Child Psychiatry Hum Dev (2014) 45:12–23 DOI 10.1007/s10578-013-0374-x
ORIGINAL ARTICLE
Friendship Quality and Social Information Processing in Clinically Anxious Children J. R. Baker • J. L. Hudson
Published online: 27 February 2013 Springer Science+Business Media New York 2013
Abstract The association between perceived friendship quality (FQ) and social information processing (SIP) was examined in three groups of children and their close friends aged 7–12 years: 16 anxiety disordered children with social phobia (SP); 12 anxiety disordered children without SP (NoSP); and 32 nonclinical children. Positive and negative FQ positively associated with target children’s positive and negative responding on a vignette measure of SIP. SP children reported lower positive SIP than No-SP but not nonclinical children; and this was the only group difference in SIP. Target children and their friends were similar in negative but not positive SIP. Following discussion about the vignette with a close friend, all target children increased in positive SIP; negative SIP did not change. Lower FQ and a more socially anxious friend predicted higher negative target child SIP postdiscussion. Close friendships play an important role in the SIP of both clinical and nonclinical children. Keywords Friendship quality Social information processing Child anxiety Social phobia
Introduction Children with anxiety disorders are more likely to experience peer neglect, victimization, rejection, and dislike than Present Address: J. R. Baker (&) DCRC, School of Psychiatry, University of New South Wales, Level 3, AGSM Building, Sydney, NSW 2052, Australia e-mail:
[email protected] J. R. Baker J. L. Hudson Department of Psychology, Centre for Emotional Health, Macquarie University, Sydney, NSW 2109, Australia
123
nonclinical children [25], but they are no less likely to have at least one close mutual friendship [10]. The quality of this close friendship, or effectively how positive a child rates his or her friendship on features such as guidance, validation, intimacy or companionship and how negative a child rates his or her friendship on conflict and betrayal, is an important buffer in adjustment and in internalising symptoms specifically. Controlling for overall peer acceptance and number of friends, 10–14 year olds with at least one friend who offered support, protection, and intimacy, were less likely to display teacher-rated internalizing, externalising and social problems than those with a lower-quality friendship [46]. Furthermore, perceived friendship quality (FQ) showed an especially strong association with adjustment when overall peer acceptance and number of friends was low. For children with anxiety disorders at greater risk of such a scenario, a high quality friendship may be especially beneficial. Indeed, there is preliminary evidence that a high quality close friendship predicts better treatment outcome in clinically anxious children [2]. No study has explicitly looked at the FQ of clinically anxious children in comparison to controls, however community samples consistently purport shy, socially anxious or socially withdrawn children to have lower FQ than control children (see [25] for a review). Thus it is a reasonable conjecture that clinical samples of anxious children may also report friendships of lower quality than control children. How FQ impacts a child’s anxiety and consequent adjustment is an important research endeavour. Recent research suggests that FQ can influence a child’s social information processing (SIP) [6]. SIP is defined as how a child first encodes and interprets a social event, decides upon situational goals, evaluates possible responses to the situation and ultimately selects a strategy [11]. How children navigate their social-cognitive processes is an important
Child Psychiatry Hum Dev (2014) 45:12–23
influence on social adjustment [11], and negative processing biases have been clinically and experimentally identified as causal mechanisms through which child anxiety develops and is maintained [35]. In accord with this, one of the primary goals of cognitive behavioural therapy with anxious youth is to restructure negative cognitions. The skills provided by a good quality friendship such as perspective taking and social skills likely advantage such SIP. The positive feelings that high quality friendships can promote both about one’s self and others [42] are likely to contribute to more positive schemata, which when accessed during an encounter are also likely to encourage more positive SIP. Low quality friendships might conversely encourage negative cognitions. This may be especially true for anxious children who tend to engage in negative SIP [35]. It might also be astute to examine friends’ SIP or friends’ anxiety, and their relation to target children’s SIP. The concept has been explored with seventh graders and rejection sensitive cognitions, or the tendency to anxiously expect, readily perceive and overreact to experiences of possible rejection [7]. Higher rejection sensitivity was associated with a higher fear of negative evaluation in general, but the association was strongest for youths whose best friends were high in rejection sensitivity as opposed to moderate or low [7]. Field and Lester [15] describe a model how parents’ SIP can impress upon a child’s SIP. The same tenets appear relevant to how a friend’s SIP or anxiety could influence a child’s SIP. Akin to trials in a bias modification procedure, if over multiple learning experiences parents (or friends) consistently model or reinforce threatening interpretations to ambiguous stimuli through verbal information or vicarious learning such as fearful facial expressions or subtle behavioural reactions, this may increase the likelihood of the child acquiring a similar propensity towards making threatening interpretations in ambiguous situations. The effect may be augmented in high anxious children [30]. There is also the friendship specific process of corumination, or the extensive discussing and reaffirming of problems of negative affect between anxious friends [33]. Just as co-rumination might explain anxiety contagion between friends [37], discussing anxious attributions and interpretations of interpersonal events might lead to contagion of anxious SIP. This is somewhat experimentally illustrated in the FEAR paradigm—family enhancement of avoidant responses [3]. Children responded to hypothetical situations before and after a family discussion about the situations. Anxious children increased in avoidant responding following the discussion, whilst nonclinical children decreased in avoidant responding [3]. Analysis of the discussions showed that parents of anxious children were more likely than parents of nonclinical children to reciprocate avoidance with avoidance [12].
13
Utilising a design similar to the FEAR paradigm [3], the present study combined the literature on children’s FQ and the literature on children’s SIP to propose several unique hypotheses in relation to children with anxiety disorders. For the baseline measures (i.e. measures completed before the discussion with a close friend), it was hypothesised that clinically anxious children would have lower FQ (i.e. lower positive FQ and higher negative FQ) than nonclinical children. Secondly, it was predicted that target children would demonstrate similar SIP to their close friend. Next, the relationship between target child SIP and FQ was examined. Specifically, it was predicted that there would be a positive correlation between positive FQ and positive SIP and a positive correlation between negative FQ and negative SIP. Anxious children were also expected to report more negative SIP and less positive SIP than nonclinical children. Next, the main motivation behind the experimental design was the hypothesis that SIP would change between pre and post-discussion for all target children, and that this change would differ across the three groups. Lastly, the study was interested in what might predict this change in SIP. It was hypothesised that change in target child SIP from pre to post discussion would be predicted by FQ, or by close friends’ anxiety. Specifically, the present study asked; does discussion with a friend from a good quality friendship predict a better adjusted (i.e. more positive or less negative) SIP? Conversely, does discussion with a friend reporting high anxiety lead to more maladjusted (i.e. less positive or more negative) SIP? Throughout the study, there was also a subsidiary exploratory hypothesis that clinically anxious children with a social phobia diagnosis may behave differently to clinically anxious children without a social phobia diagnosis. Recent findings suggest evidence for a cognitive content specificity, in that children with a specific anxiety disorder are more likely to make a threat-related interpretation in situations pertinent to that disorder compared to children with a different anxiety disorder (e.g. [29]). For example, social anxiety symptoms of young adolescents showed a significantly greater correlation with negative appraisals of social than non-social situations [28]. Thus in the present study which utilised one ambiguous vignette relevant to social threat— selected because it had most potential and relevancy for peer discussion—it was expected that clinically anxious children with social phobia (referred to throughout the text as SP children) may perform differently on the SIP task than clinically anxious children without social phobia (referred to throughout the text as No-SP children). Recent findings have also emerged that clinically anxious children with SP may be significantly different to clinically anxious children without SP on notable peer and friendship measures (e.g. [20, 36, 45]). In light of this, and
123
14
Child Psychiatry Hum Dev (2014) 45:12–23
in an endeavour to maintain statistical consistency, participants were also analysed as three groups (SP, No-SP and nonclinical children) for the FQ analyses.
