Coyne's (1976) interpersonal theory of depression posits that depressed individuals see a cycle of negative interpersonal exchanges that triggers increases in their depressive symptoms. More specifically, Coyne proposes that initially nondepressed but mildly dysphoric individuals hope reassurance from others in decree to alleviate their doubts about their own worth and lovability. Others initially respond beside genuine concern and support. However, the individual perceives this initial support as not enough and consequently escalates his or her symptoms in an action to secure more reassurance and implementation. Although others continue to provide support, they originate to experience feelings of irritation and guilt, prevailing to a separation between the content and affective quality of their kind statements. This discrepancy increases the individual's fear of rejection, which within turn leads to a further escalation of symptoms contained by order to restore the intuition of security. This downward spiral continues until any others withdraw from the individual or the individual seek treatment.
Joiner and colleagues posit that individual differences exist in the tendency to want reassurance from others (e.g., Joiner, Metalsky, Katz, & Beach, 1999a, p. 270). Excessive reassurance seeking is defined as "a relatively stable tendency to excessively and consistently seek assurances from others that one is loveable and worthy, regardless of whether such assurances enjoy already been provided." According to Joiner and colleagues (e.g., Joiner, Metalsky et al., 1999a, p. 270), excessive reassurance seeking constitutes the central feature of Coyne's (1976) supposition, as it serves as the vehicle through which the "distress and desperation of depression is transmitted from one person to another," triggering distrustful outcomes for all.
Although Coyne's (1976) model be originally intended to explain how the transactional relationship between reassurance seeking and depressive symptoms serves to maintain pre-existing depressive symptoms, Joiner and colleagues (e.g., Joiner, Katz, & Lew, 1999) enjoy proposed an extension of Coyne's model in which they conceptualize "excessive reassurance seeking" as a maladaptive interpersonal style that serves as a vulnerability factor to depression. More specifically, Van Orden, Wingate, Gordon, and Joiner (2005, p. 149) posit that excessive reassurance seeking behaviors are driven by importantly accessible, maladaptive, cognitive-interpersonal scripts (e.g., "If I get the impression bad, later I ask my parents if they love me") that become activated when individuals experience concerns more or less their self-worth and/or future. As such concerns are probable to be activated following the episode of stressors, Joiner and colleagues conceptualize the relationship between excessive reassurance seeking and depressive symptoms within a diathesis-stress framework. In other words, excessive reassurance seeking is hypothesized to serve as a vulnerability factor (e.g., diathesis) that interacts next to the occurrence of stressors to predict increases in depressive symptoms.
Joiner (1994a) also posits that excessive reassurance seeking is feasible to play a role in the "contagion" of depression. In a meta-analytic review of 36 studies examining contagious depression, Joiner and Katz (1999) concluded that findings provided robust support for the hypothesis that depressive symptoms in one personality predict the development of depressive symptoms surrounded by others in his or her environment. Joiner posited that individuals who possess lofty levels of reassurance seeking are imagined to be more susceptible to "contagious depression" than are their low reassurance seeking counterparts. More specifically, Joiner hypothesizes that the experience of depressive symptoms is likely to organize individuals to withdraw from their social environment. As individuals who possess excessive reassurance seeking tendency rely on significant others to assuage their concerns about their own level of lovability and others' degree of dependability, they are possible to experience increases in depressive symptoms when significant others become less available to assuage such doubts.
Research near adult populations have generated support for plentiful of the hypotheses derived from Joiner and colleagues' extension of Coyne's (1976) interpersonal theory. First, reassurance seeking have been demonstrated to be a valid, replicable, and cohesive construct, distinct from other interpersonal variables such as common dependency, doubt in others' sincerity, and dependence on close others (Davila, 2001; Joiner & Metalsky, 2001). Second, individuals who exhibit large levels of reassurance seeking own been shown to exhibit difficult levels of depressive symptoms than individuals who exhibit low level (e.g., Joiner, 1994a; Joiner & Schmidt, 1998). Third, excessive reassurance seeking has be shown to interact with stressors to predict increases in depressive symptoms (e.g., Katz, Beach, & Joiner, 1999; Joiner & Metalsky, 2001). Finally, reassurance seeking have been found to play a role contained by the contagion of depression, with individuals who exhibit big levels of reassurance seeking mortal more likely than their low reassurance seeking counterparts to develop depressive symptoms when interacting near a depressed partner or roommate (Joiner, 1994a; Katz et al., 1999).
Far less research have examined the relationship between excessive reassurance seeking and depressive symptoms in children. Preliminary research has shown that excessive reassurance seeking is associated next to depressive symptoms (Joiner, 1999) and depressive disorders (Joiner, Metalsky, Gencoz, & Gencoz, 2001) among youth psychiatric inpatients. Although such findings are consistent with hypotheses from Joiner and colleagues' extension of Coyne's interpersonal suggestion of depression (Joiner, 1994a; Joiner et al., 1999; Joiner, Metalsky et al., 1999a), several limitations should be noted. First, both studies have be conducted with youth psychiatric inpatient sample, and while providing preliminary support for the applicability of Joiner and colleagues' model to younger populations, exclusive use of inpatient samples may be a cut due to the lack of generalizability to community sample and the greater depressive severity and comorbidity levels observed among psychiatric inpatient youth (Newman, Moffitt, Caspi, & Silva, 1998). Second, both studies utilized cross-sectional designs. Consequently, although such studies illustrate that excessive reassurance seeking is a correlate of depressive symptoms, they do not provide a direct try-out of the hypothesis that excessive reassurance seeking serves as a vulnerability factor to depressive symptoms. Finally, given the relatively small sample sizes (i.e., n = 68-72) and wide open age ranges (i.e., 7-17 years of age) utilized in both studies, the authors be unable to provide a powerful nouns of whether the relationship between excessive reassurance seeking and depressive symptoms is moderated by age.
