Exploring Parental Predictors of Child Anxiety: The Mediating Role of Child Interpretation Bias
- 1.1k Downloads
Separate lines of research have shown that higher levels of parental overcontrol and parental anxiety are related to higher levels of child anxiety. The mechanisms of transmission, however, are poorly understood.
It has been theorized, though not empirically tested, that parental overcontrol and anxiety increase children’s interpretation bias by signaling to the child that the environment is threatening (e.g. through modeling or restriction of autonomy), thus increasing the child’s anxiety level.
The present study investigated this theory using 75 parent–child dyads (parents aged 27–52, 82 % female; children aged 7–12, 52 % female, 80 % Caucasian). All parents were diagnosed with a primary anxiety disorder, while no child was diagnosed with an anxiety disorder.
Children’s interpretation bias, measured using ambiguous stories, was shown to partially mediate the relation between parental overcontrol and child anxiety and completely mediate the relation between parental anxiety and child anxiety. There was no significant relation between parental overcontrol and parental anxiety.
Findings partially support theoretical models, which posit that higher levels of overcontrol and parental anxiety signal to children that their environment is threatening, perhaps increasing their threat appraisal of ambiguous situations and increasing their anxiety. Implications and directions for future research are discussed.
KeywordsChild Parent Cognitions Parenting Overcontrol Anxiety
There is strong empirical evidence that a cognitive bias toward threatening stimuli is associated with higher levels of anxiety in children (Hadwin et al. 2006). While interpretation bias has been measured in different ways, findings using ambiguous stories have consistently shown that anxious children show an increased tendency to interpret ambiguous situations as threatening (Barrett et al. 1996; Chorpita et al. 1996; Creswell and O’Connor 2006; Dineen and Hadwin 2004; Muris et al. 2003a). Furthermore, children with higher levels of anxiety are faster and require less information to conclude that ambiguous stories will be threatening (Muris and van Doorn 2003; Muris et al. 2000). Bögels and Zigterman (2000) reported that anxious children (ages 9–18; n = 15) interpreted ambiguous stories as more threatening than clinical controls (ages 9–17; n = 15; children diagnosed with oppositional defiant disorder, attention deficit-hyperactivity disorder, or conduct disorder). In a 1-year longitudinal study of a community sample (n = 65), Creswell and O’Connor (2011) found that children’s (10–11 years of age) anticipated distress response to ambiguous stories predicted an increase in anxiety symptoms over time. A similar finding was reported by Warren et al. (2000). Thus, it appears that interpretation bias may play a significant role in the development of anxiety disorders in children (Creswell et al. 2011). As such, the development of anxious interpretative biases in children, and their role in developmental models of anxiety, becomes a vital focus of research.
While genetic influences, like personality characteristics, may play a role in the development of interpretative biases (Eley and Zavos 2010), and in turn anxiety, there is a growing body of literature examining environmental, explicitly parental, factors. Specifically, there is evidence that parental anxiety and parental interpretation bias are subsequently related to interpretation bias in children (Creswell et al. 2011; Field and Lester 2010). Lester et al. (2009) reported that parents showed a similar interpretation bias for ambiguous situations involving themselves and their child (as measured by ambiguous stories pertaining to their own environment and that of their children). Furthermore, anxious, compared to nonanxious, parents demonstrated higher interpretation bias in ambiguous situations for both themselves and their children (Lester et al. 2009). As such, anxious parents may view their own and their child’s environment as more threatening than nonanxious parents and may communicate that threat to their child. In support of this hypothesis, Creswell et al. (2005) reported that parent and child responses to ambiguous situation questionnaires were highly correlated in a clinical sample of 27 parent–child dyads. Analogous results were found using a non-clinical sample (Creswell and O’Connor 2006).
Despite emerging evidence of the association between parental anxiety, parental interpretation bias and child interpretation bias, less is known about the influence of parental behaviors and child interpretation biases. One such behavior, parental overcontrol, defined as an excessive amount of involvement in a child’s daily routine and activities, or emotional experiences, and an encouragement of dependence on the parents (Barber 1996), is thought to signal to children that their environment is threatening and that they will be unable to cope or master such situations on their own. Consequently, children whose parents are more overcontrolling may perceive ambiguous or nonthreatening situations as threatening; therefore increasing their level of anxiety (Chorpita and Barlow 1998; Hudson and Rapee 2004; Wood 2006). Conversely, allowing children autonomy is thought to allow them to gain mastery over their environment, thereby reducing their perceptions of threat and decreasing their level of anxiety (Affrunti and Ginsburg 2012; Wood et al. 2003). One study found that attention bias (as measured through a visual search task) partially mediated the relation between maternal overcontrol and separation anxiety in sample of 129 non-referred children aged 6–14 years old (Perez-Olivas et al. 2008). Although this study used only nonanxious children in its sample, was restricted to separation anxiety as its only outcome measure, and the effect of parental overcontrol on interpretation (as opposed to attention) bias was not explored, it suggests that interpretation bias may mediate the relation between both parental anxiety and parental overcontrol and child anxiety.