Methods Participants The present sample consisted of 60 participants and their close friends; 28 children with anxiety disorders and 32 nonclinical children. All children were aged between 7 and 12 years. Children from the anxious sample presented for assessment and treatment at Macquarie University Centre for Emotional Health. The primary diagnosis of an anxiety disorder set the criteria for assignment to the clinical group. Diagnoses and clinical severity ratings (CSRs) were made by clinical psychologists, or psychology graduates, under the supervision of senior clinical psychologists, using the anxiety disorders interview schedule for DSM-IV-child and parent version (ADIS-C/P; [38]). CSRs of 4–8 indicate the presence of a disorder. For the purposes of the hypotheses, the clinical group was further subdivided into two groups: clinically anxious children with SP in their diagnostic profile (n = 16; 57.1 %) and clinically anxious children without SP in the diagnostic profile (n = 12; 42.9 %), forming the SP group and the no social phobia (No-SP)
Table 1 Demographic composition of groups Demographic
Social phobia (n = 16) % (n)
No social phobia (n = 12) % (n)
Nonclinical (n = 32) % (n)
Gender: female
56.3 (9)
58.3 (7)
62.5 (20)
Ethnicity: Australian
92.9 (13)
81.8 (9)
93.8 (30)
Family income: [$104,000
60.0 (9)
49.9 (6)
68.8 (22)
Family structure: two parent
87.5 (14)
83.3 (10)
93.8 (30)
Table 2 Diagnostic composition of clinical sample
Diagnosis
No-SP group (n = 12) Anywhere in profile % (n)
Primary diagnosis % (n)
Anywhere in profile % (n)
Generalised anxiety disorder
50.0 (8)
81.3 (13)
66.7 (8)
100 (12)
Social phobia
43.8 (7)
100 (16)
–
–
–
43.8 (7)
56.3 (3)
66.7 (8)
6.3 (1)
31.3 (5)
–
8.3 (1)
Obsessive compulsive disorder
–
13.9 (10)
8.3 (1)
33.3 (4)
Mood disorder
–
6.3 (1)
–
–
Oppositional defiant disorder
–
12.5 (2)
–
8.3 (1)
ADHD (including subtypes)
–
6.3 (1)
–
–
Separation anxiety disorder
123
SP group (n = 16) Primary diagnosis % (n)
Specific phobia
ADHD attention deficit hyperactivity disorder, SP clinical social phobia group, No-SP clinical no social phobia group
group respectively. Children did not need to have a primary diagnosis of SP for qualification into the SP group. The view was that a primary diagnosis of SP is qualitatively the same as a secondary or tertiary diagnosis of SP and that the same set of symptom criteria applies. A SP diagnosis, be it primary, secondary or tertiary or so forth, may have an impact on perceived FQ. Nonclinical participants were recruited from the community via advertisements circulated to local independent schools and youth clubs, requesting ‘‘confident, worry-free children who have never sought help from a mental health professional’’. No diagnosis of any disorder as based on the ADIS-C/P [38] set the criteria for assignment to the nonclinical group. Nonclinical participants may have demonstrated some sub-clinical anxious symptoms but these were not systematically recorded. For this reason, the control group is purposely referred to as ‘‘nonclinical’’, as opposed to ‘‘nonanxious’’. The study formed part of a collective research session, for which target children received $50 for their participation. Table 1 provides the demographic composition of the three groups. No demographic data were collected for the friend. One-way ANOVA’s showed no significant differences between groups in age (SP: M = 8.88, SD = 1.75; No SP: M = 9.25, SD = 1.55; nonclinical M = 9.31; SD = 1.58), F (2, 59) = .404, p [ .05, or family income (ps [ .05). v2 tests showed no significant difference between groups on gender, family structure or ethnicity (p [ .05). No diagnostic data were collected for the friend. The diagnostic composition of the clinical sample is provided in Table 2. The study was most interested in the change in SIP from pre to post discussion with the close friend, and how this change might differ across groups (i.e. SP, No-SP group, and Nonclinical). Using the SPSS MANOVA procedure for power calculations [43], the sample size of 60 children would have provided power of over .8 to detect an average difference in change of .5 over a SD of 1, based on an estimated correlation of .8 between pre- and post-SIP.
Child Psychiatry Hum Dev (2014) 45:12–23
15
Measures
Spence Children’s Anxiety Scale (SCAS; [41])
Friend Nomination Form: Child and Parent
A 45-item Likert-type (0 = ‘‘never’’, 3 = ‘‘always’’) questionnaire designed to assess child reported anxiety symptoms relating to six DSM-IV defined sub-scales; SP (a = .74), separation anxiety (a = .74), panic attack/agoraphobia (a = .82), obsessive compulsive disorder (a = .76), generalised anxiety and physical injury (a = .77). The Spence Child Anxiety Scale for Parents (SCAS-P; [41]) corresponds with each item on the SCAS, less six positive filler items. In addition to the good internal consistency for the total scale and each subscale, and test– retest reliability [19], strong correlations between child and parent reports support concurrent validity of the scale [41]. Cronbach alphas for the present composite SCAS reports (and SP subscale) were excellent; target child a = .90 (a = .83); mother of target child a = .78 (a = .75); and friend a = .89 (a = .79). No parent report of friends’ anxiety was obtained.