The goal of the current study be to provide a prospective test of the vulnerability and contagion hypotheses of Joiner and colleagues' extension of Coyne's (1976) interpersonal suggestion of depression (Joiner, 1994a; Joiner et al., 1999; Joiner, Metalsky et al., 1999a) in a sample of children and rash adolescents. Given that the transition from childhood to early young adulthood represents a time characterized by immense change surrounded by school environment, social relationships, erudite activities, cognitive nouns, and physiological development, it is probable that age-related differences in the relationship between excessive reassurance seeking and depressive symptoms emerge during this transition interval. One possible reason for the emergence of such differences is that the cognitive-interpersonal script hypothesized to drive excessive reassurance seeking behavior do not become consolidated until early young adulthood after repeated learning experiences own reinforced them (Hammen & Zupan, 1984). Providing indirect support for such a hypothesis, several researchers have found that cognitive vulnerability factor such as negative self-complexity (e.g., Abela & Veronneau-McArdle, 2002) and a depressogenic attribu-tional style (e.g., Abela, 2001; Nolen-Hoeksema, Girgus, & Seligman, 1992) do not fire up to moderate the relationship between stress and increases in depressive symptoms until early youth.
A second possible reason for the emergence of such age-related differences is that reassurance seeking behaviors may dwindle normative during the transition from childhood to early young adulthood. It is important to memo that in directive for reassurance seeking behavior to be considered excessive, one must compare rates of such behavior to rates that would be expected given an individual's age. As the normative base rates of reassurance seeking behaviors may become increasingly lower during this transition, such behaviors may be more imagined to be viewed as developmentally atypical, which may consequently increase the chance that others respond to such requests with rejection to some extent than with reassurance. Providing indirect support for such a hypothesis, research examining whether dependency, a similar nonetheless distinct cognitive-interpersonal construct, confers vulnerability to depression in youth suggests that elevated levels of dependency may be both normative and adaptive surrounded by children. More specifically, although elevated levels of dependency hold been found to serve as a vulnerability factor to depressive symptoms surrounded by both adolescents and adults (see Zuroff, Santor, & Mon-grain, in press), dependency has not be found to confer vulnerability to depression in children (e.g., Abela, Sakel-laropoulo, & Taxel, within press; Abela & Taylor, 2003). Further, although elevated levels of dependency own been found to be associated beside impairment in social functioning in adolescents and adults (e.g., see Zuroff et al., in press), elevated level of dependency have be found to be positively associated with social functioning in children (Fichman, Koestner, & Zuroff, 1996). Last, research using child and hasty adolescent sample reports a significant association between dependency and age (e.g., r = -.38, p < .001, Abela & Taylor, 2003) with hasty adolescents reporting lower levels of dependency than children.
In proclaim to provide a powerful test of the vulnerability and contagion hypotheses of Joiner and colleagues' extension of Coyne's (1976) interpersonal supposition of depression, we utilized a sample of parents near a history of major depressive episodes and their children. Such a taste is advantageous for several reasons. First, children of parents next to a history of major depressive episodes are four to six times more possible than other children to develop depressive symptoms (Goodman & Gotlib, 2002). Consequently, the use of such a sample is probable to maximize the number of children who experience increases in depressive symptoms during the course of the study. Second, a past history of leading depressive episodes is one of the best predictors of future depressive symptoms (e.g., Belsher & Costello, 1988). Consequently, the use of such a taster is likely to maximize the number of parents who experience depressive symptoms during the course of the study. Last, previous research have demonstrated that there is a temporal association between mother and child depressive symptoms (Hammen, Burge, & Adrian, 1991). Consequently, parents near a history of major depressive episodes and their children represent an just what the doctor ordered group in which to examine the contagion hypothesis.
The procedure involved an initial laboratory assessment where children completed measures assessing reassurance seeking and depressive symptoms. The procedure also involved a series of follow-up assessments, every 6 weeks during the following year, in which children completed measures assessing depressive symptoms and the episode of hassles. During respectively assessment, parents completed measures assessing depressive symptoms. The use of a multiwave longitudinal design allowed us to take an idiographic approach toward examining Joiner and colleagues' vulnerability and contagion hypotheses. More specifically, we examined whether the slope of the relationship between hassle and depressive symptoms within children adjectives across children as a function of excessive reassurance seeking. In addition, we examined whether the slope of the relationship between children's depressive symptoms and their parents' depressive symptoms assorted across children as a function of excessive reassurance seeking. One advantage of utilizing such a multiwave idiographic approach is that by obtain repeated assessments of levels of hassle and depressive symptoms within individuals over an extended time of year of time, we are able to congregate a relatively reliable estimate of each child's level of stress reactivity (e.g., his or her slope of the relationship between hassles and depressive symptoms). Given that vulnerability-stress theories are essentially theories of differential stress-reactivity, such an idiographic approach represents an great way to interview their vulnerability hypotheses. A second advantage of utilizing an idiographic approach is that for respectively child, high level of hassles can be operationalized contained by reference to his/or her own have it in mind level of hassle. Similarly, for each parent, lofty levels of depressive symptoms can be operationalized surrounded by reference to his or her own suggest level of depressive symptoms. Such an approach toward operationalizing respectively of these variables is likely to minimize the impact of individual differences in the reporting of hassle and parental depressive symptoms on findings.
We hypothesized that children who exhibit high level of reassurance seeking would report greater elevations contained by depressive symptoms following elevations contained by hassles than children who exhibit low level of reassurance seeking. In addition, we hypothesized that children who exhibit big levels of reassurance seeking would report greater elevation in depressive symptoms following elevation in their parents' level of depressive symptoms than children who exhibit low levels of reassurance seeking. Hypotheses be tested using both contemporaneous and time-lagged analyses. The use of time-lagged analyses allowed for a powerful examination of the direction of the effects obtain in our contemporaneous analyses. In other words, such analyses allowed us to distinguish between (1) a vulnerability-stress model where excessive reassurance seeking moderates the association between elevations contained by hassles/parental depressive symptoms at Time n - 1 and elevations contained by children's depressive symptoms at Time n and (2) a differential exposure model in which excessive reassurance seeking moderates the association between elevation in children's depressive symptoms at Time n - 1 and elevation in hassles/parental depressive symptoms at Time n (e.g., Bolger & Zuckerman, 1995). In adjectives analyses, we examined whether associations were moderated by children's age and/or sexual characteristics.
METHOD
Participants
Participants were recruit through ads placed surrounded by local newspapers as resourcefully as through posters displayed throughout the greater Montreal area (additional details are provided surrounded by Abela, Skitch, Auerbach, & Adams, 2005). The final sample consisted of 140 children (69 boys and 71 girls) and one of their parents (88 mothers and 14 fathers). In adjectives cases, the participating parent met criteria for either a current (n = 48) or ancient (n = 54) major depressive episode as assessed using the Structured Clinical Interview for the DSM-IV (First, Gibbon, Spitzer, & Williams, 2001). Participating parents have a mean age of start of their first major depressive episode at 20.8 years (SD = 10.5). Of the parents, 13.7% reported going through too various depressive episodes to count, and 15.7% reported having gone through a single episode. The remaining 70.6% of parents next to a history of major depressive episodes reported have gone through a mean of 6.1 episodes (SD = 8.0).