Participants were 74 parent–child dyads. All parents met criteria for a DSM-IV anxiety diagnosis, generalized anxiety disorder (n = 52), social phobia (n = 10), obsessive compulsive disorder (n = 4), panic disorder with agoraphobia (n = 4), panic disorder without agoraphobia (n = 3), and specific phobia (n = 1). The presence or absence of diagnoses was determined by trained evaluators using the Client ADIS (Brown et al. 1994). Participating parents were primarily female (82.7 %) and ranged in age from 27 to 52 (M = 40.75, SD = 5.22). The majority of parents had a college degree or higher (88 %), a family income of 80,000 or more (76 %), and were married (85.3 %).
Children ranged in age from 7 to 12 years of age (M = 8.86, SD = 1.64) and were primarily Caucasian (82.2, 5.5 % African American). There was an even split between male and female children (52.1 % female). Twenty-three percent of the children had total scores over 25 (the suggested clinical cut off) on the Screen for Child Anxiety-Related Emotional Disorders—Child Version (SCARED-C); the range of scores was 0–41. None of the children were diagnosed with an anxiety disorder, or any other psychiatric or medical condition needing treatment, or were receiving psychological or pharmacological treatment aimed at reducing anxiety. This study selected a sample of non-anxious children as previous research has implicated excessive child anxiety as influencing parenting behaviors (Hudson and Rapee 2004; Hudson et al. 2009).
Families were recruited as part of a larger study examining the impact of an anxiety prevention program on non-anxious offspring of anxious parents (Ginsburg 2009). All families who contacted the study completed a preliminary phone screen to determine their eligibility, prior to an in-person evaluation. Families that were deemed eligible based on this phone screen were scheduled for an in-person assessment in which all the measures of the present study were administered. Prior to completing their initial evaluation, all participants, both children and parents, completed a written informed assent/consent.
Parental anxiety was measured using the State-Trait Anxiety Inventory (Trait Version) (STAI; Spielberger et al. 1983), a 20-item questionnaire measuring the stable, enduring symptoms of anxiety. The measure uses a 4-point Likert scale from 1 (almost never) to 4 (almost always) and yields a total score. Scores range from 20 to 80, where higher scores indicate greater anxiety. The STAI correlates highly with other measures of adult anxiety (rs = .73–.85) and has shown excellent test–retest reliability (rs = .73–.86). The internal consistency for this scale in the current sample was .91.
Parental overcontrol was measured using child reports on the Egna Minnen Beträffende Uppfostran—My memories of upbringing—Child version (EMBU-C; Muris et al. 2003a, b, c), a 40-item scale used to asses perceptions of parental behaviors. The questionnaire includes 4 subscales, each with 10 items; overprotection/control, emotional warmth, rejection, and anxious rearing. Each item is answered using a 4-point likert scale from 1 (no, never) to 4 (yes, most of the time). For the purposes of this study only the overprotection/control subscale was used. Scores range from 10 to 40, where higher scores indicate greater overcontrol. A sample item of this subscale is “your parents watch you very carefully.” The internal consistency for the 10-item subscale for the current sample was .65.
Child Interpretation Bias
Child interpretation bias was measured using five hypothetical, ambiguous stories using experimental procedures adopted from Barrett et al (1996), Muris et al. (2003a, b), and Lu et al. (2007). Two stories consisted of five sentences, one story consisted of four sentences, and the final two stories were three sentences. Stories consisted of two social stories (e.g. giving an oral report in front of the class, approaching unfamiliar children playing a game) and three generalized anxiety stories (e.g. riding your bike on a busy street, encountering your teacher on the playground, losing a book at school). Children were read stories aloud, one sentence at a time, by a trained independent evaluator. After each sentence, children were instructed to give a rating for how threatening they perceived the story would be (i.e., will the story have a bad or scary ending), on a scale of 0–4, where 0 = “not threatening at all”, 1 = “a little threatening”, 2 = “somewhat threatening”, 3 = “quite threatening”, and 4 = “very, very threatening.” Although children were told to focus on what would happen at the end of the story, threat ratings were obtained throughout the procedure. At the end of each story, children were read the story a second time, without interruption, and were then asked to provide an open ended answer to what would happen next in that situation. Scoring for this study involved only a determination of the threat rating. The threat rating was the mean rating of all responses, except for the open ended responses, totaled across the five stories. The range for the mean threat rating was from 0 to 2.9. The internal consistency for this measure was .91.
