Children’s attachment-related narratives following US Gulf Coast hurricanes: Linkages with understanding and teacher stress

Abstract

The central focus of this study was the perceptions of emotional security among 64 elementary school-aged children exposed to the hurricanes that affected the US Gulf Coast in 2005. Specifically, we examined the representational qualities of attachment, exploration, and caregiving as assessed with a narrative story-stem task in relation to parental reports of children’s exposure to the hurricanes, their knowledge of hurricanes, and their teacher’s exposure to the hurricanes. Knowledge of hurricanes was assessed with a new narrative method representing hurricane conditions wherein children were asked to tell stories about what hurricanes are and what damage they could do. Children’s narrative representations of attachment, exploration, and caregiving were unrelated to parental reports of their children’s exposure to the hurricanes but were significantly related to their knowledge of hurricanes, specifically the effects of hurricanes on people, and to teachers’ reported loss of property related to the hurricanes. The findings suggest that a core component of children’s representational models is the capacity for empathy for the experience of others.

Keywords

attachment disaster hurricane knowledge narratives

In 2005, two major hurricanes tore through the Gulf of Mexico and made landfall in the southern US. Hurricane Katrina became the costliest natural disaster in US history, at approximately US$125b; it killed over 1800 people, and flooded over 80 percent of New Orleans (Graumann et al., 2006). Hurricane Rita came ashore in southern Louisiana only 1 month after Katrina. It totally destroyed several small coastal communities and the cost of its damage was estimated at over US$12b (Knabb et al., 2011). The combined devastation brought by Hurricanes Katrina and Rita hit the state of Louisiana particularly hard. Katrina displaced approximately 645,000 people in Louisiana, over 250,000 from New Orleans alone (Graumann et al., 2006), a disproportionate number of whom were poor (Gabe et al., 2005). Approximately 275,000 people in Louisiana were left homeless by both storms (Kent, 2006).

Within the affected communities, schools suffered extensive damage. Of the total 2000 primary schools in all of Louisiana, 1500 were in parishes most affected by the hurricanes of 2005, and 1445 of these were damaged by the storms (Kent, 2006). Over 40 schools were destroyed, and over 835 were significantly damaged (Louisiana Recovery Authority, 2006). The damage to schools was an important dimension of the impact of the storms on the lives of the children who attended them and the teachers who worked there. The current study investigated dimensions of school children’s emotional security in relation to their exposure to the storms and their abilities to articulate an understanding of the ways in which the hurricanes affected people and communities.

Stress, trauma, and exposure to natural disasters

Children exposed to natural disasters may face many of the same psychological risks as those exposed to other trauma-inducing circumstances. In general, greater proximity to extremely frightening circumstances heightens risk for traumatic outcomes for children (Pynoos et al., 1998). The psychological impact of natural disasters may be complicated, however, by the context of the events as these may play out over time. For example, Scheeringa and Zeanah (2008) found that children who remained in New Orleans during Hurricane Katrina experienced higher rates of post-traumatic stress disorder (PTSD) than children who evacuated, but the rate of PTSD among evacuees was also very high (62.5% and 43.5%, respectively). They linked the high rates of PTSD in evacuated children to other aspects of the context of the disaster, including frightening evacuations, return to devastated neighborhoods after the storm, and seeing their homes destroyed. The symptoms of PTSD for many children may gradually subside, yet Scheeringa and Zeanah’s assessment for traumatic effects began approximately 6 months following the storm. In a study of the impact of Hurricane Andrew in 1992, LaGreca et al. (1996) found severe PTSD symptoms in approximately 18 percent of children 7 months after the event and 12 percent of children 10 months after the event.

Observations of children’s behavior following the hurricanes of 2005 were extended to their classrooms by Buchanan et al. (2009). They found that elementary school teachers in hurricane-affected areas reported, approximately 3 months after the storms, significantly more spontaneous, hurricane-related activities led by children compared to teachers in three states not impacted by the storms. They also found evidence that teachers in Louisiana tended to be more active in calling attention to and encouraging children to discuss their personal storm-related experiences, which may well have assisted them in making sense of their experience through comparisons and affirmations with others sharing similar experiences. Louisiana teachers did not, however, provide more direct instruction related to the hurricanes, contrary to expectations. Some teachers expressed a desire to shield students from material that might be upsetting, and to reestablish routines as quickly as possible (Buchanan et al., 2009).

