also related to other psychopathological problems (Pereira 2009; Wood et al. 2003).

Research has also indicated that childhood anxiety is more clearly associated with a high-control style of parenting than with low parental warmth, although the magnitude of association between parental control and anxiety has been found to be highly variable (McLeod et al. 2007). One proposed explanation for this variability concerns the diversity of the operationalization of the parental control construct, which involves various elements: psychological control (Caron et al. 2006), negative control (Booth et al. 1994), involvement (Hudson and Rapee 2001), intrusive behavior (Adam et al. 2004), and restriction (Krohne and Hock 1991). Among these different forms of parental control, etiological models of childhood anxiety place the greatest emphasis on the aspect of overprotection. Briefly, high levels of overprotection are posited to increase a child’s perception of danger (Hudson and Rapee 2001), reduce perceived control over threats (Chorpita et al. 1998; Rapee 1997), and restrict a child from acting on opportunities to explore his or her environment and to develop coping skills to address threat and unexpected adversity (Barlow 2002).

Additionally, support has been found for another parenting dimension, referred to as “anxious rearing” (Van Brakel et al. 2006; Muris et al. 2003). Given the high co-existence of anxiety disorders in parents and children, it is likely that anxiety and specific fears in children and adolescents may also be learned through modeling and transmission of negative information (Rachman 1977). The anxious rearing style is used to label parents who describe situations to their children as dangerous and irresolvable, encourage or reinforce catastrophic interpretations, overtly express fear or anxiety, and model avoidant coping behavior (Wood et al. 2003).

Although several studies have explored the links between parental rearing styles and childhood anxiety, little is yet known about the processes through which these parental factors increase children’s anxiety levels. The triple-vulnerability model of anxiety development (Barlow 2000, 2002) posits the influence of two psychological vulnerabilities for anxiety: (a) low perceived control as a generalized psychological vulnerability, which is characterized by a sense of lack of control over stimuli, situations, and anxious feelings, and (b) stimulus and situation-related threat cognitions as a specific psychological vulnerability.

Existing research supports the association between perceived control and anxiety in youths. For example, anxious children display lower levels of perceived control over external and internal anxiety-related events compared to non-anxious children (Pereira et al. 2012a, b; Weems et al. 2003). Research on the development of children’s anxiety control beliefs points to the influence of parents. Chorpita and Barlow (1998) theorized that among the aforementioned set of family characteristics, warmth and control are important for the development of control-related cognitions in children. A child whose parents lack warmth and are high on control, thereby limiting the child’s autonomy, may be particularly prone to develop a diminished sense of control over his or her environment.

Only a few empirical studies have actually tested Chorpita and Barlow’s (1998) model, and these studies have made use of different measures of perceived control. An initial investigation by Chorpita et al. (1998) revealed that the external locus of control mediated the relationship between parental control and negative affect in a mixed sample of clinically referred and non-clinically referred youths. Two further studies with children explored the mediating role of perceived control specific to threatening situations, but this research yielded rather inconsistent results. A study by Becker et al. (2010) relied on the reports of clinically anxious and non-anxious mothers of youths aged 6–14; their findings supported the existence of a mediating role for perceived control over anxiety-related situations on the effect of maternal anxiety and controlling rearing behavior on childhood anxiety. In contrast, a study by Muris, Meesters, Schouten, and Hoge (2004) found no evidence of a mediating effect of perceived control over anxiety in the associations between parental control and rejection and children’s anxiety in a non-clinical sample of children and adolescents aged 11–14 years. These divergent findings can be explained by differences in study methodologies—in particular, the sample characteristics, such as children’s age and clinical status. Studies that have found support for a mediational model (Chorpita et al. 1998; Becker et al. 2010) have been conducted with younger children. Note that Chorpita and Barlow (1998) have argued that the mediational role of perceived control is especially relevant for younger children because their cognition is still in the process of formation. Additionally, the studies that have found a significant mediation effect of control beliefs have been conducted in mixed samples, including both clinical and nonclinical children or parents, which may have facilitated the detection of the hypothesized effect because of increased variability in the measurement of the pertinent constructs.