Children's ages range from 6 to 14 (mean = 9.8, SD = 2.3, median = 10). Parents' ages ranged from 27 to 53 (mean = 40.3, SD = 6.4, median = 41). The taster was 84.3% Caucasian. The mother tongue of the participant included English (68.7%), French (9.8%), Spanish (2.9%), and other languages (18.6%). At matching time, all of the participant were fluent surrounded by English. Of the parents, 14.7% were single, 43.1% be married, 9.8% were separated, 27.5% be divorced, 1.0% was widowed, 3.9% be none of the above. The median family income range from $30,000 to $45,000.
Procedure
Phase 1 of the study involved an initial laboratory assessment. Two research assistants met with one parent-child duo at a time. Parents completed a consent form and a demographics form. Children were told that their taking part was voluntary and they could choose not to share. All children decided to share. A research assistant read the Reassurance-Seeking Scale for Children (RSSC; Joiner, 1999) and the Children's Depression Inventory (CDI; Kovacs, 2003) aloud to the child while the child followed along and responded to questions using his/her own copy. Verbal command was chosen to ensure that children couched the instructions for each questionnaire, to prevent children from moving too smartly through questionnaires, and to build rapport near the children so that they would feel comfortable asking question. Parents completed the Beck Depression Inventory (BDI; Beck & Steer, 1987).
Phase 2 of the study involved a series of eight telephone follow-up assessments. Assessments occur every 6 weeks during the year following the initial assessment. At each assessment, a research assistant vocally administered the following questionnaires to children: (1) CDI and (2) Hassles Scale for Children (CHAS; Kanner, Feldman, Weinberger, & Ford, 1987). In adornment, a research assistant verbally administered the Beck Depression Inventory (BDI; Beck & Steer, 1987) to parents. One hundred and thirty-three children and their participating parents completed the Phase 2 assessments. The average number of follow-up assessments completed by participant was 4.79 (SD = 2.13). The number of follow-up assessments completed be not significantly associated with the following Time 1 variables: parental depressive symptoms (r = .11, ns), children's depressive symptoms (r = .09, ns), children's age (r = -.17, ns), and children's femininity (r = -.01, ns). At the same time, highly developed levels of reassurance seeking be associated with the completion of a greater number of follow-up assessments (r = .18, p < .05). The seven children who did not complete any Phase 2 assessments did not significantly differ from the 133 children who completed assessments on any Time 1 child or parent variables.
Phase 3 of the study occur 1 year after Phase 1 and involved a final laboratory assessment. A research assistant read the RSSC and CHAS aloud to the child while the child followed along and responded to questions using his/her own copy. In direct to examine the accuracy of children's withdraw of hassles, parents also completed a CHAS. At the finish of the assessment, participants be fully debriefed. Parents and children were compensated $180 for time lost and expenses incurred while participating in the current study. One hundred and six children and their participating parents completed the Phase 3 follow-up assessment. The 34 children who did not complete the Phase 3 follow-up assessment did not differ significantly from the 106 children who completed this assessment on any Time 1 child or parent variables.
Measures
The Structured Clinical Interview for the DSM-IV (SCID-I; First et al., 2001). The SCID-I is a semistructured clinical interview designed to arrive at current and lifetime DSM-IV diagnoses. The current study employed the affective disorders module and the psychotic peak in establish to allow for the diagnosis of all DSM-IV mood disorders. The SCID-I has be shown to yield reliable diagnoses of depressive disorders (Zanarini et al., 2000) and is frequently used in clinical studies of depression in adults. SCID-I interviews be conducted by the principal investigator, doctoral students in clinical psychology, and the project coordinators. All diagnosticians completed an intensive training program for administering the SCID-I and K-SADS interviews and for assigning DSM-IV and RDC diagnoses. The training program consisted of attending approximately 80 hr of didactic instruction, listening to audiotaped interviews, conducting practice interviews, and endorsement regular exams (85% or above). The principal investigator (PI) held weekly supervision sessions for the interviewers. The PI also reviewed interviewers' notes and tape in command to confirm the presence or absence of a diagnosis. Discrepancies be resolved through consensus meetings and best estimate procedures.
Beck Depression Inventory (BDI; Beck & Steer, 1987). This 21-item self-report inventory assesses the severity of depressive symptoms present inside the past 2 weeks. Each item is rate on a 0-3 scale near higher score indicating more severe depressive symptoms. The BDI has shown honourable concurrent validity when compared next to psychiatric ratings of depression in clinical populations (Beck & Steer, 1987). We obtained alphas range from .89 to .93 (mean = 0.91) across administrations, indicating glorious internal consistency.
Children's Depression Inventory (CDI; Kovacs, 2003). The CDI is a 27-item self-report questionnaire that measures the cognitive, affective, and behavioral symptoms of depression. For each item, children are asked whether it described how they be thinking and feeling surrounded by the past week. Items are score from 0 to 2 with a high score indicating greater symptom severity. The CDI possesses a giant level of internal consistency and distinguishes children next to major depressive disorders from nondepressed children (Saylor, Finch, Spirito, & Bennett, 1984). We obtain alphas ranging from .79 to .87 (mean = 0.83) across administration, indicating moderate to high internal consistency.
The Reassurance-Seeking Scale for Children (RSSC; Joiner, 1999). The RSSC is a modified revision of the Reassurance-Seeking Scale (RSS; Joiner & Metalsky, 1995) reworded for use with children. The revised ascend consists of four questions (e.g., I other need to ask my parents and friends if they really support about me). Items are score from 0 to 3 with a difficult score indicating greater reassurance seeking. RSSC scores exhibit moderate to dignified levels of internal consistency (Joiner, 1999; Joiner, Metalsky, Gencoz et al., 2001). Regarding authority, higher RSSC score have be found to be associated with, nonetheless distinct from, similar interpersonal constructs such as loneliness (r = .48, p < .01; Joiner, Metalsky, Gencoz et al., 2001) and dependency (r = .30, p < .001; Abela et al., 2005). In addition, RSSC score have be found to be not associated with insecure attachment to parents (r = .11, ns) and peers (r = .08, ns; Abela et al., 2005), suggesting that RSSC score are not confounded with the assessment of interpersonal variables indicative of a maladaptive interpersonal environment.