Target child and friend friendships were identified using an established two-step nomination [31]. The first step asked children to list the ‘‘good friends you spend time with … really like to be around, know well, spend a lot of time together, and talk to’’’. There was no restriction on number of friends and children were not limited to naming school friends. Instructions specified not to include adults or relatives or opposite-sex friendships. The second step asked children to review this list and indicate their ‘‘very best friend’’. An additional step asked children to name their second and third very good friend. Lastly, children circled the name of the friend (i.e. first, second or third) whom they had chosen to accompany them to the research session (if not the first best friend it was often for logistical reasons). The parent version followed an analogous procedure; parents listed on an unlined page their child’s good friends, then from this list identified their child’s best, second and third good friends. An additional three forced-choice items were added to the questionnaire to identify friendship duration, frequency of contact and place of contact between dyads. Friendship Quality and Closeness Questionnaire [33] Is a shortened version of Parker and Asher’s [31] wellestablished Friendship Quality Questionnaire (FQQ). It comprises three items from each of the six subscales of the FQQ that showed the highest factor loadings (see [33]). The six subscales are: help and guidance (e.g. ‘‘we share things with each other’’), intimate exchange (e.g. ‘‘we tell each other private things’’), validation and caring (e.g. ‘we make each other feel important and special’’), companionship and recreation (e.g. ‘‘we do fun things together a lot’’), conflict resolution (e.g. ‘‘we make up easily when we have a fight’’) and conflict and betrayal (e.g. ‘‘we argue a lot’’). Replicating Rose [33], an additional emotional closeness subscale incorporated three items of the affective bond subscale of the Friendship Qualities Scale, a = .77, [8]; and four items of the Emotional Closeness scale, a = .76–.81, [9]. The final questionnaire totalled 25 items. Children indicated on a 5-point Likert scale (ranging from ‘‘not at all true’’ to ‘‘really true’’) how true a particular quality was of their friendship with the child with whom they attended the research session. Two scores were computed for each report; a positive FQ score reflecting the 20 positive items, and a negative FQ score reflecting the conflict and betrayal subscale. The current sample yielded very high Cronbach a for the positive score (child a = .92; friend a = .88), and negative score (child a = .81; friend a = .85).
Anxiety Disorders Interview Schedule for DSM-IV: Children and Parents Versions (ADIS-CP) [38] This is a semistructured interview designed specifically for the diagnosis of anxiety and other related disorders in 6–16 year olds. Diagnoses were assigned on a combination of information about symptoms and interference from both the child and parents. Children were assigned a principal diagnosis, representing the most distressing/interfering current problem, and any additional diagnoses for which they met DSM-IV criteria. The ADIS-C/P has demonstrated excellent interrater reliability at the Centre for Emotional Health (k = .92; [27]), test–retest reliability (k = .84) and sensitivity to treatment effects for children with anxiety [39]. The Children’s Evaluation of Everyday Social Encounters Questionnaire [4] Multistage vignette measures of SIP [4], offer greater predictive power over single or more general indices of social cognition [11]. The ChEESE-Q is a multi-stage vignette designed to assess the stages of SIP described by Crick and Dodge [11], and has been shown to predict internalizing symptoms [26]. Of the six hypothetical ambiguous social vignettes, one was selected for the present study: ‘‘You’ve just joined a club (like Scouts) and are going to your first meeting. You see that all of the other kids are looking at you when you come in, and one kid asks who you are’’. Aside from an initial open-ended question orienting children to the vignette, questions pursued a 5-point likert scale, from ‘‘definitely not’’ to ‘‘definitely’’ (unless otherwise stated).
123
16
As described in Luebbe et al. [26], interpretation is assessed with a causal attribution and an intent attribution question. For goal formation/clarification, the child rates the extent to which they would have each of the six listed goals in the situation. For response construction, the child indicates how likely they would enact each of the eight problem-solving solutions listed. For response selection the child indicates which potential solution they would most likely do. For behavioural enactment efficiency, the child rates (from ‘‘not at all well’’ to ‘‘very well’’) how well they could perform the chosen response. Response justification and goal attainment efficiency loaded onto a different factor (positive response evaluation style) not applicable to the primary hypotheses and were thus excluded. Consistent with Bell et al. [4], two subscales were created: a negative information-processing and responding factor (comprising negative causal and intent attributions; avoidant and distress expression goals and justifications; and passive avoidant, active avoidant and negative responding); and a positive information-processing and responding factor (comprising positive attributions; solution-focused, facesaving, and relationship-focused goals; positive efficacy evaluations; and prosocial responding). To maintain constancy with the positive and negative FQ distinction, the factors are referred to as positive and negative SIP—positive and negative SIP respectively. Four-week test–retest reliability estimates for the two factors are satisfactory (rs = .60–.68) and internal consistencies are high (negative a = .87; positive a = .84; [4]). The current sample yielded acceptable Cronbach a for the two factors; negative SIP (child pre and postdiscussion a = .74/.86; friend pre and postdiscussion a = .79/.77); positive SIP (child a = .78/ .81; friend a = .63/.75).
Child Psychiatry Hum Dev (2014) 45:12–23
SCAS, ChEESE-Q and FQ questionnaire. Nonclinical children and their mothers completed the ADIS-C/P at the research session. Target children and their mothers completed the ADIS-C/P at a separate appointment prior to the research session. The two friends were then brought together and left alone to discuss the ChEESE scenario for 3-min. Each child then completed the ChEESE-Q again. The ChEESEQ responses are referred to as pre and postdiscussion, respectively. Participants were fully debriefed and presented with one movie voucher each. The study was one of a battery of tasks in a collective research session. The research session totalled between 1.5 and 2 h; with the present SIP task taking about 40 min.
Results Plan of Analyses Preliminary analyses were conducted on the friend data: the number of close friends nominated; the concordance between mother and target child as to how close the accompanying friend was to the target child; the degree of friendship reciprocation between the target child and friend; and friendship descriptives such as duration and frequency of contact. Next, important covariates of age and gender were analysed regarding their relationship to FQ and anxiety symptoms. Thereafter, analyses were conducted in relation to hypotheses concerning group effects in FQ, similarities between friends in SIP, and the relationship between FQ and SIP. Lastly, analyses looked at target children’s SIP pre and post discussion with the close friend, and the predictors of SIP change.
Procedure Friend Nominations The procedures were approved by the Macquarie University Human Ethics Committee. Prior to the research session, interested families were sent the Friend Nomination Form-Child and an Information and Consent form to complete. A follow-up phone call established the nominated friend’s attendance and contact details, to whom an Information and Consent Form was sent, inviting them to participate in a study on ‘‘good friendships’’. For the purpose of confidentiality, no reference to anxiety, to the Centre for Emotional Health or to the target child’s involvement in treatment was made. Written consent was collected on arrival at the university, and participants directed to separate rooms to complete their questionnaires. Mothers completed the Friend Nomination Form-Parent and SCAS-P; target children completed the SCAS, ChEESE-Q and FQ questionnaire; and friends completed the Friend Nomination Form-Child,
123
Target children were asked to list the names of their good friends. The number of close friends nominated did not differ across groups, F (2, 59) = .90 p [ .05 (SP: M = 4.56, SD = 2.25; No-SP: M = 5.25, SD = 3.55; nonclinical: M = 5.88, SD = 3.50). Mothers were also asked to list the names of the target child’s good friends. The number of target children’s close friends as nominated by the child, and the number of target children’s close friend as nominated by the mother were significantly correlated, r = .581, p \ .001. However, the number of children reported by the mother significantly differed by group, F (2, 59) = 3.76, p = .029. Bonferroni post hoc comparisons indicated that parents of SP children reported their child to have significantly less close friends (M = 3.94, SD = 2.18) than parents of No-SP (M = 5.00, SD = 2.09), p = .025, and nonclinical children (M = 5.72, SD = 2.11), p = .025.