With respect to specific psychological vulnerability (stimulus and situation-related threat cognitions), a fair amount of research has consistently shown that interpretative biases are positively correlated with anxiety in children (e.g., Barrett et al. 1996; Chorpita et al. 1996; Weems et al. 2007). However, research on the influence of parents on children’s interpretative biases is sparse, although it is generally assumed that specific parental behaviors may promote or reinforce children’s cognitive experiences of anxiety in particular situations (Craske 1999). In an observational study by Chorpita et al. (1996), clinical and nonclinical children were asked to generate interpretations in response to hypothetical, ambiguous scenarios. Parents completed the same procedure, after which the family members discussed their responses together to reach a final interpretation. During the family discussions, parents’ modeling, prompting, and rewarding of anxiety were assessed. The main finding of this study was that anxious expressions exhibited by the parents during the family discussions prompted children to interpret the ambiguous scenarios in a more negative manner, although these results failed to achieve statistical significance, most likely because of the study’s small sample size. The results of a more recent study also support the mediating role of children’s interpretational biases in the relationship between parental variables (anxiety and control) and children’s anxiety. From these findings, it can be concluded that parental anxiety and control contribute to children’s anxiety through their influence on children’s interpretative biases (Affrunti and Ginsburg 2012).

One important shortcoming of the research conducted to date on the role of parental influences in the pathogenesis of childhood anxiety is the excessive focus on maternal behaviors. The role of fathers has been largely neglected, although there are good reasons to believe that fathers play a unique role in the development of anxiety problems in youths. For example, Bögels and Phares (2008) have posited that the relationship between parenting and children’s anxiety is moderated by parental gender and the developmental stages of children. For example, during middle childhood, a close interpersonal relationship with the mother and the encouragement of independence by the father seem to be important protective factors, which prevent the child from developing excessive anxiety. Studies that simultaneously examine the roles played by both paternal and maternal parenting on children’s anxiety are rare; to date, little evidence for a differential role of fathers and mothers has been found (see Verhoeven et al. 2012).

The present study addresses this gap in the literature by examining the relationships between relevant parental dimensions (trait anxiety, overprotection, and emotional support) and anxiety symptoms in children, considering independent influences of mothers and fathers. Furthermore, extending previous work, anxiety control beliefs and interpretative biases (negative cognitive errors and threat interpretations over ambiguous situations) were assessed, which provided an opportunity to explore their unique role in mediating the relationship between parental variables and childhood anxiety. Consistent with Chorpita and Barlow’s model (1998) and previous research (Becker et al. 2010; Chorpita et al. 1998), we hypothesized that children’s control beliefs would mediate the relationship between parental control behavior and children’s anxiety symptoms. In addition, we expected that interpretative biases would act as a mediator in the relationship between parental anxiety and control and children’s anxiety symptoms (Affrunti and Ginsburg 2012).

Method

Participants

Children were selected from an initial sample of 905 children who participated in a universal screening of anxiety symptoms carried out in 15 public, private, and cooperative primary schools situated in semi-urban and urban areas in the vicinity of Lisbon, Portugal. To ensure sufficient representation of anxious children in the sample and to form an equivalent normative group of non-anxious children, two groups were selected according to their scores on a standardized anxiety questionnaire (see below): children who scored above the 80th percentile (n = 41) and children who scored below the 50th percentile (n = 39). Sampling was conducted to ensure that the groups did not differ in terms of age and had a balanced gender distribution.

Thus, the total sample consisted of 80 children (33 boys and 47 girls) and their mothers and fathers. The children were aged between 7 and 12 years with a mean age of 9.84 years (SD = 1.23). Most children came from intact families, and parents had completed 12 years of education on average. Different socio-economic levels were represented in the sample (36.4 % low SES, 33.8 % medium SES, and 29.9 % medium–high and high SES).

Instruments

Child Anxiety

The Screen for Child Anxiety Related Emotional Disorders-Revised version (SCARED-R; Muris et al. 1999) is a self-report questionnaire with 69 items for assessing symptoms of the following anxiety disorders in children: separation anxiety disorder, generalized anxiety disorder, panic disorder, social phobia, school phobia, specific phobia, obsessive–compulsive disorder, and acute or posttraumatic stress disorder. Children rate how frequently they experience each symptom (e.g., “I worry about going to school”) on a 3-point scale: 0 (never or almost never), 1 (sometimes), or 2 (often). Scores for each dimension of anxiety and an overall score are obtained. The Portuguese version of the SCARED-R (Pereira et al. 2011) shows high levels of internal consistency and good test–retest reliability. In the present sample, the child and mother versions of the SCARED-R were employed, and both showed high internal consistency (α = 0.96 for children and α = 0.93 for mothers).