Child anxiety was measured using the SCARED-C (Birmaher et al. 1997). The SCARED-C is a 41-item measure of pediatric anxiety shown to differentiate between clinically anxious and nonanxious psychiatrically ill youth (Birmaher et al. 1997). Children answer questions using a 3-point Likert scale indicating to what degree a statement about themselves is true, from 0 (not true) to 2 (very or often true). The SCARED-C yields a total score, obtained by summing the 41 items, and five subscale scores which correspond to some of the DSM-IV anxiety disorders (subscales are panic, generalized anxiety, separation anxiety, social anxiety and school phobia). For the purposes of this study the SCARED-C Total score was used to assess overall anxiety levels. Possible total scores range from 0 to 84, where higher scores reflect higher overall levels of child anxiety. Internal consistency for the total score for this sample was .89.
Descriptive and Correlational Analyses
Range, means, standard deviations and first order correlations for all variables
Variable (n = 74)
Child interpretation bias
Child Interpretation bias rating
To examine the first research question, namely, does child interpretation bias mediate the relation between parental factors (anxiety and overcontrol) and child anxiety, we ran mediation analyses. Mediation was tested by determining the significance of the indirect effect of the independent variable (parental overcontrol and parental anxiety, X) on the dependent variable (child anxiety, Y) through the mediator (child interpretation bias rating, M), quantified as the product of the effects of Y on M and M on X, deducting the effect of Y. The Sobel test was used to determine if the indirect effect was statistically significant (Baron and Kenny 1986; Sobel 1982). As Fig. 1 shows, the total effect of parental overcontrol on child anxiety was significant (t = 4.06, p < .01). Also, there was a significant effect of parental overcontrol on child interpretation bias rating (t = 2.77, p < .01) as well as child interpretation bias rating on child anxiety, when controlling for parental overcontrol (t = 3.51, p < .01). This resulted in a significant indirect effect (z = 2.17, p = .03). That is, child interpretation bias rating was a significant mediator of the relation between parental overcontrol and child anxiety (Fig. 1). Despite significant mediation, the direct effect (c′; t = 3.08, p < .01) remained significant, suggesting that child interpretation bias rating was only a partial mediator.
As Fig. 1 further shows, the total effect of parental anxiety on child anxiety was significant (t = 2.48, p = .02). There was also a significant effect of parental anxiety on child interpretation bias rating (t = 2.33, p = .02) as well as child interpretation bias rating on child anxiety, when controlling for parental anxiety (t = 3.95, p < .01). This resulted in a significant indirect effect (z = 2.07, p = .04). That is, child interpretation bias rating was a significant mediator of the relation between parental anxiety and child anxiety (Fig. 1). The direct effect of parental anxiety on child anxiety (c′; t = 1.58, p < .12) was not significant, suggesting that child interpretation bias rating completely mediated this relation. There was no significant association between parental anxiety and parental overcontrol (t = .633, p = .53). Additionally, there was no significant interaction effect of overcontrol with either parental anxiety (t = .434, p = .67) or child interpretation bias rating (t = −.081, p = .94) on child anxiety.
To examine our second aim, regarding the relative strength or importance of the parental variables within this model, we tested the significance of the independent association between parental overcontrol and child interpretation bias and parental anxiety and child interpretation bias using Fisher’s r-to-Z transformation (Cohen and Cohen 1983). Correlation coefficients were determined by running linear regressions between either parental overcontrol or parental anxiety and child interpretation bias, while controlling for the other independent variable. That is, the regression between parental overcontrol and child interpretation bias controlled for parental anxiety (r = .07) and the regression between parental anxiety and child interpretation bias controlled for parental overcontrol (r = .13). According to Fisher’s r-to-Z transformation, used to compare the two correlation coefficients, the two associations were not significantly different (z = .34, p = .73). In other words, there was no difference in the strength of association between child interpretation bias and parental anxiety and parental overcontrol.