Child Psychiatry Hum Dev (2014) 45:12–23 Table 3 Friendship features across groups: closeness of the accompanying friend as based on mother and child report; the concordance between mother and child on the closeness of the accompanying friend; and friendship reciprocation
Friendship feature
17
SP (n = 16) Child % (n)
Mother % (n)
No-SP (n = 12)
Nonclinical (n = 32)
Child % (n)
Child % (n)
Mother % (n)
68.8 (22)
71.9 (23)
Mother % (n)
Reported closeness of friend Best friend
87.5 (14)
87.5 (14)
75 (9)
75 (9)
Second good friend
6.3 (1)
Third good friend
6.3 (1)
12.5 (2)
8.3 (1)
16.7 (2)
21.9 (9)
15.6 (5)
0 (0)
17.6 (2)
8.3 (1)
9.4 (3)
12.5 (4)
Concordance 100 %
93.8 (15)
91.7 (11)
93.8 (30)
6.3 (1)
8.3 (1)
6.3 (2)
Perfect
75 (9)
93.8 (15)
78.1 (25)
Close friend
35 (3)
6.3 (1)
21.9 (7)
One status step apart Friendship reciprocation SP clinical social phobia group, No-SP clinical no social phobia group
The majority of parents were concordant with their child, in that they agreed as to how close the accompanying friend was to the target child, i.e. best, second, or third close friend (see Table 3). A v2 analysis confirmed no significant group difference in mother or children nominations as to the status of the the accompanying friend (i.e. best, second or third close friend): v2 (4, n = 60) = 6.02, p [ .05; v2 (4, n = 60) = 9.99, p [ .05, respectively. Further, number of close friend nominations and accompanying friend status for mother and child report were not significantly related to child or friend FQ (ps [ .05). Friendship Reciprocation Invited close friends were also asked to list the names of their good friends. Friends of SP, No-SP and nonclinical children nominated an average of 4.88 (SD = 2.42), 5.33 (SD = 3.03), and 4.81 (SD = 2.66) ‘‘good friends’’, respectively. The average number of nominations was not found to be related to FQ, or to differ between group or reciprocation (ps [ .05). Reciprocation refers to the agreement between target child and friend as to whom they each nominated as their close friend. If the accompanying friend also nominated the target child as their ‘‘very best’’ friend, then it was deemed to be a ‘‘perfectly reciprocated’’ friendship. If the accompanying friend nominated the target child who invited them as their ‘‘second close friend’’ then this was coded as ‘‘second close friend’’. There were no ‘‘third’’ close friendships or nonreciprocated friendships—i.e. friendships in which the friend did not nominate the target child in their list of good friends. A v2 test reported no significant difference between groups on reciprocation status, v2 (6, n = 60) = 11.20, p [ .05. Descriptives are provided in Table 3. There was no indication of differences in target child or friend report of FQ, reported demographics, or target child or friend
symptom measures between children in perfectly reciprocated and children in ‘‘second close’’ friendships (ps [ .5). Thus perfectly reciprocated and second close friendships were collapsed together. Friendship Features All dyads had been friends for 1 year or more and were within 1 chronological year of each other. The majority of dyads saw each other five times a week or more (SP: 50.0 %, n = 8; No-SP: 66.7 %, n = 8; nonclinical: 59.4 %, n = 19), and attended the same school (SP: 75.0 %, n = 12; No-SP: 66.7 %, n = 8; nonclinical: 84.4 %, n = 27). SP dyads had been friends for an average of 4.88 years (SD = 2.42), No-SP friends for 5.33 years (SD = 3.03), and nonclinical friends for 4.81 years (SD = 2.66). One-way ANOVA and v2 analyses confirmed no significant differences between groups on the described features (p [ .05). Age and Gender Relationships with Friendship Quality and Anxiety One-way ANOVAs were conducted to assess for differences in anxiety between groups as measured by the SCAS and SCAS-P (see Table 4). Although the three groups did not significantly differ on age or gender, previous literature has indicated associations between age and gender, and FQ and anxiety; One-way ANOVAs reported no significant relationships between gender and the FQ or SCAS measures (ps [ .05). There were no significant correlations between age and child FQ (r = .168, p [ .05) or friend FQ (r = .168, p [ .05). Age was not related to the SCAS or SCAS-P total scores (ps [ .05), but did show a significant positive correlation with the SCAS SP subscale (r = .306, p = .017). Age was thus entered as a fixed factor into multilevel analyses.