Child Cognitive Variables

Cognitive errors. The Children’s Negative Cognitive Errors Questionnaire (CNCEQ; Leitenberg et al. 1986 ) is a self-report questionnaire containing 24 items that measure four types of cognitive errors: catastrophizing, overgeneralization, personalizing, and selective abstraction. Each item consists of a hypothetical vignette followed by a negative interpretation that represents one of the four cognitive errors that are being evaluated. Children were asked to what degree the interpretation provided matched their interpretation of the hypothetical situation (e.g., You forgot your homework. Your teacher tells the class to turn in their homework. You think: “The teacher is going to think that I don’t care, and I won’t pass”) using a 5-point scale ranging from 1 (“nothing like I would think”) to 5 (“exactly what I would think”). Scores for each of the cognitive errors and a total cognitive error score were obtained. The Portuguese version of the questionnaire shows a high degree of internal consistency for the overall score (Pereira et al. 2012b). This study used only three of the scales (catastrophizing, overgeneralization, and personalizing) because, in a previous study, these cognitive errors were found to be most relevant for childhood anxiety (Pereira et al. 2012b). The Cronbach’s α of the total score for these three scales in the present sample was 0.93.

Threat Interpretations. The Ambiguous Situation Questionnaire for Children (ASQ-C; Pereira et al. 2010) is a self-report questionnaire that aims to assess children’s interpretations and action plans in response to ambiguous situations. This measure was largely based on the ambiguous situations interview employed by Barrett et al. (1996), to which the following modifications were made: (a) efforts were made to increase the ambiguity of some of the original vignettes; (b) new vignettes were added to cover common anxiety topics for children of this age (e.g., separation anxiety); and (c) considering the fact that children may simultaneously offer different interpretations for the same situation, children were not forced to choose one of the possible interpretations but rather were asked to score to what extent various interpretations applied to them. The final ASQ-C consisted of six ambiguous scenarios, two of which covered the domains of separation anxiety, social phobia, and specific phobia. For each scenario (e.g., “You are at school waiting for your mother. Your mother agreed to pick you up at 5:00 pm, but it is 5:15 pm, and she’s not there”), children were prompted for the following: (1) to identify their level of concern on a scale ranging from 1 (“not worried”) to 5 (“extremely concerned”); (2) to categorize three different thoughts presented (two representing a threat interpretation, e.g., “She is late because she may have had an accident,” and one representing a non-threat interpretation, e.g., “She is delayed because there is a lot of traffic”) in terms of similarity to their own thoughts, again employing a 5-point scale (1—exactly the same, 2—very similar, 3—more or less similar, 4—somewhat similar and 5—nothing similar); and finally, (3) to report what they would do to resolve the situation. In the present study, a total score for threat interpretations was obtained, which showed high internal consistency (α = 0.87).

Anxiety control beliefs. The short form of the Anxiety Control Questionnaire for Children (ACQ-C; Weems 2005) is a self-report questionnaire with 10 items for assessing control beliefs over external and internal anxiety-related events. The children were asked to rate their agreement with each item on a 5-point scale from 0 (none) to 4 (very much). An overall score for control beliefs can be obtained by summing the ratings for all items (e.g., “I can usually calm myself down when I want to”). The Portuguese version of the questionnaire (Pereira et al. 2012a) has shown high degrees of internal consistency for the overall score, and this was true in the present sample (α = 0.83).

Parental Variables

Parental anxiety. The State Trait Anxiety Inventory (STAI; Spielberger et al. 1970) is a widely used measure of state and trait anxiety. In the present study, we only used the trait scale, which is composed of 20 items (e.g., “I lack self-confidence”) that have to be rated on a 4-point Likert scale. The STAI has demonstrated good internal consistency, good test–retest reliability, and convergent and discriminant validity (e.g., Barnes et al. 2002; Spielberger et al. 1980). The scale had high internal consistency in the current sample (α = 0.92 for mothers and α = 0.87 for fathers).