The way in which children understand extremely frightening events is likely to influence their perceptions of how the events can affect them, and, ultimately, their sense of vulnerability in relation to the events (Cordón et al., 2004). The information processing literature has provided a framework for understanding linkages between children’s knowledge of events and the psychological impact those events may have. According to this perspective, memories are encoded and stored within a psychological context that includes previously acquired knowledge and understanding associated with the event (Salmon and Bryant, 2002). Avoidance of knowledge or stimuli pertaining to an extremely frightening event is a common, albeit not necessarily optimal, form of coping (Coates and Gaensbauer, 2009), and is recognized as one of the symptom clusters for a diagnosis of PTSD (American Psychiatric Association, 2000). On the other hand, having accurate knowledge about a frightening event can displace uncertainty or factual errors that may fuel anxiety-related reactions. For some, the acquisition of knowledge about a frightening event may even promote feelings of mastery, a way to assert a form of control over an event that has drawn its power partially from its mystery and, therefore, be associated with more adaptive coping (Fivush and Sales, 2006).

Current views about effective psychotherapy for children exposed to trauma commonly include facilitation of the child’s construction of a “trauma narrative” about the event (Coates and Gaensbauer, 2009). The theory underlying this approach is that trauma memories, because of their overwhelmingly frightening nature, are not completely processed by children, that is, integrated into long-term memory that is accessible to recall, interpretation, and understanding. When extremely frightening memories are not available in this way, the intense emotions and physiological states associated with them can lead to destabilization and dysregulation of emotion and behavior, with little understanding of the reasons for this (Coates and Gaensbauer, 2009). A major goal of psychotherapy for these children is thus to create an understanding, usually through narrative means, of the feeling state and circumstances of the past frightening event, and to make clear that the event is not an actual threat in the present (Hodges et al., 2000). The therapeutic use of narrative methods to facilitate children’s representations of their experiences with adversity, according to Hodges et al. (2000), “allows them eventually to make their own sense of their experience” (p. 452).

More knowledge about a very frightening event, and the capacity to articulate this, may also reflect more exposure to adults who speak with the child about the meaning of the event, and this may reflect a relationship that is highly attentive to the child’s need for making sense of such an event (Salmon and Bryant, 2002). Fivush and associates have provided a rich research literature on the beneficial role of maternal “scaffolding” of reminiscence for children’s recall of events (Fivush and Nelson, 2006). Through this process, the child is provided an opportunity to share fears and have questions answered, and experience having his or her perception understood and acknowledged. The intimacy of sharing perceptions and reactions to frightening events greatly assists with re-regulation of physiological states aroused through the stress response. The acquisition of internal capacities for emotion regulation is considered a major developmental imperative, predictive of a wide range of psychological adaptations and interpersonal skills. The first lessons in emotion regulation are widely understood to be learned in the intimacy of attachment relationships (Rosenblum et al., 2009).

Attachment, emotion regulation, and empathic sensitivity

Bowlby (1982) conceptualized attachment as a behavioral system operating in a context of other complementary behavioral systems, notably exploratory and caregiving systems. Attachment security emerges when these behavioral systems operate more or less in synchrony, where the growing child’s alternating expressions for exploration and attachment are met with corresponding sensitive care that supports and attends to the child’s varying needs. As the growing child experiences distress from various unpredictable sources, the consistent presence of a sensitive and responsive caregiver promotes the child’s confidence in the availability of care.

In such a caregiving environment, the infant’s increasing capacities for internalized emotional and physiological regulation provide a foundation for many social and academic achievements throughout childhood (Rosenblum et al., 2009). A child’s abilities, for example, to maintain attention, persist in problem-solving, delay gratification, understand the perspectives of others, and participate in cooperative activities all require self-regulation. Personal experience with having one’s own states of distress effectively responded to appear to be facets of what is one interpersonal process, a “shared” understanding of emotional experience, which allows for the eventual development of explicit capacities for empathic responding to the distress of others. Such capacities, according to Stern and colleagues, represent the essential “intersubjective” nature of human psychological experience and understanding (Stern, 2008).

Whereas younger children, in infancy and into the preschool years, typically need the comfort of direct contact with attachment figures in times of great distress, as children mature they can rely increasingly for comfort and self-regulation on expectations derived from earlier experiences of care with attachment figures. Bowlby (1973) referred to “internal working models” of attachment relationships as the organized memories of experiences with attachment figures in times of activated need. Over time, the internal working models of a securely attached child provide “reasonably good simulations of himself and parents in interaction” that reflect histories of reliable care integrated with on-going updating reflecting maturation and the changing circumstances of the relationships (Bowlby, 1988: 130).

Among the innovations in developmental science of recent decades, new methodologies have been created designed to specifically assess qualities of internal working models. One impetus for this work has been the basic problem that to understand the internal perceptions of young children, direct self-report approaches are inadequate. As Main et al. (1985) wrote, the “move to the level of representation” in the measurement of developmental experience allows assessment of internal experience, and direct evaluation of the qualities and characteristics of internal working models.