123
18
Child Psychiatry Hum Dev (2014) 45:12–23
Table 4 Mean target child anxiety across groups Dunnett C post hoc \.05
Anxiety measure
Social phobia (n = 16) M (SD)
No social phobia (n = 12) M (SD)
Nonclinical (n = 32) M (SD)
F values
SCAS
32.50 (15.00)
28.17 (7.79)
17.78 (7.79)
F (2, 29) = 2.73, p = .062a
– a
30.81 (14.63)
34.58 (8.92)
11.56 (6.01)
F (2, 21) = 41.65, p \ .001
SCAS Soc
6.13 (3.76)
2.00 (1.28)
3.88 (2.67)
F (2, 31) = 10.68, p \ .001a
SP [ No-SP; No-SP \ nonclin
SCAS-P Soc
8.44 (3.16)
5.33 (2.54)
3.47 (2.29)
F (2, 24) = 15.76, p \ .001a
SP [ No-SP, SP [ nonclin; No-SP [ nonclin
Friend SCAS
30.75 (10.64)
33.33 (19.64)
30.25 (14.37)
F (2, 26) = .19, p [ .05
–
5.00 (2.780
5.83 (4.00)
5.25 (3.34)
F (2, 25) = .12, p [ .05
–
SCAS-P
Friend SCAS Soc
SP [ nonclin; No-SP [ nonclin
SCAS Spence Children’s Anxiety Scale, SCAS-P Spence Children’s Anxiety Scale-Parent, Soc social subscale, SP clinical social phobia group, No-SP clinical no social phobia group, nonclin nonclinical group a
Assumption of homogeneity of variance violated (p B .027), thus the Welch test was used
Actor–Partner Interdependence Model Multilevel modelling was used within the framework of the actor–partner interdependence model for distinguishable dyads [24]. The dyads were distinguishable in that there was some meaningful manner by which to distinguish the two individuals across dyads, i.e. target child or close friend. To enable the analysis to be carried out with the mixed linear model in SPSS 18, a stacked dataset was constructed consistent with the guidelines in Kenny et al. [24]. A level-1 variable contains separate scores for each individual. Level-2 is the dyadic variable and contains a single summary score that includes data contributed by both target child and friend. A score in the level-2 variable is the same for all the level-1 units nested within the level-2 unit. An alpha level of .05 was used for all statistical analyses [21, 32]. Follow-ups of overall main effects were carried out using an adjusted Bonferroni a value of .017 (i.e. .05 divided by 3—the number of comparisons). Friendship Quality The main aim of the FQ analyses was to assess for group differences in positive and negative FQ. The design also Table 5 Mean target child and friend friendship quality across groups
Friendship quality
permitted exploration of any differences across partner, i.e. between target child and friend. The dependent variable was FQ. The main effects of group (i.e. SP, No-SP or nonclinical) and partner were entered with number of comorbid diagnoses as a covariate. Age was centred at the mean (9.23 years). Interaction terms were used to explore whether group or partner effects differed according to the age of the dyad. The three-way interaction term of group 9 partner 9 age did not show significance (ps [ .05) and was removed from analyses. Gender and all two- and threeway interactions with partner and group were originally entered into each analysis, but did not show significance (ps [ .05) and thus removed. Attentive to the restricted sample size, group was also exploratory collapsed into two levels; clinical (i.e. SP and No-SP) and nonclinical. Findings were no different to the three levels of group; thus the SP, No-SP and nonclinical grouping was kept. Descriptives are provided in Table 5. There were no significant main effects of group or age on positive FQ (ps [ .05). There was a significant main effect of partner, F (1, 56) = 6.53, p = .013, and a significant partner 9 age interaction, F (1, 56) = 6.02, p = .017 (see Fig. 1). For younger dyads (7.66 years; 1 SD below the mean age), friends reported significantly higher FQ than the target
Social phobia (n = 16)
No social phobia (n = 12)
Nonclinical (n = 32)
Target M (SD)
Friend M (SD)
Target M (SD)
Friend M (SD)
Target M (SD)
Friend M (SD)
63.81
69.50
66.50
72.75
67.42
69.44
(9.83)
(10.23)
(16.04)
(11.50)
(11.70)
(10.79)
2.19
1.81
1.17
1.58
2.63
2.56
(1.64)
(1.68)
(1.59)
(2.43)
(2.51)
(2.92)
Reported Positive Negative
123
Child Psychiatry Hum Dev (2014) 45:12–23
19
children and friends in positive SIP (ps [ .05). However, there was a significant correlation between target child negative SIP and friend negative SIP (r = .46, p \ .001). Relationships Between Target Child Social Information Processing and Friendship Quality
Fig. 1 Positive friendship quality report for target child and friend (n = 60) ±1 SD from the mean age of 9.23 years. EMM estimated marginal means
child F (1, 56) = 13.48, p = .001. For older dyads (10.79 years; 1 SD above the mean age), perception of FQ did not significantly differ between target child and friend, F (1, 56) = .09, p = .760. To explore the possibility that this interaction was a reflection of friendship duration, additional analyses were conducted with duration included in the model. There were no significant main effects or interactions involving duration of friendship (ps [ .05), and the significant partner 9 age interaction for FQ upheld. There were no other significant interactions for positive FQ (ps [ .05). There were no significant main effects or interactions for negative FQ (ps [ .05). The SP and No-SP children were also analysed as a single group versus the nonclinical group with the view that it might provide more power to detect any differences between clinically anxious and nonclinical children. The partner 9 age interaction remained, but there were no significant main effects or interactions of Group (ps [ .05). Effect Size Concerned that the absence of any group effects for FQ may be due to the limited sample size and lack of power, effect size analyses were conducted. In line with the approach described by Snijders and Bosker [40], the random variation for the model with the group factor included was compared to a null model (intercept only) without the group factor included. Group was revealed to account for 1.4 % of the variance in positive FQ and 1.4 % of the variance in negative FQ. The marginal amount of variance contributed by group to the FQ models suggests that the null group effect is due more to it being a small effect than due to limited sample size. Similarities in Social Information Processing Between Friends The main aim of the Pearson correlations was to explore whether target children and their friends engaged in similar SIP. There were no significant correlations between target
The main aim of the Pearson correlations was to explore whether target child SIP was related to FQ. Prediscussion target child SIP and FQ (positive and negative) for both target child report (i.e. level-1 FQ) and dyadic report (i.e. level-2 FQ) were entered into the matrix. Given the partner 9 age interaction observed for FQ report, correlations were also split by age (7–9 and 10–12 year olds) to check for any differing results on the dyadic FQ correlations. Age did not impact the pattern of correlations. There was a significant positive correlation between child positive SIP and target child positive FQ (r = .42, p \ .001), and a significant positive correlation between target child negative SIP and target child negative FQ (r = .30, p = .010). There were no other significant correlations between FQ and SIP (ps [ .05). Group Differences in Social Information Processing There were two main aims of the mixed between-within subjects ANOVA. Firstly, to assess for group differences in target child SIP, and secondly to assess for change in SIP from pre and postdiscussion and whether this change differed across groups. To enable positive and negative SIP to be entered into the same model, factors were averaged and negative SIP reversed (i.e. a higher score now reflected a more positive SIP). There was a significant main effect of group, F (2, 57) = 3.95, p = .025. Pairwise comparisons revealed that the SP group was significantly lower in positive responding than the No-SP group (p = .007), but not significantly less so than the nonclinical group (p [ .05). The No-SP and nonclinical children did not significantly differ in responding (p [ .05). Regards within-subjects, the assumption of sphericity was highly violated (ps \ .001), thus only the multivariate tests were considered. There was no significant main effect of time (p [ .05), but a significant main effect of factor, Wilks k = .32, F (1, 57) = 58.80, p \ .001. There was also a significant time 9 factor interaction, Wilks k = .83, F (2, 56) = 5.65, p = .006. Pairwise comparisons revealed that across groups, positive SIP significantly increased from pre to postdiscussion (p = .001). There was no significant difference in negative SIP from pre to postdiscussion (ps [ .05). There were no other significant interactions (ps [ .05). The analysis was repeated with inclusion of the friend data (i.e. an additional two-way partner factor) to explore if this influenced findings; findings did not differ. In summary, SP target children were much