Parental overprotection and concern. The Anxiety and Overprotection Scale (EASP; Pereira et al. 2013) is a self-report scale with 20 items for evaluating parental anxiety, concern, and overprotection. In this study, this scale was adopted in an attempt to fill the need for a specific instrument capable of measuring various components of parental control and parental overprotection. The EASP contains three scales: (1) parental anxiety and concerns (10 items, e.g., “I only feel calm when my child is near me”); (2) parental overprotection (7 items, e.g., “To stop my child from making mistakes, I sometimes do things for him”); and (3) support of children’s positive coping behaviors (3 items, e.g., “I tell my son that the best way to address his fears is facing them”). Parents are asked to rate to what extent each statement corresponds with how they commonly think, feel, and behave toward their child, using a scale from 0 (none) to 4 (very much). The scale has good psychometric qualities (Pereira et al. 2013). In the present study, we used a sum of the parental overprotection and parental anxiety and concerns scales, which exhibited good internal consistency (anxiety and concerns scale: α = 0.93 for mothers and α = 0.90 for fathers; overprotection scale: α = 0.90 for mothers and α = 0.81 for fathers).

Parental emotional warmth. The Egna Minnen Betraffende Uppfostran for Children (EMBU-C; Castro et al. 1993) is a scale originally developed in Spanish for the purpose of measuring children’s perceptions of parental rearing behaviors in three dimensions: Emotional Warmth (14 items; e.g., “Your parents tell you they like you and hug you or kiss you”), Rejection (8 items; e.g., “Your parents say they do not like the way you behave at home”) and Control (10 items; e.g., “Your parents tell how you should dress and cut your hair”). The Portuguese version of EMBU-C contains 32 items, which have to be answered on a 4-point Likert scale (1 = no, never; 2 = yes, but seldom; 3 = yes, often; 4 = yes, most of the time). For each item, children first rated their father’s and then their mother’s behavior. In the present study, only the emotional support scale of the Portuguese version of EMBU-C (Canavarro and Pereira 2007) was used. Using the present data, the internal consistencies for the Emotional Warmth scale (α = 0.84 for mother and α = 0.87 for father) were good.

Procedure

This study was approved by the General Administration for Innovation and Curriculum Development in Lisbon, Portugal, and organized in collaboration with the school boards and teachers of the schools involved. Children were recruited from 15 Portuguese public and private schools. Our objective was to select a sample that varied in terms of anxiety levels (by including highly anxious and non-anxious children). The recruitment was conducted in two phases. After receiving an informed consent form, 83 % of the parents allowed their children to participate in the first phase of the study, a universal screening procedure at school during which children’s anxiety symptoms were assessed twice over a one-month interval. All of the children who scored above the 80th percentile on both assessments were selected and included in the highly anxious group. To form a normative group, some of the children scoring below the 50th percentile were also selected, depending on their age, sex, and grade (to create equivalent groups). All of the parents were then contacted and asked whether they granted permission for their child to participate in the second phase of the study. A clear majority of parents (i.e., 87 %) again responded positively to this request and signed informed consent forms permitting their child to participate in the second phase of the study.

The remainder of the instruments (the CNCEQ, the ASQ-C, and the ACQ-C) were later administered collectively, again during regular classes, in the presence of the teacher and one or two members of the research team. The administration of the questionnaires lasted for a maximum of 45 min. The questionnaires completed by the parents (the EASP and the STAI) were sent home and returned to the researcher by the teacher.

Statistical Analysis

First, we computed correlations among various measures to explore how variables were associated with each other. In the case of child anxiety, one total score was computed. The literature on discrepancies among informants on child psychopathology has generally revealed that there is a fairly low agreement between ratings provided by parents and children (for a review, see De Los Reyes and Kazdin 2004), and this agreement is also observed for childhood anxiety (Pereira et al. 2011; Weitkamp et al. 2010). Following the procedure adopted in other studies (e.g., Becker et al. 2010), total scores of children’s anxiety in this study were obtained by standardizing and summing the scores of the parent’s and children’s versions of the questionnaires. These total scores were used as indicators of child anxiety symptoms in all subsequent analyses.