In directive to examine the psychometric properties of the RSSC in the current study, we first examined item-total correlations. (4) All four items exhibited item-total correlations greater than .72 at Time 1 (mean = 0.78) and .69 at Time 2 (mean = 0.77). In order to examine the factor structure of the RSSC, we conducted an exploratory principal axis factor analysis beside prom ax rotation on children's responses to the four items of the RSSC. Consistent with findings from studies using developed samples (e.g., Joiner & Metalsky, 2001), a one-factor solution emerge. All four items exhibited factor loadings greater than 0.57 at Time 1 (mean = 0.69) and 0.49 at Time 2 (mean = 0.67). We obtained have it in mind interitem correlations of .47 and .45 and alphas of .78 and .77 at Times 1 and 2, respectively, indicating moderate internal consistency. Test-retest reliability over the 1-year interval was 0.37 (p < .001), suggesting that reassurance seeking exhibited a moderate scope of stability in the current example over the 1-year follow-up interval.
Hassles Scale for Children (CHAS; Kanner et al., 1987). The CHAS is a list of 39 hassle that children may experience. Children rate how often respectively event happened to them during olden times 6 weeks on a scale of 0 (never) to 4 (all the time). A total evaluation is obtained by summing responses on adjectives items with better scores indicating a greater number of hassle. The majority of items focus on specific external hassles so as to prevent confounding the assessment of hassle and depressive symptomatology (Kanner et al., 1987). High CHAS scores hold been found to be associated beside increases in depressive symptoms over time in both third- and seventh-grade school children, both on their own and contained by interaction with vulnerability factor (e.g., Abela, 2001; Abela & Veronneau-McArdle, 2002). The mean test-retest reliability across the 1-year follow-up interval be 0.66, suggesting that children's levels of hassle exhibited moderate stability. (5)
In order to examine the rightfulness of children's reports, both parents and children completed the CHAS at the final follow-up assessment. Parents are likely to be aware of the happening of many of the items reported on the CHAS (e.g., "Your mother and father be fighting" or "Your mother or father were silly at you for getting a bad college report," etc.). Consequently, we would expect parent and child reports to be associated with one another. At one and the same time, there are several items on the CHAS that parents may not be capable of report accurately on (e.g., "Kids at school tease you" or "You didn't know the answer when the teacher call on you,"). Consequently, we would expect the strength of the association between parent and child reports only to be moderate. As expected, parents' and children's score were moderately associated near one another (r = .45, p < .001).
RESULTS
Descriptive Data
Means, standard deviations, and Pearson correlations between Time 1 measures and children's age and gender are presented surrounded by Table I. Pearson point biserial correlations between children's gender and adjectives Time 1 measures are also included in Table I.
As age was significantly associated beside reassurance seeking, we computed means and standard deviations for the RSSC separately for respectively age group. The pattern of results suggested that normative level of reassurance seeking behaviors decline during the transition from childhood to adolescence. (6) Thus, what may constitute excessive level of reassurance seeking for a 14-year-old is likely to represent normative level for a 6-year-old. As the hypotheses of the current study center around the relationship between "excessive" reassurance seeking and depressive symptoms, we consequently computed standardized RSSC scores (ST_RSSC) for respectively age group. Thus, high ST RSSC score reflect giant levels of reassurance seeking in comparison to the average smooth of reassurance seeking exhibited by same age peers. ST_RSSC scores a bit than raw RSSC score are utilized in adjectives subsequent analyses.
The Diathesis-Stress Hypothesis
To test our hypothesis that children who demonstrate glorious levels of reassurance seeking would report greater elevation in depressive symptoms following elevation in hassle than children who exhibit low levels of reassurance seeking, we utilized multilevel modeling. Analyses be carried out using the SAS (version 8.1) MIXED procedure and maximum likelihood estimation. Our dependent inconstant was within-subject fluctuations in CDI score during the follow-up interval (FU_CDI). As FU_CDI is a within-subject variable, CDI score were centered at respectively participant's mean such that FU_CDI reflect upward or downward fluctuations in a child's rank of depressive symptoms as compared to his or her mean rank of depressive symptoms. Our primary predictors of FU_CDI were AGE, reassurance seeking (ST RSSC), and fluctuations in CHAS score during the follow-up interval (FU_HASSLES). As AGE and ST_RSSC scores are between-subject predictors, AGE, and ST_RSSC score were standardized prior to analyses. As FU_HASSLES is a within-subject predictor, CHAS score were centered at respectively participant's mean prior to analyses such that FU_HASSLES reflect upward or downward fluctuations in a child's even of hassles as compared to his or her be going to level of hassle.
For all analyses presented, preliminary models be first examined testing whether GENDER exhibited a primary effect or served as a moderator of any relationships. No significant effects involving GENDER were obtain. In addition, for adjectives analyses presented, preliminary models were also tested examining whether Time 1 CDI score served as a moderator of any relationships (Joiner, 1994b). No significant interactions involving Time 1 CDI scores be found. Last, for all analyses presented, preliminary models be also tested examining whether Time 1 parental diagnostic status served as a moderator of any relationships. No significant interactions involving parental diagnostic status were found. Consequently, for the sake of simplicity, results are presented lone for models including AGE, ST_RSSC, and FU_HASSLES.
When fitting hierarchical linear models, one must specify appropriate mean and covariance structures. It is far-reaching to note that penny-pinching and covariance structures are not independent of one another. Rather, an appropriate covariance structure is essential in order to purchase valid inferences for the parameters surrounded by the mean structure. Overparametrization of the covariance structure can front to inefficient estimation and poor assessment of standard errors (Altham, 1984). On the other hand, too much restriction of the covariance structure can head to invalid inferences when the assumed structure does not hold (Altham, 1984).
In our first set of analyses, we were interested in examining the effects of AGE, ST RSSC, and FU_HASSLES on children's CDI score during the follow-up interval. Consequently, in line near Diggle, Liang, and Zeger's (1994) recommendation that one use a "saturated" model for the be a sign of structure while searching for an appropriate covariance structure, we chose a tight structure that included AGE, ST_RSSC, FU_HASSLES, and all two- and three-way interactions. Four extramural effects were also included here initial mean structure. First, surrounded by order to control for individual differences in baseline level of depressive symptoms, children's Time 1 CDI scores (T1_CDI) be included in the model. Second, in order to statement for the possible correlation in response variables between siblings from one and the same family, uninformed effects for children (RE_CHILD; random intercept) nested inside families (RE FAMILY; jumbled intercept) were included in the model. Finally, given that FU_HASSLES is a within-subject predictor whose effect is expected to alter from participant to participant, a random effect for slope (RE_SLOPE) be included in the model.