123
20
Child Psychiatry Hum Dev (2014) 45:12–23
Table 6 Target child social information processing pre and postdiscussion across groups SIP
Social phobia (n = 16)
No social phobia (n = 12)
Nonclinical (n = 32)
Pre M (SD)
Post M (SD)
Pre M (SD)
Post M (SD)
Pre M (SD)
Post M (SD)
Negative
23.19
22.94
19.5
18.67
22.16
23.28
(6.11)
(5.13)
(6.36)
(6.72)
(5.94)
(9.11)
Positive
33.75
37.00
39.67
41.42
37.59
39.13
(6.10)
(6.18)
(8.71)
(6.40)
(4.76)
(6.70)
processing,
Pre
prediscussion,
SIP social information postdiscussion
Post
lower in positive SIP than No-SP target children. Across groups, target children increased in positive SIP following the discussion with their friend. Descriptives are provided in Table 6. The SP and No-SP children were also analysed as a single group versus the nonclinical group with the view that it might provide more power to detect any differences between clinically anxious and nonclinical children. The time 9 factor interaction remained, but there were no significant main effects or interactions of Group (ps [ .05). Predictors of Change in Target Child Social Information Processing The main aim of the multiple regression was to explore whether changes in target child negative and positive SIP pre to postdiscussion were related to: FQ (target child and dyadic); or friend report of anxiety (as measured by the SCAS and SCAS SP subscale). Target child prediscussion positive or negative SIP were entered into analyses with postdiscussion positive and negative SIP respectively as a control. Friend anxiety was also entered into the regression as an interaction term with child anxiety and as an interaction term with group to examine whether a combination of both target child and friend anxiety predicted target child postdiscussion SIP. The friend anxiety 9 target child anxiety and friend anxiety 9 Group interactions did not significantly predict postdiscussion target child SIP (ps [ .05). Dyadic or target child FQ (negative or positive), friend’s anxiety or friend’s prediscussion SIP did not significantly predict target child positive SIP postdiscussion (ps [ .05). Dyadic FQ (negative or positive) and friends’ prediscussion SIP did not significantly predict negative SIP postdiscussion (ps [ .05). Friend SCAS did not significantly predict negative SIP postdiscussion, unstandardised coefficient B = .080, t (59) = 1.97, p = .053, but higher friend report of social anxiety did significantly predict higher target child negative SIP postdiscussion, B = .39, t (59) = 2.18, p = .034. Higher target
123
child positive FQ report also significantly predicted lower negative SIP postdiscussion, B = -.11, t (59) = -2.28, p = .026. Target child negative FQ did not significantly predict negative SIP postdiscussion (p [ .05).
Discussion Contrary to expectations, FQ did not differ across SP, NoSP and nonclinical dyads. This is inconsistent with literature in community samples purporting shy, socially anxious or socially withdrawn children to have lower FQ than control children [25]. However, nonclinical samples of anxious children can only inform the literature on clinical samples of anxious children. It could be that the friendships of clinically anxious children differ from the friendships of nonclinically anxious children. For example, although a little bit older than the present sample, the relationship between anxiety and friendship was found to differ between clinical and nonclinical 9–15 year olds [10]. Whilst having a best friend relative to not having a best friend was associated with lower social anxiety for nonclinical children, social anxiety was unrelated to having a best friend for clinically anxious children. Alternatively, the unexpected results could be due to a methodological bias. Across groups, friends reported higher positive FQ than target children in younger but not older dyads. It is reasonable that friends differ in their perception of FQ [16], and that younger children are less reliable in their reports than older children [13]. However the incongruence in younger dyads appears to be due to friends’ reporting especially high FQ which then declines with age. This is in opposition to the trend for FQ to increase with age [34]; as evidenced by target child FQ in the present study. The reasoning appears to be methodological. Target children were accompanied by their mother, whilst typically friends were not. This may have had no bearing on older friends, but for younger friends this may have introduced an apprehensive compliance that resulted in an overreporting of FQ. Attending the research session without their mother present, may have also augmented friends’ anxiety. This would explain the finding that friends scored more similar—although still within a subclinical range—to the clinical participants than the nonclinical participants. Whilst the high anxiety of the friends was consistent across all groups, the finding demands cautious interpretation of the results regarding friends’ anxiety. Nonetheless, moving towards the SIP hypotheses, both SP and No-SP children did not differ from nonclinical children in positive SIP. This is consistent with prior research documenting a lack of difference in positive thinking between clinically anxious and nonclinical children [5]; and moreover the general phenomenon of ‘‘the
Child Psychiatry Hum Dev (2014) 45:12–23
power of nonnegative thinking’’ [22]. This is the suggestion that for both children and adults it is the lower frequency of negative thoughts as opposed to the presence of positive ones that differentiates normal from disordered groups. What is inconsistent with prior research then is the lack of difference between the clinical (SP and No-SP dyads) and nonclinical children in negative SIP. There is evidence that a positive emotional state can ‘‘remove’’ interpretation biases for threat in children with anxiety disorders [18]. Having a close friend present at the research session, may have unintentionally induced a positive emotional state and diminished the otherwise typical, negative cognitions in anxious children. The presence of a friend may also have motivated anxious children’s tendency to ‘‘fake good’’ [22]. This would explain the finding that anxious children’s self-reports (not parent reports) of anxiety were below the clinical range [41] and undifferentiated from nonclinical children. An alternative explanation for the absence of any group differences in negative SIP may be due to the measure of SIP employed. In comparison to the typical thought-based measures (e.g. ‘‘what do you think about this?’’) that have demonstrated a difference between clinical and nonclinical children in ‘‘nonnegative thinking’’, the present multistage measure included items whereby the child indicated how likely and how well they would do several behavioural options. Nonnegative thinking may apply purely to thoughts surrounding an event but not regards