Narrative story-stem techniques

Among the better known of measures in this tradition, narrative story-stem protocols provide structured story stems depicting common family-related stressors with the aid of simple doll figures and props, to which children are asked to freely respond. The foundation for this method was provided by two original protocols developed by Bretherton and colleagues, the Attachment Story Completion Task (ASCT; Bretherton, Ridgeway et al., 1990) and the MacArthur Story-Stem Battery (MSSB; Bretherton, Oppenheim et al., 1990). In subsequent years, many new stories have been developed by many researchers working around the world, but the essential structure and methods of the original protocols have been retained.

Decades of research with this method have shown that children’s enacted representations of close relationship qualities, as told through their spontaneous narratives, can be reliably assessed and are associated with relevant developmental experience, particularly dimensions of relationship qualities such as attachment security, family conflict, and peer relationships (see reviews in Bettmann and Lundahl, 2007; Page, 2001; Woolgar, 1999). Narrative assessments with traumatized children exposed to maltreatment have shown that they represent significantly more frightening and violent imagery, and that the structures of their narratives are more likely to be incoherent, suggesting emotional and behavioral dysregulation and incomplete processing of frightening memories (Heller et al., 2006; Hodges et al., 2000; MacFie et al., 2001). Narrative assessments have also been shown to correspond with interactional qualities observed in the home (Bretherton et al., 2013; Dubois-Comtois and Moss, 2008; Goodman et al., 1998), and actual behavior changes parents made in response to their participation in a parenting intervention (Toth et al., 2002).

Narrative story-stem techniques with children have only had limited applications in studies of assessments of caregiver characteristics likely to influence the quality of interactions with the child, although this is a growing area of the research literature. Children’s narrative representations of mother in authoritative roles has been inversely linked with self-reports of their psychological distress (Oppenheim et al., 1997); narrative enactments of aggression, avoidance, and danger have been linked to mothers’ real-life exposure to domestic violence and PTSD symptoms (Schechter et al., 2007); higher coherence and lower aggression in children’s narratives have been linked to mothers’ self-reports of emotional expressiveness (Laible, 2006); and in several studies, mothers’ depression has been linked to children’s narrative representations, including representations of self-worth (Bretherton et al., 2013), attachment, and family conflict, for both maternal and paternal depression (Cummings et al., 2008), and representations of family aggression (Goodman and Aber, 2010). Thus, some evidence exists that the psychological states of caregivers are associated with qualities of relationship representations created by young children in narrative story-stem protocols.

The influence of new caregiving relationships

According to attachment theory, children essentially apply expectations of caregiving qualities to new relationships with people in caregiving roles, using internal working models to understand and predict what to expect from unfamiliar relationships. Bowlby (1988) also theorized that as the child matures, internal working models of caregiving relationships are likely to be updated in response to new experience in such relationships. Internal working models are thus not static constructions but reflect lived experience. A limited literature addressing caregiving relationships outside the home provides some evidence for this process in early childhood. Buyse et al. (2011) investigated the question of whether qualities of the teacher relationship with a child might account for some of the variance in aggressive behavior with peers in school, beyond the contribution of attachment to mother. They found that teacher self-reports of closeness with the child interacted with children’s attachments to mothers, such that when relationships with teachers were closer, children were less likely to be aggressive, even when attachments to mothers were less secure. They also demonstrated that observed teacher sensitive responding in the classroom predicted teacher self-reports of closeness with children.

Similarly, using a data sub-set of the National Institute of Child Health and Development (NICHD) study of childcare, Watamura et al. (2011) found that high quality childcare provided a buffering effect on child internalizing and externalizing behavior problems for children with lower quality/low sensitivity caregiving at home. It is likely that the operating explanatory mechanism for these findings lies in the ways in which children incorporate new relational experience into representational models. As internal working models are altered with accommodation to new experience, updated expectations toward others in close relationships become evident in observed changes in behavior.