Two multiple regression analyses were conducted to examine the contributions of maternal and paternal variables to anxiety symptomatology, as perceived by children and mothers. We ran the models separately for mothers and fathers. These analyses were performed hierarchically in two steps. In the first step, we entered the demographic variables (the child’s gender and age), and in the second step, we added the parental variables (trait anxiety, emotional support and overprotection). All variables were forced to enter.

Subsequently, we conducted a joint analysis to explore the relative contributions of the maternal and paternal variables and the child’s cognitive vulnerabilities to anxiety symptoms as perceived by children and mothers. These analyses were performed hierarchically in four steps. In the first step, we entered the demographic variables (child gender and age). In the second step, we added the parental variables that had shown significant effects in the previous analysis. Next, we included children’s anxiety control beliefs, and finally, we included the children’s interpretive biases (negative cognitive errors and threat interpretations). Children’s cognitive variables were entered in different steps because we wanted to explore their independent roles in the mediation of the parental variables.

Finally, we conducted a mediational analysis to examine whether children’s cognitive variables mediated the relationship between parental factors and child anxiety, following the method described by Baron and Kenny (1986). The Sobel test was used to determine if the indirect effect was statically significant.

Results

Relationships Among Measures

Table 1 displays mean scores and standard deviations for various constructs, as well as the intercorrelations among these variables. The correlational analyses revealed a moderate but significant positive association between trait anxiety and overprotection/concern for the mothers but not for the fathers. In addition, no significant association between trait anxiety and emotional support was found, and this lack of association appeared to hold for both parents. Furthermore, it was found that children’s cognitive errors were significantly associated with mothers’ levels of trait anxiety and overprotective parenting. Children’s threat interpretations were also moderately associated with mothers’ trait anxiety, whereas children’s control perceptions were not related to any of the mothers’ variables. For the fathers, only the correlation between emotional support and children’s cognitive errors reached statistical significance.

Table 1 Mean scores and standard deviations for various questionnaires included in the study, as well as correlations among these variables

Most importantly, the analyses also demonstrated significant positive associations between mothers’ trait anxiety, mothers’ overprotection and concern, and fathers’ overprotection and concern on the one hand and children’s anxiety symptoms on the other hand. Finally, all cognitive variables were also significantly related to children’s anxiety. As expected, children’s interpretative biases were positively linked to children’s symptoms, whereas children’s anxiety control beliefs revealed a negative association with such symptoms.

Regression Analysis: Maternal and Paternal Effects on Children’s Anxiety

Two multiple regression analyses were conducted (Table 2). First, a regression model was conducted to explore the independent contributions of paternal variables to child’s anxiety symptomatology. Paternal overprotection and concern had a unique and significant effect on children’s anxiety symptoms, with the model explaining 14 % of the total variance. Second, a hierarchical multiple regression was performed to examine the independent contributions of maternal variables to children’s anxiety symptoms. In this regression, a significant effect was observed for trait anxiety, with maternal trait anxiety being positively related to children’s anxiety. Variables included in the final model explained 22 % of the total variance.

Table 2 Results of the hierarchical multiple regression analyses: models predicting children’s anxiety from mothers’ variables and fathers’ variables

Fathers’ and Mothers’ Conjoint Effects and the Mediating Role of Children’s Cognitive Variables

Finally, we conducted a conjoint analysis to explore mothers’ and fathers’ relative contributions to children’s anxiety and to explore the mediating role of children’s cognitive variables (control beliefs and interpretative biases). The results (Table 3) show that mothers’ trait anxiety and fathers’ overprotection and concern have significant effects on children’s anxiety. The results also show that all of the child’s cognitive variables were significant predictors in the model, supporting the incremental validity of each one. Finally, when children’s interpretative biases variables were entered in the model, mothers’ trait anxiety no longer reached significance, suggesting a mediational effect of these interpretative biases on the relationship between maternal trait anxiety and children’s anxiety. Variables included in the final model explained a substantial amount of the total variance, r 2 = 0.59.