a plan of action. The behavioural items may also help explain the unexpected finding that SP children reported lower positive SIP than No-SP children. Anecdotal testament is often made to the sensitive and caring nature of anxious children. This may motivate particularly prosocial intentions for both SP and No-SP children. When it comes to thoughts regards actual behaviour or self-efficacy in performing the behaviour however, fear of evaluation or lower social skills associated with a SP diagnosis [1, 41] may be too debilitating. SP children may not be so avoidant as to ‘‘ignore’’ the protagonist or ‘‘leave’’ the situation (negative options which could paradoxically draw attention to themselves); but the positive SIP options to ‘‘talk to the kid’’ and ‘‘tell them your name’’ may be just ‘‘too much’’ for the SP child. This unexpected finding presents an interesting question for future research. There was no difference between groups for the influence of a discussion with a close friend. Unlike the family studies that found a group 9 time effect, in that discussion with parents enhanced avoidant responding only for anxious and not aggressive and nonclinical children [3], discussion with a close friend did not selectively enhance negative SIP for anxious children and not nonclinical children. Instead discussion with a close friend increased positive SIP for all target children. This generic positive
21
‘‘push’’ of the friend might testify to the important socialising role of peers [17]. Conversely, discussion with a friend had no effect on target children’s negative SIP. This distinction is important in that negative thinking may be more strongly related to adjustment than the presence of positive thinking [44], with one hypothesis that concurring negative cognition may hinder the otherwise potentially facilitating effect of positive cognition [23]. Thus whilst a close friend might encourage positive SIP, how meaningful this is within the context of negative cognitions is an endeavour for future research. Moreover, without a control group of children simply completing the vignette measure again, without close friend discussion, one cannot rule out that the positive ‘‘push’’ of friends, might simply be an artifact of completing the same measure twice. It is interesting then that friends were similar to each other in negative but not positive SIP. Dyadic discussion of anxious attributions and interpretations might encourage mutual engagement in negative SIP. However this reasoning is somewhat inconsistent with the finding that friends’ SIP did not predict target children’s SIP postdiscussion. Rather friends’ social anxiety predicted target children’s negative SIP postdiscussion. The more socially anxious a friend, the more negative target children’s SIP was postdiscussion (with friends’ global anxiety showing a similar trend). The present study is small and exploratory, but it could be that having an anxious friend may exacerbate or maintain a child’s anxiety by encouraging negative SIP. The associations between higher positive FQ and more positive SIP and higher negative FQ and more negative SIP, are similar to the associations found between perceived FQ and SIP in overweight 10–15 year olds [6]. However, the current study recognises a shared method variance concern. Nonetheless the finding holds intuitive theoretical support in that high quality friendships can be reasoned to encourage more positive experiences and schemata, which when accessed during the processing of an event engage more positive SIP [11]. The finding adds to the literature on the developmental importance of a good quality friendship [46]; in this instance by encouraging adaptive SIP. A limitation of the current study is that SIP was assessed via children’s responses to one vignette. Disparate findings across the content and design of a measure are commonplace, both in terms of SIP style and in amenability of a belief to modification [14]. Given another situation close friends might have had little or an opposite influence. Anxiety is a heterogeneous construct and although the current vignette was specifically selected for its amenability to friend influence, a vignette of alternative content may have evidenced different group findings. Moreover, the vignette described a situation most likely encountered by both target child and friend previously. It would be
123
22
interesting to see how friends relate and influence each other’s SIP prospectively without confounds of past experience and memories. This original experimental study has established a tentative role for close friendships in children’s SIP that, with replication across a wider and ideally more naturalistic design of measures could lend itself to important developmental implications. Indeed viewing these findings on a broader level, the absence of any difference between the clinical and nonclinical groups across the analyses suggest that children with anxiety disorders may be quite similar to nonclinical children in their close friend experiences. However, before it is argued that there may not be a specific aetiological role for friends and FQ in models of child anxiety, but rather a comprehensive role for friends and FQ in child development more generally, replication of the original study design with a larger sample size and greater power is needed. Summary The association between perceived FQ and SIP was examined in three groups of children and their close friends aged between 7 and 12 years: 16 anxiety disordered children with SP; 12 anxiety disordered children without SP (No-SP); and 32 nonclinical children. There were no group differences in FQ. Positive and negative FQ positively associated with target children’s positive and negative responding on a vignette style multistage measure of SIP. It is reasoned that a high quality friendship likely encourages more positive schemata, and in turn, more positive SIP. SP children reported lower positive SIP—than No-SP but not nonclinical children; and this was the only group difference in SIP. Target children and their friends were similar in negative but not positive SIP. Dyadic discussion of anxious attributions and interpretations might encourage mutual engagement in negative SIP. However, friend SIP did not predict target children SIP following dyadic discussion about the vignette. Lower FQ and a more socially anxious friend predicted higher negative target child SIP postdiscussion. This finding has developmental implications in that a socially anxious friend or low quality friendship may exacerbate or maintain a child’s anxiety by encouraging a child’s negative SIP. Discussion with a close friend about the vignette increased positive SIP for all target children, but given that negative SIP is most strongly related to adjustment, the meaningfulness of this finding is queried. Nonetheless, this novel experimental design expands the important role that close friendships might play in the SIP of both clinical and nonclinical children.
123
Child Psychiatry Hum Dev (2014) 45:12–23