Commonly used covariance structures in studies in which multiple responses are obtain from the same individual over time (and consequently within-subject residuals over time are imagined to be correlated) include compound symmetry, first-order autoregressive, heterogeneous autoregressive, and banded Toeplitz. In order to select one of these covariance structures for our analyses, we fitted models utilizing each structure and chose the "best" fit base on Akaike information criterion (AIC and AICC) and Schwarz Bayesian criterion (BIC). In all cases, the best fit be a heterogeneous autoregressive structure.
After choosing the appropriate covariance structure, we next examined the random-effects component of our model. Nonsignificant random-effect parameter were delete from the model prior to examining the fixed-effects component. The one exception was RE_FAMILY. Regardless of this parameter's horizontal of significance, it was retained in the model to sketch for any nonindependence of data resulting from the inclusion of siblings in the study. With respect to variable effects, the ARH(1) parameter (r = .25, p < .001) and RE_CHILD (p < .001) were significant and thus be retained in the model. RE_SLOPE was not significant and consequently be deleted from the model prior to examining the fixed effects.
When examining the fixed-effects component of the model, we used a process of towards the back deletion. More specifically, we first examined the AGE x ST_RSSC x FU_HASSLES interaction. Given that this three-way interaction was not significant, it be deleted and the model be re-estimated. We next examined the two-way interactions. Two of the three two-way interactions be not significant: AGE x ST_RSSC ([beta] = -0.219, SE = 0.380, F(1,127) = 0.33, ns) and AGE x FU_HASSLES ([beta] = 0.004, SE = 0.010, F(1, 601) = 0.14, ns). Consequently, these interactions were delete and the model was re-estimated.
Results next to respect to the fixed-effects component of the model are presented in the top panel of Table II. Of primary rush, a significant two-way, cross-level interaction emerged between ST RSSC and FU_HASSLES. (7) In instruct to examine the form of this interaction, the model summarized in the top panel of Table II was used to work out predicted CDI scores for children exhibiting any low or high level of reassurance seeking (plus or minus 1.5 SD) who are experiencing either low or high-ranking levels of hassle in comparison to their own average rank of hassles (plus or minus 1.5 x aim within-subject SD). The results are presented in the upper panel of Fig. 1. As both FU_CDI and FU_HASSLES are within-subject variables centered at each participant's niggardly, slopes are interpreted as the increase in a child's CDI score that would be expected, given that he or she score one point higher on the CHAS.
[FIGURE 1 OMITTED]
Analyses be conducted for each ST_RSSC condition examining whether the slope of the relationship between hassle and depressive symptoms significantly differed from 0. Analyses indicated that children exhibiting high level of reassurance seeking reported higher level of depressive symptoms when experiencing high level of hassles than when experiencing low level of hassles, t(602) = 5.03, p < .001. At equal time, level of depressive symptoms did not alter as a function of level of hassle for children exhibiting low levels of reassurance seeking, t(602) = 0.25, ns. Planned comparisons of the slopes of the relationship between hassle and depressive symptoms revealed that the slope was significantly greater contained by children exhibiting high level of reassurance seeking (slope = 0.080) than in children exhibiting low levels of reassurance seeking (slope = 0.004; t(602) = 2.78, p < .01).
In demand to provide a stringent test of our hypothesis that lofty levels of reassurance seeking would be associated next to greater elevations within depressive symptoms following elevations contained by hassles, we also examined two second models. First, we conducted time-lagged analyses examining whether AGE, ST_RSSC, and/or their interaction moderated the relationship between fluctuations in hassles at Time n - 1 and fluctuations contained by depressive symptoms at Time n. (8) With respect to the random-effects component of the model, RE_CHILD (p < .01) was significant. Neither RE_FAMILY nor RE_SLOPE, however, be significant.
Results with respect to the fixed-effects component of the model are presented surrounded by the bottom panel of Table II. A significant three-way cross-level interaction was obtain between AGE, ST_RSSC, and FU_HASSLES. In order to examine the form of the three-way interaction, the model summarized in the bottom panel of Table II be used to calculate predicted CDI score for children (ages = 6, 7, 8, 9, 10, 11, 12, 13, or 14) demonstrating either low or elevated levels of reassurance seeking (plus or minus 1.5 SD) who are experiencing any low or high level of hassles contained by comparison to their own average level of hassle (plus or minus 1.5 x mean within-subject SD). The results of such calculation are presented in Table III.
Analyses be conducted for each AGE x ST_RSSC condition examining whether the slope of the relationship between hassle at Time n - 1 and depressive symptoms at Time n significantly differed from 0. Analyses indicated that children between the ages of 9 and 14 who exhibited high level of reassurance seeking reported higher level of depressive symptoms 6 weeks following the experience of a high plane of hassles than 6 weeks following the experience of a low plane of hassles. At alike time, level of depressive symptoms at Time n did not change as a function of level of hassle at Time n - 1 for either (1) children between the ages of 6 and 14 who exhibited low level of reassurance seeking or (2) children between the ages of 6 and 8 who exhibited high level of reassurance seeking.
For each AGE (ages = 6, 7, 8, 9, 10, 11, 12, 13, or 14), we conducted planned comparisons of the slopes of the relationship between hassle at Time n - 1 and children's depressive symptoms at Time n for children demonstrating high and low level of reassurance seeking. For children between the ages of 11 and 14, the slope was significantly greater within children exhibiting high level of reassurance seeking than in children exhibiting low levels of reassurance seeking. For children between the ages of 6 and 10, the slopes for children exhibiting large and low levels of reassurance seeking did not significantly differ.
Second, we conducted time-lagged analyses examining whether AGE, ST_RSSC, and/or their interaction moderated the relationship between fluctuations in depressive symptoms at Time n - 1 and fluctuations in hassle at Time n. With respect to the random-effects component of the model, RE_FAMILY (p < .01) was significant. Neither RE_CHILD nor RE_SLOPE be significant. Thus, both effects were delete from the model prior to examining the fixed-effects component. With respect to the fixed-effects component of the model, significant effects were not obtain for any variables.