To date, the explicit study of associations of caregiving experiences outside the home with direct assessments of internal working model characteristics, specifically via narrative methods, has been very limited. A major focus for such research would logically be children’s day-care providers or teachers, to examine the extent to which qualities identified in these relationships would be identifiable in internalized representations. Only three studies are known to us where assessments of teachers’ perceptions of their relationships with children were examined in relation to children’s narrative representations. Each of these studies used the same instrument, the Student–Teacher Relationship Scale (Pianta, 1994), to assess the child–teacher relationship via teacher report. Among these, Page and Bretherton (2001) found that teachers’ reports of conflict with children were inversely associated with children’s narrative representations of maternal caregiving. Rydell et al. (2005) found that Swedish children who enacted narratives coded as secure, compared to those coded as avoidant, tended to have teachers who assessed their relationships as closer, more secure, and less conflictive. Vu and Howes (2012) found that children who created narrative representations of security in mother–child interactions tended to also have teachers who reported feeling emotionally closer to them, suggesting that secure children tend to transfer similar relational expectations to their teachers. Furthermore, these authors adapted four traditional story stems by having a teacher figure assume the role of central attachment figure in the story-stem plot, a “first” in this published literature. Children’s narrative representations characterized by “hyperactivation” (over-attention to negative elements, fearful or anxious behavior) were significantly associated with teacher reports of conflict in the relationship. None of these studies provided analyses to discern the directionality of these findings. We are aware of no studies that have examined whether aspects of teachers’ psychological well-being, which might be likely to influence relationships with children, were analyzed in relation to children’s narrative representations. Examinations of linkages and influences between qualities of children’s relationships with caregivers outside the home and their representations of caregiving relationships may contribute to an important expansion of our understanding of the developmental processes associated with emergence of emotional security.

This study is an attempt in this direction. Among other primary questions, this study examines the concordance between teachers’ psychological experience and children’s narrative representations. Its broader focus represents a shift from most of the research done with this narrative method, to address questions aimed at how and the extent to which circumstances associated with extreme stress outside the home influence the nature of children’s internal working models of attachment relationships.

Aims of the study

Based on Bowlby’s conceptualization of attachment, exploratory, and caregiving behavioral systems, the following three primary hypotheses were proposed:

Children’s narrative representations of attachment of children to parents, children’s exploratory behavior, and parental caregiving qualities would be linked inversely to the intensity of their experiences of exposure to the hurricanes of 2005 according to parental reports, specifically the degree of physical displacement or other forms of harm, as these events may have altered their expectations of care.

Children’s narrative representations of attachment, exploration, and caregiving would be positively linked to their expressed knowledge about the hurricanes, based on the supposition that more knowledge would likely reflect more awareness, more active coping, better understanding, better capacities for self-expression, and a reflection of their caregiving environments through the involvement of adults in caregiving roles who facilitated this understanding.

The emotional well-being of teachers exposed to harmful effects of the hurricanes would vary positively with children’s narrative representations of attachment, exploration, and caregiving, based on the supposition that emotional well-being would be likely to contribute to teachers’ capacities for positive responsiveness and engagement with children, while emotional distress would be likely to be associated with the absence of these qualities.

Methods

This study was part of a larger comparative study of teachers’ classroom responses in four states, and Louisiana students’ knowledge of hurricane effects and activities following the hurricanes of 2005 (Buchanan et al., 2009). This sub-study was designed to examine children’s understanding of the hurricanes using a newly adapted narrative method, described below, in relation to their experiences and their assessed narrative representations of relationship qualities.

Sample

A total of 64 children (35 males/29 females), aged 4–9 years (mean = 6 years), living in south Louisiana participated. Data regarding children’s exposure to the hurricanes were provided by 57 parents or guardians (89% of the sample) via survey instruments. The great majority of parent reporters were biological mothers (49 mothers, 5 fathers, 1 grandmother, 1 aunt, 1 step-father). Approximately 44 percent of parents/guardians who provided information concerning their children’s exposure to the hurricanes in 2005 reported that their child was in a community directly hit by one of these, and approximately the same number reported that their child had to evacuate as a result of the storms. Approximately 79 percent of children had “firsthand experience” of the hurricane, according to parent reports. The majority of families were described as headed by two biological parents. For approximately half the sample, parents’ highest level of educational attainment was high school. Approximately 50 percent of responding parents reported family incomes between US$20,000 and US$75,000. Sociometric data at the level of the school indicate that approximately 50 percent of the children in this sample were in classrooms where 74 percent of children received free or reduced school lunch, a commonly used index of extremely low income. In all, 54 parents reported their child’s ethnicity, and of these 43 (approximately 80%) were reported as Caucasian, 7 (13%) as African American, 2 as Native-American, and 1 each as Hispanic and Asian American.

A total of 15 teachers, of grades pre-kindergarten through Grade 3, participated. Of these, all but one reported living in a community affected by the hurricanes and/or their own evacuations and temporary stays in shelters set up for the hurricanes.