Table 3 Results of the hierarchical multiple regression analyses: model predicting children’s anxiety from parental variables and cognitive vulnerabilities

An additional analysis was conducted to explore whether interpretative biases mediated the relation between mothers’ trait anxiety and children’s anxiety. As Fig. 1 shows, the total effect of mothers’ trait anxiety on children’s anxiety was significant (c; t = 3.75, p < 0.001). There was also a significant effect of mothers’ trait anxiety on children’s cognitive errors (1a; t = 3.66, p < 0.001), as well as on children’s threat interpretations (2a; t = 2.85, p = 0.006). Finally, there was a significant effect of children’s cognitive errors on children’s anxiety when controlling for the effects of mothers’ trait anxiety and children’s threat interpretations (1b; t = 2.65, p = 0.01) and a significant effect of children’s threat interpretations on children’s anxiety after controlling for the effects of mothers’ trait anxiety and children’s cognitive errors (2b; t = 3.10, p = 0.003). This effect resulted in a significant indirect effect of children’s cognitive errors (z = 2.14, p = 0.032) and a significant indirect effect of children’s threat interpretation (z = 2.09, p = 0.037). The direct effect was no longer significant (c′; t = 1.81, p = 0.074), showing that the effect of mothers’ trait anxiety on children’s anxiety was fully mediated by children’s interpretative biases.

Fig. 1
figure 1

Model of the mediating role of interpretative bias in the relationship between mothers’ trait anxiety and children’s anxiety: standardized betas. c The total effect of the independent variable on the dependent variable, b the effect of the mediator (mediator 1 child cognitive factors, mediator 2 child threat interpretation) on the dependent variable while controlling for the independent variable and the other mediator, and c’ the direct effect of the independent variable on the dependent variable while controlling for the two mediators

Discussion

Over the past few years, a growing body of empirical studies has revealed a significant relationship between parenting and children’s anxiety symptoms (Affrunti and Ginsburg 2012; McLeod et al. 2007; Wood et al. 2003). However, this literature has largely ignored the processes responsible for this relationship and has almost exclusively focused on the mother’s role. Most theoretical models of the influence of parents on children’s anxiety do not take into account that mothers and fathers may affect children’s anxiety in different ways. This study examined the relation between mothers’ and fathers’ trait anxiety, parenting styles (overprotection and emotional support), and children’s anxiety symptomatology, while exploring the role of cognitive vulnerabilities in mediating these relationships.

The present results show that different maternal and paternal variables were significantly associated with children’s anxiety. Mothers’ trait anxiety was the only variable with a significant effect on children’s anxiety after controlling for the other parental factors. In fathers, only overprotection and concern were significantly associated with children’s anxiety when controlling for the effects of fathers’ trait anxiety and emotional support.

The findings only provide partial support for the hypotheses regarding the role of children’s cognitive variables in mediating the relationship between parental variables and children’s anxiety. Namely, the relationship between mothers’ trait anxiety and children’s anxiety was fully mediated by children’s interpretative biases. There are several hypotheses to explain the influence of mothers’ trait anxiety on children’s anxiety mediated through children’s interpretative biases. Mothers’ anxious rearing styles, although not assessed directly in this study, can be useful in understanding this process. Compared to fathers, mothers are more involved in children’s education (Paulson and Sputa 1996), spend more time with them (Lamb 2000), and are more emotionally expressive than fathers (Halberstadt et al. 1995); for these reasons, mothers have more opportunities than fathers to model certain cognitions and behaviors. Therefore, anxious mothers express more anxious cognitions and behaviors, which will influence the ways children interpret and think about their environment and adversities. This occurs through the influence of interpretative biases, cognitive errors, and threat interpretations, and not through anxiety control beliefs. One previous study (Becker et al. 2010) identified some mediation through control beliefs, but this study was conducted on clinically anxious mothers.

In contrast, we hypothesized a non-cognitive, behavioral path for fathers’ influences on children’s anxiety. Higher levels of fathers’ controlling behaviors in clinical anxious children compared to those in a normal control group were also found in other studies (Barrett et al. 2005; Bögels et al. 2008; Greco and Morris 2002), although research that has also included an evaluation of maternal overprotection has documented a difference of similar magnitude in mothers. The findings of the current study suggest that fathers of more anxious children offer only limited encouragement toward children’s independence and autonomy. Overprotection can prevent children from having opportunities to explore their environment and to develop coping skills in difficult situations. Compared to mothers, fathers may have a more important role in children’s exposure to risky situations, which is consistent with the model described by Bögels and Phares (2008).