References 1. APA (1994) Diagnostic and statistical manual of mental disorders, 4th edn. American Psychiatric, Arlington 2. Baker JR, Hudson JL (2013) Friendship quality predicts treatment outcome in children with anxiety disorders. Behav Res Ther 51(1): 31–36. doi:10.1016/j.brat.2012.10.005 3. Barrett PM, Rapee RM, Dadds MM, Ryan SM (1996) Family enhancement of cognitive style in anxious and aggressive children. J Abnorm Child Psychol 24(2):187–203. doi:/10.1007/BF01 441484 4. Bell DJ, Luebbe AM, Swenson LP, Allwood MA (2009) The children’s evaluation of Everyday Social Encounters Questionnaire: comprehensive assessment of children’s social information processing and its relation to internalizing problems. J Clin Child Adolesc Psychol 38(5):705–720. doi:10.1080/1537441090310 3585 5. Bogels SM, Zigterman D (2000) Dysfunctional cognitions in children with social phobia, separation anxiety disorder, and generalized anxiety disorder. J Abnorm Child Psychol 28(2): 205–211. doi:10.1023/A:1005179032470 6. Bowker JC, Spencer SV, Salvy S-J (2010) Examining how overweight adolescents process social information: the significance of friendship quality. J Appl Dev Psychol 31(3):231–237. doi:10.1016/j.appdev.2010.01.001 7. Bowker JC, Thomas KK, Norman KE, Spencer SV (2011) Mutual best friendship involvement, best friends’ rejection sensitivity, and psychological maladaptation. J Youth Adolesc 40(5):545–555. doi: 10.1007/s10964-010-9582-x 8. Bukowski WM, Laursen B, Hoza B (2010) The snowball effect: friendship moderates escalations in depressed affect among avoidant and excluded children. Dev Psychopathol 22(4):749–757. doi:10.1017/S095457941000043X 9. Camarena PM, Sarigiani PA, Petersen AC (1990) Gender-specific pathways to intimacy in early adolescence. J Youth Adolesc 19(1):19–32. doi:10.1007/BF01539442 10. Chansky TE, Kendall PC (1997) Social expectancies and selfperceptions in anxiety-disordered children. J Anxiety Disord 11(4): 363 11. Crick NR, Dodge KA (1994) A review and reformulation of social information-processing mechanisms in children’s social adjustment. Psychol Bull 115(1):74–101. doi:10.1037/0033-2909. 115.1.74 12. Dadds MR, Marrett PM, Rapee RM (1996) Family process and child anxiety and aggression: an observational analysis. J Abnorm Child Psychol 24(6):715–734. doi:10.1007/BF01664736 13. Edelbrock C et al (1985) Age differences in the reliability of the psychiatric interview of the child. Child Dev 56(1):265–275. doi: 10.2307/1130193 14. Field A, Hamilton S, Knowles K, Plews E (2003) Fear information and social phobic beliefs in children: a prospective paradigm and preliminary results. Behav Res Ther 41(1):113–123. doi:10.1016/S0005-7967%2802%2900050-5 15. Field AP, Lester KJ (2010) Learning of information processing biases in anxious children and adolescents. In: Hadwin JA, Field AP (eds) Information processing biases and anxiety: a developmental perspective. Wiley-Blackwell, Malden, pp 253–278 16. Furman W (1998) The measurement of friendship perceptions: conceptual and methodological issues the company they keep: friendship in childhood and adolescence. Cambridge University Press, New York, pp 41–65 17. Harris JR (1995) Where is the child’s environment? A group socialization theory of development. Psychol Rev 102(3):458–489. doi:10.1037/0033-295X.102.3.458
Child Psychiatry Hum Dev (2014) 45:12–23 18. Hughes AA, Kendall PC (2008) Effect of a positive emotional state on interpretation bias for threat in children with anxiety disorders. Emotion 8(3):414–418. doi:10.1037/1528-3542.8. 3.414 19. Ishikawa S-i, Sato H, Sasagawa S (2009) Anxiety disorder symptoms in Japanese children and adolescents. J Anxiety Disord 23(1):104–111. doi:10.1016/j.janxdis.2008.04.003 20. Ginsburg GS, La Greca AM, Silverman WK (1998) Social anxiety in children with anxiety disorders: relation with social and emotional functioning. J Abnorm Child Psychol 26(3):175–185. doi:10.1023/A:1022668101048 21. Jaccard J, Guilamo-Ramos V (2002) Analysis of variance frameworks in clinical child and adolescent psychology: issues and recommendations. J Clin Child Adolesc Psychol 31(1): 130–146. doi:10.1207/153744202753441747 22. Kendall PC, Chansky TE (1991) Considering cognition in anxiety-disordered children. J Anxiety Disord 5(2):167–185. doi: 10.1016/0887-6185%2891%2990027-Q 23. Kendall PC, Howard BL, Epps J (1988) The anxious child: cognitive-behavioral treatment strategies. Behav Modif 12(2): 281–310. doi:10.1177/01454455880122007 24. Kenny DA, Kashy DA, Cook WL (2006) Dyadic data analysis. Guilford Press, New York 25. Kingery JN, Erdley CA, Marshall KC, Whitaker KG, Reuter TR (2010) Peer experiences of anxious and socially withdrawn youth: an integrative review of the developmental and clinical literature. Clin Child Fam Psychol Rev 13(1):91–128. doi:10.1007/s10567009-0063-2 26. Luebbe AM, Bell DJ, Allwood MA, Swenson LP, Early MC (2010) Social information processing in children: specific relations to anxiety, depression, and affect. J Clin Child Adolesc Psychol 39(3):386–399. doi:10.1080/15374411003691685 27. Lyneham HJ, Abbott MJ, Rapee RM (2007) Interrater reliability of the anxiety disorders interview schedule for DSM-IV: child and parent version. J Am Acad Child Adolesc Psychiatry 46(6): 731–736. doi:10.1097/chi.0b013e3180465a09 28. Magnusdottir I, Smari J (1999) Social anxiety in adolescents and appraisal of negative events: specificity or generality of bias? Behavi Cogn Psychother 27(03):223–230 29. Micco JA, Hirshfeld-Becker DR, Henin A, Ehrenreich M (2013) Content specificity of threat interpretation in anxious and nonclinical children. Cogn Ther Res 37:78–88. doi:10.1007/ s10608-012-9438-7 30. Muris P, Field AP (2008) Distorted cognition and pathological anxiety in children and adolescents. Cogn Emot 22(3):395–421. doi:10.1080/02699930701843450 31. Parker JG, Asher SR (1993) Friendship and friendship quality in middle childhood: links with peer group acceptance and feelings of loneliness and social dissatisfaction. Dev Psychol 29(4): 611–621. doi:10.1037/0012-1649.29.4.611
23 32. Pocock SJ (1997) Clinical trials with multiple outcomes: a statistical perspective on their design, analysis, and interpretation. Control Clin Trials 18(6):530–545. doi:10.1016/S0197-2456 (97)00008-1 33. Rose AJ (2002) Co-rumination in the friendships of girls and boys. Child Dev 73(6):1830–1843. doi:10.1111/1467-8624.00509 34. Rubin KH, Wojslawowicz JC, Rose-Krasnor L, Booth-LaForce C, Burgess KB (2006) The best friendships of shy/withdrawn children: prevalence, stability, and relationship quality. J Abnorm Child Psychol 34(2):143–157. doi:10.1007/s10802-005-9017-4 35. Salemink E, van den Hout M, Kindt M (2010) How does cognitive bias modification affect anxiety? Mediation analyses and experimental data. Behav Cogn Psychother 38(1):59–66. doi: 10.1017/S1352465809990543 36. Scharfstein L, Alfano C, Beidel D, Wong N (2011) Children with generalized anxiety disorder do not have peer problems, just fewer friends. Child Psychiatry Hum Dev 42:712–723. doi: 10.1007s10578-011-0245-2 37. Schwartz-Mette RA, Rose AJ (2011) Co-rumination mediates contagion of internalizing symptoms in youths’ friendships. In: Paper presented at the Society for Research into Child Development, Montreal 38. Silverman WK, Albano AM (1996) The anxiety disorders interview schedule for children for DSM-IV: child and parent versions. Psychological Corporations, San Antonio 39. Silverman WK, Saavedra LM, Pina AA (2001) Test–retest reliability of anxiety symptoms and diagnoses with anxiety disorders interview schedule for DSM-IV: child and parent versions. J Am Acad Child Adolesc Psychiatry 40(8):937–944. doi:10.1097/00004583-200 108000-00016 40. Snijders TAB, Bosker RJ (1999) Multilevel analysis: an introduction to basic and advanced multilevel modeling. Sage, London 41. Spence SH (1998) A measure of anxiety symptoms among children. Behav Res Ther 36(5):545–566. doi:10.1016/S00057967%2898%2900034-5 42. Sullivan HS (1953) The interpersonal theory of psychiatry. Tavistock, London 43. Taylor A (2011) MANOVA procedure for power calculations (SPSS). J Mod Appl Stat Methods 10(2):741–750 44. Treadwell KR, Kendall PC (1996) Self-talk in youth with anxiety disorders: states of mind, content specificity, and treatment outcome. J Consult Clin Psychol 64(5):941–950. doi:10.1037/0022006X.64.5.941 45. Verduin TL, Kendall PC (2008) Peer perceptions and liking of children with anxiety disorders. J Abnorm Child Psychol 36(4):459–469. doi:10.1007/s10802-007-9192-6 46. Waldrip AM, Malcolm KT, Jensen-Campbell LA (2008) With a little help from your friends: the importance of high-quality friendships on early adolescent adjustment. Soc Dev 17(4): 832–852. doi:10.1111/j.1467-9507.2008.00476.x
123