Recruitment and procedures

Teachers were recruited through contact with school principals in areas of Louisiana that were particularly impacted by the hurricanes of 2005. Some schools most heavily impacted by the hurricanes could not be included because at the time of the survey, in November 2005, they had not yet re-opened. For the larger multi-state study, 30 percent of surveys (n = 344) were returned by teachers in Louisiana. The 15 teachers from whose classrooms the children of this study participated were selected through stratified random sampling from among those participating in the larger study. Teachers distributed parent survey packets to the children in their classrooms (n = 280), and of these, 164 (43%) parents initially agreed to participate. A total of 121 children were initially interviewed, but usable narrative assessments were obtained for only 64, resulting mostly from time pressures to gather data. Children were interviewed during school hours, typically in an unoccupied schoolroom, on two separate, consecutive days, one to complete the NSST and the other for the “hurricane stories,” described below, approximately 7–8 months after the storms.

Measures

Narrative Story-Stem Task

The Narrative Story-Stem Task (NSST) is a generic term, adapted from Buchsbaum et al. (1992), used to describe a semi-projective assessment protocol for young children aged approximately 4–9 years. NSST protocols consist of standardized story stems, enacted with simple figures and props, to which the child is invited to respond by creating spontaneous narratives. NSST protocols generally assess children’s representations of close relationship qualities, although individual protocols differ with respect to the specific story stems included. The plot elements of NSST story stems are intended to stimulate reactions to specific relational challenges. For example, one story-stem sequence used in the ASCT (Bretherton, Ridgeway et al., 1990) is designed to stimulate children’s represented experience with separations and reunions by presenting the mother and father leaving and returning for an over-night trip and the grandmother staying with the children. Other story stems address represented caregiving, peer cooperation and conflict, empathic responsiveness, loss, and moral judgment, among other themes. At the conclusion of the presentation of the story stem, the child is asked to “show me and tell me what happens next,” and the narrative the child creates is analyzed for themes of interest. Warm-up and “wind-down” stories are presented at the beginning and end of the protocol, and not coded, to provide orientation and transition.

The NSST story stems used in the present study, with their provenances, are presented in Table 1.

Table 1.

Narrative Story-Stem Protocol.

1. Spilled Juice: The family is sitting at the table and little Jane/Robert reaches for some juice and spills the pitcher on the floor.^a

2. Hurt Knee: The family is taking a walk in the park. Little Jane/Robert tries to climb a “high, high rock” and falls off it, crying, “I’ve hurt my knee, I’m bleeding!”^b

3. Monster in the Bedroom: It’s bedtime and the mother tells little Jane/Robert to go to bed. He or she goes into the bedroom and cries out, “There’s something scary in my room! There’s something scary in my room!”^b

4. Departure: The mother and dad are going to go on a trip for 3 days, and say to the children, “See you in three days, Grandma will stay with you.”^b

5. Reunion: The mother and dad return from their trip.^a

6. Uncle Fred: The mother sits on the couch, crying because Uncle Fred has died. The younger child stands facing her, some distance away.^c

7. Ball Play: Jane/Robert is playing with his or her ball with his or her friend, Sally/Pete. Suddenly, Jane/Robert cries, “Ouch! That hurt my hand!” Experimenter asks: Why do you think Sally/Pete did that?^d

From the Attachment Story Completion Task (Bretherton, Ridgeway et al., 1990).

From the Attachment Story Completion Task (Bretherton, Ridgeway et al., 1990), adapted by Granot and Mayseless (2001).

From Zahn-Waxler et al. (1994).

Adapted from Warren et al. (2000).

As noted in Table 1, the ASCT was the principal protocol used, although with modifications for older children for three of the original five story stems, as provided by Granot and Mayseless. Two other stories in the present protocol (Uncle Fred and Ball Play) were adapted from Zahn-Waxler and colleagues and Warren and colleagues, respectively.

The approach to coding the NSST taken in the present study was developed by Page (2007), where three principal scales are used, consisting of 7 points each, reflecting Bowlby’s (1982) original conceptualization of the complementary nature of attachment, exploratory, and caregiving behavioral systems. Coding of children’s narratives involves creating verbatim transcripts, with descriptions of children’s behavior associated with their narratives, and applying codes to transcripts while viewing the narrative enactments on videotape. As a first step, 20 descriptive codes are applied each time these occur in the narrative. These codes reflect interactive behavior such as prosocial and conflictive peer interactions, parental nurture and discipline, attachment behavior, parental affection and conflict, and so on. (The initial development of this approach can be found in Page and Bretherton (2001).) The descriptive codes as applied in each transcript are then translated into ratings on the three scales of attachment, exploration, and caregiving according to decision rules contained in the coding manual (Page, 2007), and these ratings become the variables used in data analyses. Prior research with this coding system demonstrated associations of the three NSST scales with mothers’ and teachers’ ratings of children’s behavior with peers (Page et al., 2011).