Thus, the pattern of results obtain when examining the two time-lagged models suggests that children between the ages of 11 and 14 who showed high level of reassurance seeking exhibited elevations contained by depressive symptoms following elevations surrounded by hassles to some extent than exhibited elevations surrounded by hassles following elevation in depressive symptoms.
The Contagion Hypothesis
Similar analyses be conducted in charge to test our hypothesis that illustrious reassurance seeking would be associated with greater elevation in children's depressive symptoms following elevation in their parents' rank of depressive symptoms. As fluctuations in parental depressive symptoms is a within-subject erratic, BDI scores be centered at each parent's imply prior to analyses such that FU_BDI reflects upward or downward fluctuations surrounded by a parent's level of depressive symptoms as compared to his or her denote level of depressive symptoms.
Once again, for adjectives analyses presented, preliminary models were first examined trialling whether GENDER exhibited a main effect or served as a moderator of any relationships. No significant effects involving GENDER be obtained. In increment, for all analyses presented, preliminary models be also tested examining whether Time 1 CDI scores served as a moderator of any relationships (Joiner, 1994b). No significant interactions involving Time 1 CDI score were found. Last, for adjectives analyses presented, preliminary models were tested examining whether Time 1 parental diagnostic status served as a moderator of any relationships. No significant interactions involving parental diagnostic status be found. Consequently, for the sake of simplicity, results are presented only for models including AGE, ST_RSSC, and FU_BDI.
Similar to our previous analyses, we chose a plan structure that included AGE, ST_RSSC, BDI, and all two-and three-way interactions as the initial model for select the (residual) covariance structure. Once again, T1_CDI, RE_CHILD, RE_FAMILY, and RE_SLOPE were included in the model. In adjectives cases, the best fit was a heterogeneous autoregressive structure.
With respect to chaotic effects, the ARH(1) parameter (r = .22, p < .001) and RE_CHILD (p < .01) were significant. Neither RE_FAMILY nor RE_SLOPE, however, be significant. With respect to the fixed-effects component of the model, the AGE x ST_RSSC x FU_BDI interaction was not significant ([beta] = -0.009, SE = 0.021, F(1, 587) = 0.19, ns). Consequently, it be deleted and the model be re-estimated. Two of the three two-way interactions were not significant: AGE x ST_RSSC ([beta] = -0.377, SE = 0.388, F(1, 126) = 0.76, ns) and AGE x FU_BDI ([beta] = 0.027, SE = 0.020, F(1, 588) = 1.94, ns). Consequently, they be deleted and the model be re-estimated.
Results with respect to the fixed-effects component of our model are presented surrounded by the top panel of Table IV. A significant two-way cross-level interaction was obtain between ST_RSSC and FU_BDI. (9) In order to examine the form of the two-way interaction, the model summarized in the top panel of Table IV be used to calculate predicted CDI score for children showing either low or elevated levels of reassurance seeking (plus or minus 1.5 x between-subject SD) whose parents are experiencing any low or high level of depressive symptoms in comparison to their average even of depressive symptoms (plus or minus 1.5 x mean within-subject SD). The results of such calculation are presented in lower panel of Fig. 1. As both FU_CDI and FU_BDI are within-subject variables centered at respectively participant's mean, slopes are interpreted as the increase in a child's CDI gain that would be expected given that his or her parent scored one point greater on the BDI.
Analyses were conducted for respectively ST_RSSC condition examining whether the slope of the relationship between hassles and depressive symptoms significantly differed from 0. Analyses indicated that children exhibiting illustrious levels of reassurance seeking reported highly developed levels of depressive symptoms when their parents be experiencing high level of depressive symptoms than when their parents were experiencing low level of depressive symptoms, t(589) = 4.17, p < .0001. At the same time, rank of depressive symptoms did not vary as a function of stratum of parental depressive symptoms for children exhibiting low levels of reassurance seeking, t(589) = -0.44, ns. Planned comparisons of the slopes of the relationship between parental depressive symptoms and children's depressive symptoms revealed that the slope be significantly greater in children exhibiting glorious levels of reassurance seeking (slope = 0.138) than in children exhibiting low level of reassurance seeking (slope = -0.017; t(589) = 2.65, p < .001).
In order to provide a stringent question paper of our hypothesis that high level of reassurance seeking would be associated with greater elevation in children's depressive symptoms following elevation in parents' level of depressive symptoms, we also examined two additional models. First, we conducted time-lagged analyses surrounded by which we examined whether AGE, ST_RSSC, and/or their interaction moderated the relationship between fluctuations in parental depressive symptoms at Time n - 1 and fluctuations in children's depressive symptoms at Time [n.sup.6]. With respect to the random-effects component of the model, RE_CHILD was significant (p < .01). Neither RE_FAMILY nor RE_SLOPE, however, be significant.
Results with respect to the fixed-effects component of the model are presented within the bottom panel of Table IV. A significant three-way cross-level interaction was obtain between AGE, ST_RSSC, and FU_BDI. In order to examine the form of the three-way interaction, the model summarized in the bottom panel of Table IV be used to calculate predicted CDI score for children (ages = 6, 7, 8, 9, 10, 11, 12, 13, or 14) demonstrating either low or soaring levels of reassurance seeking (plus or minus 1.5 SD) whose parents are experiencing any low or high level of depressive symptoms in comparison to their own average height of depressive symptoms (plus or minus 1.5 x mean within-subject SD). The results of such calculation are presented in Table V.
Analyses be conducted for each AGE X ST_RSSC condition examining whether the slope of the relationship between child depressive symptoms at Time n and parent depressive symptoms at Time n - 1 significantly differed from 0. Analyses indicated that children between the ages of 9 and 14 who exhibited lofty levels of reassurance seeking reported complex levels of depressive symptoms 6 weeks following their parents experiencing lofty levels of depressive symptoms than 6 weeks following their parents experiencing low level of depressive symptoms. Further, children between the ages of 12 and 14 who exhibited low levels of reassurance seeking reported lower level of depressive symptoms 6 weeks following their parents experiencing high level of depressive symptoms than 6 weeks following their parents experiencing low levels of depressive symptoms. At indistinguishable time, children's level of depressive symptoms at Time n did not change as a function of parents' levels of depressive at Time n - 1 for any (1) children between the ages of 6 and 8 who exhibited high level of reassurance seeking or (2) children between the ages of 6 and 11 who exhibited low levels of reassurance seeking.