In contrast with previous findings indicating that during middle childhood, mothers’ emotional support is a particularly relevant for children’s anxiety (e.g., Siqueland et al. 1996), the results of this study show that parents’ emotional support do not have a significant impact on anxiety. These results are well in line with research showing that controlling styles of parenting are more clearly connected to childhood anxiety than a supportive and warm parenting style (McLeod et al. 2007).

The results also show that all three cognitive vulnerability measures offer incremental explanatory power in predicting children’s anxiety. Previous research has yielded similar findings: specifically, that anxiety control beliefs and negative cognitive errors are incrementally valid in predicting anxiety symptoms in youth (Weems et al. 2007) and in predicting anxiety disorder diagnostic status (Cannon and Weems 2010). This study adds to such evidence by showing that each of these cognitive measures (negative cognitive errors and threat interpretations of ambiguous situations) incrementally predicts anxiety symptoms in a community sample of children. Taken together, these findings are interesting, suggesting that each of these constructs represents an independent cognitive process that needs to be addressed in preventive and therapeutic interventions for children.

Limitations

There are several limitations that should be considered when interpreting the results of the current study. First, this is a cross-sectional study, and thus, caution is needed when interpreting the causality of the observed associations. The results show that maternal trait anxiety and paternal overprotection and concern have significant independent effects on children’s anxiety and that children’s interpretative biases mediate the relationship between mothers’ trait anxiety and children’s anxiety symptoms. However, it is logical to expect a cyclical relationship between parenting and child anxiety, especially when considering paternal effects. Fathers’ overprotection can lead to children’s anxiety, but children’s anxiety can also elicit paternal control. However, in the case of maternal effects, it is more difficult to posit a parallel, reverse effect because mothers’ trait anxiety is supposed to be a more stable psychological characteristic less affected by environmental factors.

Another limitation concerns the study’s exclusive reliance on a self-report methodology to evaluate parenting variables and children’s anxiety. The inclusion of other methodologies (e.g., observations) could have enriched the study, although some of the variables assessed are of an internalized nature and would be difficult to capture from the perspective of an external observer. Additionally, as mentioned before, our sample of participants was composed of families in which both parents voluntarily participated in the study. Therefore, the results may not generalize to other types of families. The inclusion of clinically referred families—for example, families with serious problems of neglect or violence—could produce different results. In these types of families, parental support and rejection may play a more central role in the etiology of anxiety and emotional problems.

Implications for Research and Practice

This study has several strengths, including its simultaneous consideration of mothers’ and fathers’ anxiety and parenting, its focus on the mediational processes that explain these parental influences, its inclusion of a more specific component of parental control, and its inclusion of several informants. The findings suggest that models concerning parental influences on children’s anxiety should consider different roles and pathways of influence for fathers and mothers.

An improved understanding of the developmental pathways of childhood anxiety can contribute to prevention and intervention efforts to reduce these problems in childhood. The results of the current study should be considered when working clinically with such families. Several empirical studies (Podell and Kendall 2011) emphasize the importance of including fathers to improve the efficacy of interventions with anxious disordered youth. This study adds to this evidence, suggesting ways that mothers and fathers could be involved. Intervention should aim to reduce the modeling of maternal anxious behaviors, which would reduce the exposure of children to the overt manifestation of mothers’ anxious cognitions and behaviors. This reduction can be achieved by helping mothers deal with their own anxiety through cognitive and behavioral strategies. Conversely, fathers could be given more salient roles in the exposure component, such as helping children gradually confront anxiety-provoking situations and encouraging independence.

This study also has implications for the theoretical models used to explain the development and maintenance of childhood anxiety. Particularly, more studies are needed to replicate the current findings and to further explore the different paths of influence for mothers and fathers. As noted earlier, one hypothesis is that maternal influences are more cognitive in nature, whereas paternal impact on children’s anxiety is more behavioral. This possibility can be related to differences between mothers and fathers in emotion regulation and the behavioral strategies they employ to address children’s anxiety and could be investigated in future studies relying on clinical samples.