The first author (T.P.), the originator of the coding system, coded all narratives. The third author (O.V.) independently coded 48 percent of the sample. Obtained inter-rater reliabilities (weighted kappa) for the three scales were as follows: Attachment, .93; Exploration, .79; and Caregiving, .85. Obtained differences among the double-coded transcripts were resolved through conference.

Children’s exposure to the hurricanes

Parents completed brief, non-standardized surveys, specifically created for this project, concerning their children’s exposure to the hurricanes in addition to family demographic data. Parents indicated how their children were exposed to the hurricanes by checking boxes associated with five conditions: Child was in one of the communities directly hit, had to evacuate, stayed in a shelter, had evacuees living in the home, and/or family lost property. A composite variable was created of all these conditions by summing them and the resulting variable ranged from 0 to 5, with higher scores reflecting greater exposure. In addition, parents rated on a 5-point scale, not included in the exposure composite, the extent to which they believed the child’s or their own life had been threatened.

Assessment of children’s knowledge of hurricanes: The Hurricane Stories

Children’s knowledge of hurricanes was assessed using a similar technique as the NSST but with six specially created story stems, referred to here as the “Hurricane Stories.” The intention of the hurricane story protocol was to assess children’s factual knowledge of technical aspects of hurricanes, in contrast to the established uses of the NSST which concern stressful relationship situations. Children were presented with five story scenarios using simple props (sand, trees, “water” in the form of confetti, farm animals, human family figures, pets, houses, work, and rescue crews). The presentation of these props was spaced across the five scenarios, and children were told that a hurricane was coming and asked what would happen to whichever props were presented in the story stem. The sixth and final story presented a new child moving to the community from someplace where there are no hurricanes and asked children what they would tell him about hurricanes (with prompts directed to wind, rain, what makes hurricanes, how hurricanes can damage houses, trees, and so on, and endanger people). (The detailed hurricane stories protocol is available from the second author (T.K.B.).)

Hurricane stories and NSST protocols were coded independently by two separate teams who had no contact with each other. Hurricane stories were coded on the basis of six scales reflecting children’s knowledge: Hurricane Effects, which especially assessed knowledge with respect to water and wind; Property Damage; Effects on People; Effects on Pets; Vocabulary; and Scientific Knowledge. Coders evaluated children’s videotaped responses using 69 statements about these effects (e.g. “trees fell on houses/cars”). A score of 2 indicated clear demonstration of understanding of the item, 1 some understanding, and 0 no demonstration of understanding. Obtained ranges for the six scales were as follows: Hurricane Effects, 1–9; Property Damage, 1–7; Effects on People, 0–14; Effects on Pets, 0–10; Vocabulary, 0–4; Scientific Knowledge, 0–15. In addition, a single, global rating of story coherence (5-point scale) was applied across the children’s story completions and was used as a control variable in multivariate analyses for narrative ability as well as shared method variance (see below). Independent inter-rater agreement among five coders on the six knowledge scales and coherence scale was evaluated as percentages, with an obtained mean of 81 percent.

Assessments of teacher well-being

Teacher well-being was assessed with two measures, a brief non-standardized survey of their exposure to the hurricanes, and the Hopkins Symptom Checklist (HSC; Winokur et al., 1984).

The survey of exposure to the hurricanes asked teachers about most of the same conditions as the parent survey of their children’s exposure (was in a community directly hit, had to evacuate, stayed in a shelter, was injured, someone close was injured, someone close died or was missing, lost property, someone close lost property). Two composite variables were created from these survey items: a composite of personal experience items (the first 5 items listed above) and a composite of loss (someone close died or was missing, loss of property). Of 15 teachers, 14 reported that they lived in communities directly hit by one of the hurricanes (one teacher’s community was hit by both), and 12 of 15 evacuated as a result. Due to lack of variance in the personal experience composite, it was not used in analyses. Three of the 15 teachers reported no losses from the hurricanes (one teacher did not provide this information, leaving 14 teachers used in this analysis). Because only one teacher reported someone close as dead or missing, this item was removed from this variable, to leave a variable for loss of property only, of the teacher and/or someone close to the teacher. The obtained range for this variable was 0–2.

The HSC consists of 25 items measuring anxiety and depression to which the respondent replies using a 4-point scale. It has a history of validation as a screener of psychological distress (Winokur et al., 1984) including extensive cross-cultural validation (see Harvard Program in Refugee Trauma, http://hprt-cambridge.org). The alpha coefficient obtained for the present study was .97.