For respectively AGE (ages = 6, 7, 8, 9, 10, 11, 12, 13, or 14), we conducted planned comparisons of the slopes of the relationship between parental depressive symptoms at Time n - 1 and children's depressive symptoms at Time n for children showing high and low level of reassurance seeking. For children between the ages of 10 and 14, the slope was significantly greater within children exhibiting high level of reassurance seeking than in children exhibiting low levels of reassurance seeking. For children between the ages of 6 and 9, the slopes did not significantly differ as a function of reassurance seeking.
Second, we conducted time-lagged analyses examining whether AGE, ST_RSSC, and/or their interaction moderated the relationship between fluctuations in children's depressive symptoms at Time n - 1 and fluctuations in their parents' level of depressive symptoms at Time n. With respect to the random-effects component of the model, RE_FAMILY (p < .01) was significant. Neither RE_CHILD nor RE_SLOPE be significant. With respect to the fixed-effects component of the model, significant effects were not obtain for any variables.
Thus, the pattern of results obtain when examining the two time-lagged models suggests that children between the ages of 11 and 14 who showed high level of reassurance seeking exhibited increases in depressive symptoms following elevations within their parents' levels of depressive symptoms fairly than that parents experienced elevations surrounded by depressive symptoms following elevations within their children's depressive symptoms.
Finally, given that it is possible that excessive reassurance seeking in children is simply associated with elevation in depressive symptoms following elevation in parental depressive symptoms because elevation in parental depressive symptoms are associated near elevations contained by hassles, we examined whether any the ST_RSSC x FU_BDI interaction or the AGE x ST_RSSC x FU_BDI continued to be significantly associated with FU_CDI score after adding FU_HASSLES into the models summarized in the top and bottom panel of Table IV. Results indicated that following the inclusion of FU_HASSLES into the models, the ST_RSSC x FU_BDI interaction continued to be significantly associated with FU_CDI score in the contemporaneous analysis ([beta] = 0.043, SE = 0.019, F(1, 574) = 4.75, p < .05) and the AGE x ST_RSSC x FU_BDI interaction continued to be significantly associated near FU_CDI in the lag analysis ([beta] = 0.060, SE = 0.025, F(1, 329) = 5.50, p < .05).
DISCUSSION
The results of the current study provide partial support for the applicability of the vulnerability-stress hypothesis of Joiner and colleagues' extension of Coyne's (1976) interpersonal theory of depression (Joiner, 1994a; Joiner et al., 1999; Joiner, Metalsky et al., 1999a) to youth. More specifically, in file with Joiner and colleagues' hypothesis that excessive reassurance seeking represents a relatively stable individual difference, children's level of reassurance seeking exhibited moderate stability over the 1-year follow-up interval. Further, in line next to Joiner and colleagues' hypothesis that excessive reassurance seeking confers vulnerability to the development of depressive symptoms following the pervasiveness of negative events, the results from our contemporaneous analyses indicated that children who exhibited glorious levels of reassurance seeking reported greater elevation in depressive symptoms following elevation in hassle than did children who exhibited low levels of reassurance seeking. At like time, results from our time-lagged analyses indicated that the strength of this association varied as a function of age. More specifically, such analyses indicated that reassurance seeking be associated with increases in depressive symptoms at Time n following increases in hassle at Time n - 1 only surrounded by children between the ages of 9 and 14. In addition, such analyses indicated that the slope of the relationship between depressive symptoms at Time n and hassle at Time n - 1 was significantly greater within high as unwilling low reassurance seeking youth only for children between the ages of 11 and 14.
The results of the current study also provide partial support for Joiner's (1994a) hypothesis that individuals who exhibit lofty levels of reassurance seeking are more adjectives to the contagion effect than individuals who exhibit low levels of reassurance seeking. More specifically, in stripe with olden research examining the timing of the emergence of depressive symptoms in parents with a history of principal depressive episodes and their children (e.g., Hammen et al., 1991), elevations within parents' levels of depressive symptoms be associated with subsequent elevation in their children's level of depressive symptoms. Further, in vein with work by Joiner, the results of contemporaneous analyses indicated that this relationship be moderated by excessive reassurance seeking with children who showed lofty levels of reassurance seeking reporting greater elevation in depressive symptoms following elevation in their parents' depressive symptoms than did children who showed low level of reassurance seeking. At the same time, once again, results from our time-lagged analyses indicated that the strength of this association sundry as a function of age. More specifically, such analyses indicated that children's levels of reassurance seeking be associated with increases in their level of depressive symptoms at Time n following increases in their parents levels of depressive symptoms at Time n - 1 single in children between the ages of 9 and 14. In incorporation, such analyses indicated that the slope of the relationship between children's levels of depressive symptoms at Time n and parents' level of depressive symptoms at Time n - 1 was significantly greater within high as challenging low reassurance seeking only for children between the ages of 10 and 14.
Several factor may account for the emergence of age-related differences contained by our time-lagged analyses. First, the results of the current study indicated that reassurance seeking behaviors were significantly associated near age, with younger children reporting greater levels of reassurance seeking behaviors than elder children. Such a finding suggests that reassurance seeking behaviors decline during the transition from childhood to early young adulthood. As base rates of reassurance seeking behaviors decline during this transition extent, elevated levels of reassurance seeking behaviors may originate to be increasingly perceived as developmentally atypical and consequently may become increasingly more apt to be responded to in a negative carriage. Second, Joiner and colleagues (Van Orden et al., 2005) have lately posited that excessive reassurance seeking behavior is driven by highly accessible, maladaptive, cognitive-interpersonal script (e.g., "If I feel impossible, then I ask my parents if they love me") that become activate when individuals experience concerns about their self-worth and/or adjectives. It is possible that the cognitive-interpersonal scripts hypothesized to drive excessive reassurance seeking behavior do not become consolidated until untimely adolescence after repeated research experiences have reinforced them (Hammen & Zupan, 1984). Finally, it is possible that age-related differences emerge during the transition from childhood to impulsive adolescence near respect to whom youth turn when they seek reassurance. Peer relationships become increasingly key during the transition from childhood to early youth (e.g., Rubin, Bukowski, & Parker, 1998), and children may then be more apt to turn to their peers than their parents for reassurance. Parents and peers, however, may swing in the comportment in which they respond to reassurance seeking behaviors beside parents displaying a higher height of tolerance than peers. Although no past research have examined the relationship between reassurance seeking behavior and peer rejection, past research have indicated that high level of immaturity-dependency are associated with peer rejection (Schwartz, McFadyen-Ketchum, Dodge, Petit, & Bates, 1999). Peers consequently may be more apt than parents to respond negatively to excessive reassurance seeking behaviors.