The twofold purpose of this study was to empirically examine whether child interpretation bias mediated the relation between parental factors (parental anxiety or overcontrol) and child anxiety and also to determine the relative importance of these parental factors in relation to child interpretation bias. Based on theoretical models (e.g. Chorpita and Barlow 1998) we hypothesized that child interpretation bias would mediate the relations between parental overcontrol and child anxiety and parental anxiety and child anxiety. Overall, our data provided partial support for this model. Consistent with previous research (Barrett et al. 1996; Chorpita et al. 1996; Creswell and O’Connor 2006; Dineen and Hadwin 2004; Muris et al. 2003a), children that ascribed higher levels of threat to ambiguous situations had higher levels of anxiety. Further, parents who exhibited higher levels of overcontrolling behaviors, such as watching their child very carefully, demanding to know what their child is doing, and not allowing their child to decide what they want to do, had children that ascribed higher levels of threat to ambiguous situations and had higher levels of anxiety. That is, interpretation bias influenced child anxiety directly, and acted as a partial intermediary for the relation of parental overcontrol and child anxiety.
Stronger mediational results were found when examining parental anxiety level in the place of parental overcontrol; as parental anxiety level increased, children’s tendency to interpret ambiguous situations as threatening also increased, as did their level of anxiety. This suggests that parents’ anxiety directly increases children’s interpretation biases, which subsequently increases their anxiety level. However, given the cross sectional nature of this study, it is equally plausible that increases in anxiety led to increases in perceptions of threat (Muris et al. 2004).
With respect to the second aim, there was no difference in the independent strength of the relations between parental anxiety and child interpretation bias and parental overcontrol and child interpretation bias, and, consistent with previous research, there was no relation between parental overcontrol and parental anxiety (Costa and Weems 2005). As such, it is likely that each exerts separate and unique effects. Specifically, the influence of parental anxiety on children’s interpretation bias may be due to genetics (see Eley and Zavos 2010), or anxious modeling (i.e., parents may voice their worries of danger and threat out loud in front of the child; Field 2006; Field et al. 2007; Murray et al. 2007a, b). Indeed, there is evidence that interpretation biases can be induced over repeated trials (Mathews and MacLeod 2002) and that this may be analogous to what a child with an anxious parent experiences as they develop (Lester et al. 2009). Parental overcontrol, as noted, may affect children’s interpretation bias by signaling that their environment is dangerous and restricting their opportunities to encounter (and thus minimize) perceived challenges or threatening situations.
These findings expand our understanding of the relation between parental anxiety, parental overcontrol, and child interpretation bias and child anxiety. In building on the work of Perez-Olivas et al. (2008), we found that interpretation bias, rather than attention bias, mediated the relation between parental overcontrol and child anxiety, and extended anxiety from one domain (separation) to a general rating of anxiety. Furthermore, we provided support for studies that have shown an association between parental anxiety and child interpretation bias (Bögels et al. 2003) but develop them further by including parental behaviors (i.e., overcontrol) and child anxiety levels in a single model. In finding a relation between parental overcontrol, parental anxiety, child interpretation bias, and child anxiety, we have provided support for the notion that parenting behaviors, as well as parental anxiety level, can be related to interpretation biases in children, and subsequently child anxiety. However, since our study only investigated parental overcontrol and parental anxiety, other parental behaviors (e.g. rejection, punishment, anxious modeling) that may influence both interpretation bias and child anxiety need to be examined.
The current study used correlational analyses and cannot claim any causal associations among these variables. Thus, it is possible that child anxiety leads to increases in child interpretation bias and higher overcontrol in parents. Furthermore, findings that parental overcontrol may be influenced by child anxiety (Becker et al. 2010; Costa and Weems 2005; Ginsburg et al. 2004; Manassis and Bradley 1994; Parker 1983) suggest that the relation between parental overcontrol, child interpretation bias and child anxiety may be reciprocal, where all three influence each other. Furthermore, because child interpretation bias accounted for 21 % of the variance in child anxiety, while parental overcontrol accounted for 18 % and parental anxiety accounted for 8 %, there are likely other variables (e.g. peer factors, coping skills, temperament) that affect the relation between these three constructs, which are worth investigating. For example, Costa and Weems (2005) found that self-reported maternal anxious attachment beliefs mediated the relation between maternal and child anxiety. However, the proportion of variance in child anxiety explained by interpretation bias is slightly higher than for other mediators of this relation, such as perceived competence (Affrunti and Ginsburg 2012).
Another limitation was the reliance on self-report measures that were completed by the child. Although children’s perceptions of these constructs are critical, using a single reporter can introduce reporter bias and cause reports to be influenced by factors such as the child’s comprehension or social desirability. The use of an independent and objective measure of child anxiety, parental overcontrol and interpretation bias could strengthen the associations and conclusions in future research.