Data analysis

Bi-variate analyses present descriptive data concerning children’s and teachers’ exposure to the hurricanes. Hierarchical multiple regressions are used to examine the three central study hypotheses. To obtain normal distributions of the NSST variables, each of the scales was reverse-scored and transformed (square root transformation of the Exploration scale, Logarithmic transformations of the Attachment and Caregiving scales). Of the six hurricane story variables, square root transformations were performed for the Effects on Pets, Effects on Property, and Vocabulary variables to obtain normality. The teacher-report HSC was also transformed via square root to obtain normality.

Results

Bi-variate analyses

Zero-order, bi-variate correlations of the main variables of interest are presented in Table 2. The three NSST scales were intercorrelated, as was expected: Attachment correlated with Exploration (r = .44) and Caregiving (r = .65), and Exploration with Caregiving (r = .56).

Table 2.

Zero-order bi-variate correlations for the main variables of interest.

	1	2	3	4	5	6	7	8	9	10
1. NSST Exploration
2. NSST Attachment	.44***
3. NSST Caregiving	.56***	.65***
4. HSC^a	Ns	Ns	ns
5. Teacher Property Loss^b	−.28*^c	Ns	−.25*	ns^d
6. Hurricane Effects	Ns	.28*	ns	−.27*	ns
7. Property Damage	.35**	.28*	.35**	ns	ns	.40***
8. Effects on People	.42***	.50***	.55***	−.32**	ns	.38**	.49***
9. Effects on Pets	.30*	.32**	ns	ns	ns	.52***	ns	.33**
10. Vocabulary	ns	ns	ns	−.37**	ns	.33**	.36**	.34**	ns
11. Scientific Knowledge	.29*	.26*	.39***	−.37**	ns	.41***	.60***	.51***	ns	.50***

NSST: Narrative Story-Stem Task; HSC: Hopkins Symptom Checklist.

n = 15.

n = 14.

NSST variables were reverse-scored to improve normality. To facilitate interpretation, the signs for findings with NSST variables are reversed to reflect their original scoring.

n = 13 due to missing data for two teachers.

p ≤ .05; **p ≤ .01; ***p ≤ .001 (two-tailed).

Among associations with the Hurricane Stories, the three NSST variables were correlated most strongly with the variable Effects on People, and not at all with Vocabulary. Of the two indicators of teacher well-being, only Loss of Property was correlated with NSST variables (Exploration and Caregiving); greater loss of property was associated with lower NSST ratings. The HSC was correlated moderately and negatively with four of six Hurricane Stories variables. Greater levels of teacher distress were associated with lower levels of children’s demonstrations of knowledge of hurricane effects.

Other statistically significant bi-variate associations were obtained for some demographic variables with NSST variables. Girls enacted significantly more attachment behavior than boys (t(62) = 2.57, p = .01). No gender differences were observed in the ratings of the other two NSST variables. No gender effects were observed for the hurricane stories variables, although girls were marginally more likely to demonstrate knowledge of Effects on People (t(62) = 1.86, p < .1). Children’s age (older) and number of children in the classroom (fewer) were also associated with higher ratings on NSST variables. In view of the large amount of missing parent-report data for family income, socio-economic status was assessed at the level of the classroom, using percent of children in the classroom receiving free or reduced school lunch. This variable was also significantly correlated, negatively, with NSST ratings.

Hypothesis 1

Individual t-tests were performed for each of the five conditions of children’s exposure to the hurricanes to examine mean differences among NSST variables (n = 57). No differences were found for any of the conditions of children’s exposure to the hurricanes and NSST variables. Zero-order correlations of the composite variable summing these conditions also yielded no statistically significant associations with NSST variables. The variable representing parents’ beliefs that theirs or their children’s lives were threatened was also unrelated to NSST variables.

Multi-variate analyses

Hierarchical regressions were performed to examine Hypotheses 2 and 3, controlling for children’s age, gender, percent of children receiving free or reduced school lunches in the classroom, number of children in the classroom, and the rating of coherence for the Hurricane Stories, as a control for cognitive capacities for narrative organization/ability as well as shared method variance between Hurricane Stories and NSST (the possibility that correlations may be influenced by similarity of method).

Hypothesis 2

Of the six scales reflecting children’s knowledge of hurricanes, only one, Effects on People, was associated, positively, with NSST variables. Table 3 presents these data.

Table 3.

Regression of NSST variables on hurricane stories, effects on people (n = 64).

Dependent variable	β (t value)	ƒ²	R² adjusted
NSST Attachment	.43 (3.16**)	.15	.28***
NSST Exploration	.32 (2.15*)	.06	.13*
NSST Caregiving	.47 (3.37***)	.18	.27***

NSST: Narrative Story-Stem Task.

p < .05; **p < .01; ***p < .001.