A similar model of age-related findings has be obtained contained by past research examining other interpersonal vulnerability factor to depression in children. For example, although dependency have been found to serve as a vulnerability factor to depressive symptoms within adults (e.g., Zuroff et al., in press), research using child sample has yield less supportive results (e.g., Abela & Taylor, 2003; Fichman et al., 1996). Consequently, researchers own hypothesized that dependency needs are normative and adaptive contained by children leading dependency to emerge as a vulnerability factor to depression merely in youth when lower levels of dependency become more normative (Fichman et al., 1996). Future longitudinal studies are needed following children through the transition from childhood to rash adolescence surrounded by order to prospectively examine the impact of developmental factor on levels of reassurance seeking. For example, the emergence of a worldwide sense of self-worth (Harter, 1990) may result in smaller quantity need to hope reassurance from others as to one's self-worth. In addition, parent and peer socialization processes may discourage elevated levels of reassurance seeking. Future research should also examine whether high-ranking levels of reassurance seeking are smaller quantity likely to organize to depressive symptoms in younger children because reassurance seeking is smaller amount likely to organize to interpersonal rejection. In other words, if reassurance seeking behavior is normative in children, significant others may be less credible to respond to such behavior in children contained by a rejecting manner.
It is vital to note that the results of our time-lagged analyses suggest that the results from our contemporaneous analyses are best interpreted in a vulnerability-stress as opposed to a differential exposure framework. More specifically, the analyses indicated that, in dash with a vulnerability-stress framework, excessive reassurance seeking be associated with increases in depressive symptoms following increases in children's level of stress. At the same time, contrary to a differential exposure model, excessive reassurance seeking be not associated with increases in children's level of hassles following increases in their level of depressive symptoms. Similarly, strong support was obtain for a vulnerability-stress as opposed to a differential exposure model when examining the contagion hypothesis. As the current study be able to demonstrate that a pre-existing interpersonal vulnerability factor interacted beside subsequently occurring stressors to predict subsequent changes within depressive symptoms, results provide perhaps the strongest support to date for the applicability of vulnerability-stress models of depression to younger populations.
It is also high-status to note that the contagion effect observed within both our contemporaneous and time-lagged analyses was independent of the plane of hassles occurring in children's lives. In other words, excessive reassurance seeking continued to be associated near elevations surrounded by depressive symptoms in elder children following elevations contained by their parents' depressive symptoms even after controlling for children's levels of hassle. Joiner (1994a) has hypothesized that depressive symptoms contained by a significant other may represent a unique stressor explicitly not adequately assessed by hassle inventories. Future research examining the contagion hypothesis is likely to benefit from examining the biological and/or psychosocial piece of equipment through which the contagion process occurs contained by older children who exhibit soaring levels of reassurance seeking (see Joiner & Katz, 1999 for a description of potential mediate mechanisms).
Several limitations of the current study should be noted. First, self-report measures were used to assess depressive symptoms. Although both the CDI and the BDI possess high-ranking degrees of reliability and rightfulness, one cannot draw conclusions about clinically diagnosed depression base on self-report questionnaires. Second, self-report measures be used to assess stress. Although measures of life events that require participant only to indicate whether or not an event occur are probably less expected to be influenced by informant bias than those that ask subjects to rate the subjective impact of each event, more sophisticated methods of analysis such as interviewing procedures that assess contextual threat may provide better assessments of stress (Brown & Harris, 1978). Third, the current study single examined the relationship between excessive reassurance seeking and depressive symptoms. Thus, we were inept to identify whether excessive reassurance seeking served as a vulnerability factor specifically to depressive symptoms. Future research should assess a broader range of psychological symptoms surrounded by order to examine the specificity of excessive reassurance seeking to depressive symptoms in hasty adolescents. Fourth, the current study did not control for additional variables that could potentially statement for the relationship between reassurance seeking and depressive symptoms. Consequently, we cannot rule out the possibility that a third variable could portrayal for the pattern of findings obtain. At the same time, it is defining to note that research next to adult populations have obtained support for the relationship between reassurance seeking and depressive symptoms even after controlling for self-esteem (e.g., Joiner, Alfano, & Metalsky, 1992), interpersonal difficulties (e.g., Potthoff, Holahan, & Joiner, 1995), attachment insecurity (Davila, 2001), and neuroticism (Joiner, Metalsky, Katz, & Beach, 1999b). Finally, the current study utilized a token of self-identified, high-risk parents (predominantly mothers) and their children. Although such a design leads to a powerful question paper of both the vulnerability and contagion hypotheses, results cannot be generalized to other populations (e.g., samples not recruit specifically on the basis of parental depression). Future research is needed examining whether similar results are obtain using a community sample of children and their parents. In appendix, future research should include a greater proportion of father in command to rule out the possibility that the contagion effect observed in the current study is constrained to mothers and their children.
Thus, the results of the current study provide partial support for the applicability of Joiner and colleagues' expansion of Coyne's (1976) interpersonal theory (e.g., Joiner, 1994a; Joiner et al., 1999) to a youth token. At the same time, results suggest that dignified levels of reassurance seeking may be normative in younger children and consequently that excessive reassurance seeking may individual emerge as a vulnerability factor to depression during the transition from childhood to early youth. Discovering that excessive reassurance seeking confers vulnerability to depression in youth is important as such a finding is credible to inform the development of potent interventions for use with younger populations. For example, Van Orden et al. (2005) put emphasis on that individuals who engage within excessive reassurance seeking are likely to benefit from interventions such as Cognitive Behavioral Analysis System Psychotherapy (CBASP; e.g., McCullough, 2003) which focus on helping clients to spot interpersonal situations in which they work in maladaptive ways within order to swot up more adaptive interpersonal behaviors (i.e., new cognitive-interpersonal scripts). As research examining the role of excessive reassurance seeking in the etiology of depression in youth accumulate, a deeper understanding of the interpersonal processes that underlie the kick-off of this disorder in younger populations will emerge. Such understanding will ultimately provide greater insight into potential avenues for the treatment and prevention of depression in youth.
No comments:
Post a Comment