Lastly, the generalizability of the study was restricted due to characteristics of the sample. The sample consisted of children 7–12 years of age. As such, it was impossible to examine the model within a developmental framework. Replication with older children is necessary to determine the relation of these constructs throughout adolescence. Also, the sample consisted of primarily of Caucasian children in two-parent upper, middle class families. Using a more diverse sample is necessary to better understand the relations between interpretation bias, parental anxiety, overcontrol and child anxiety among different racial/ethnic groups.
Results from the current study provide important information on the interplay of child interpretation bias, parental overcontrol, parental anxiety and child anxiety. The implications of these findings suggest that child interpretation bias is a mechanism of transmission between both parental anxiety and parental overcontrol, and child anxiety. This supports theoretical models, which suggest that parental overcontrol can signal threat to a child in an otherwise ambiguous situation, which increases the child’s level of anxiety (Chorpita and Barlow 1998). As such, treatments and prevention for child anxiety should focus on both child cognitive biases, such as interpretation bias, as well as parenting behaviors, such as overcontrol, that may play a role in the continuation of cognitive biases and anxiety in the child.
This research was supported by a grant from the National Institute of Mental Health (grant number R01MH077312-01) awarded to Golda S. Ginsburg.
- Becker, K., Ginsburg, G., Domingues, J., & Tein, J. (2010). Maternal control behavior and locus of control: Examining mechanisms in the relation between maternal anxiety disorders and anxiety symptomatology in children. Journal of Abnormal Child Psychology, 38(4), 533–543. doi: 10.1007/s10802-010-9388-z.PubMedCrossRefGoogle Scholar
- Birmaher, B., Khetarpal, S., Brent, D., & Cully, M. (1997). The screen for child anxiety related emotional disorders (SCARED): Scale construction and psychometric characteristics. Journal of the American Academy of Child and Adolescent Psychiatry, 36(4), 545–553. doi: 10.1097/00004583-199704000-00018.PubMedCrossRefGoogle Scholar
- Brown, T. A., DiNardo, P. A., & Barlow, D. H. (1994). Anxiety disorders interview schedule for DSM-IV. New York, NY: Graywind Publications.Google Scholar
- Cohen, J., & Cohen, P. (1983). Applied multiple regression/correlation analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum.Google Scholar
- Eley, T. C., & Zavos, H. M. S. (2010). Genetics. In J. A. Hadwin, A. P. Field, J. A. Hadwin, & A. P. Field (Eds.), Information processing biases and anxiety: A developmental perspective (pp. 209–232). Malden, MA: Wiley-Blackwell.Google Scholar
- Field, A. P., & Lester, K. J. (2010). Learning of information processing biases in anxious children and adolescents. In J. A. Hadwin, A. P. Field, J. A. Hadwin, & A. P. Field (Eds.), Information processing biases and anxiety: A developmental perspective (pp. 253–278). Walden, MA: Wiley-Blackwell.CrossRefGoogle Scholar
- Hudson, J. L., & Rapee, R. M. (2004). From anxious temperament to disorder: An etiological model of generalized anxiety disorder. In R. G. Heimberg, C. L. Turk, & D. S. Mennin (Eds.), Generalized anxiety disorder: Advances in research and practice (pp. 51–76). New York, NY: Guilford.Google Scholar
- Muris, P., Meesters, C., & von Brakel, A. (2003a). Assessment of anxious rearing behaviors with a modified version of ‘egna minnen beträffande uppfostran’ questionnaire for children. Journal of Psychopathology and Behavioral Assessment, 25(4), 229–237. doi: 10.1023/A:1025894928131.CrossRefGoogle Scholar
- Muris, P., Rapee, R., Meesters, C., Shouten, E., & Geers, M. (2003c). Threat perception abnormalities in children: The role of anxiety disorders symptoms, chronic anxiety, and state anxiety. Journal of Anxiety Disorders, 17(3), 271–287. doi: 10.1016/S0887-6185(02)00199-8.PubMedCrossRefGoogle Scholar
- Parker, G. (1983). Parental overprotection: A risk factor in psychosocial development. New York, NY: Grune & Stratton.Google Scholar
- Sobel, M. E. (1982). Asymptotic confidence intervals for indirect effects in structural equations models. In S. Leinhart (Ed.), Sociological methodology 1982 (pp. 290–312). San Francisco: Jossey-Bass.Google Scholar
- Spielberger, C. D., Gorsuch, R. L., Lushene, P. R., Vagg, P. R., & Jacobs, A. G. (1983). Manual for the state-trait anxiety inventory (form Y). Palo Alto, CA: Consulting Psychologists Press.Google Scholar