All three NSST variables were predicted by Effects on People, with stronger effects, in the medium range, for Caregiving and Attachment.

To illustrate how children represented hurricane effects on people, the story response of one child, a 9-year-old girl is presented. The interview question to which the child provided this response (coded as “clear demonstration”) began by asking the child to imagine that two children were talking, one was from “far away up north” and did not know anything about hurricanes. (This response also demonstrates understanding of how the hurricane damaged property, which was unrelated to NSST responses.) This child asks the other, “I heard when you were younger, Hurricane Katrina came. What happened then?”

Well, Hurricane Katrina messed up lots of my home, and it took things, and we had to go to the shelter, at M. High School. And I guess it messed up our skirt [note: presumably a reference to a mobile home “skirt”] and it fell off. And our cousin, they had to move into our house because their house was blown off because of Katrina. And our house got lots of damage, our radios, and our fence, and our skirt, all of that got damaged, and there was like a big commotion, and everybody was so scared and lots of people in New Orleans, they had floods; and people went to go get some charity, and FEMA they gave them houses … They’re trying to build back Katrina … trying to look into the future and don’t try to look in back. They’re trying to think about good things and not bad things about Katrina, and some of the people lost lots of their … it mostly killed lots of people and kids, and some of the kids were lost and they could not find their parents.

Hypothesis 3

Of the two measures of teacher well-being, the HSC was unrelated to NSST representations. The index of teacher-reported property loss, however, was significantly associated with all three NSST variables, as seen in Table 4. Effect sizes were small. Given the circumstances where 60 children were in 15 different classrooms, re-running the analysis using teacher (or classroom) as a fixed effect to address potential effects from clustering of the data was unsuccessful due to insufficient power. These results, therefore, should be regarded with caution.

Table 4.

Regression of NSST variables on teacher-reported loss of property (n = 60).

Dependent variable	β (t value)	ƒ²	R² adjusted
NSST Attachment	−.31 (−2.16*)	.07	.18**
NSST Exploration	−.36 (−2.48*)	.10	.14*
NSST Caregiving	−.33 (−2.34*)	.08	.19**

NSST: Narrative Story-Stem Task.

p < .05; **p < .01.

Discussion

This study investigated whether various indices of children’s exposure to the hurricanes that struck south Louisiana in 2005 would be associated with their narrative representations of attachment-related qualities. The main study questions reflect a departure from most NSST research to date in that they explore potential influences from a major environmental catastrophe on children’s internal representations of relationship qualities.

With respect to covariates, two variables, the number of children in the classroom and the school-level index of socio-economic status (percent of children receiving free or reduced school lunch), were significantly associated with narrative measures and thus included as control variables. These two variables were themselves correlated (r = .25, p < .05), suggesting that they are both community-level indices of socio-economic status. Several previous studies have found associations of family-level socio-economic status and children’s narrative representations (e.g. Goodman and Aber, 2010; Page and Bretherton, 2001). The fact that socio-economic status can be measured with community-level indices may just be an artifact of the socio-economic homogeneity of communities, and thus actually be proxies for family-level processes. However, the findings raise an intriguing question as to whether community circumstances with respect to socio-economic status may be influential in children’s organization of attachment-related representational models. The educational and developmental literatures provide evidence that the community-level effect of poverty and of classroom size present independent risks for children’s academic and social development (Sylva, 1994). The benefits to children of smaller classroom sizes appear to operate to a great extent through smaller child–teacher ratios, which would likely involve more attention and individual involvement with teachers. It is therefore plausible that such effects could be detected via narrative story-stem assessments. This should be an area for future study.

With respect to the study’s three hypotheses, contrary to expectations, Hypothesis 1 was not supported. It appears that children’s exposure to the hurricanes of 2005, as reported by their parents approximately 7 months later, had no influence on their representations of child–parent interactions with the NSST. It may be that other relational variables, particularly the parent’s own reactions to the storms and capacities to provide care for the child, influenced this process. The quality of parental support for their children, not measured in this study, may be a stable characteristic and not strongly influenced by external circumstances. Related to this, it may also be that no matter how frightening or disruptive children’s exposure to the hurricanes was, internal representations of caregiving, attachment, and exploration as measured on the NSST were not significantly affected because they tend to be stable and enduring characteristics.

The finding related to Hypothesis 2 that only children’s representations of hurricane effects on people were associated with NSST representations of attachment, exploration, and caregiving may provide further information about the composition of NSST representations. That NSST representations were not associated with technical or scientific knowledge about hurricanes, or children’s representations of property damage, but were associated with children’s enactments of people’s vulnerability to the storms suggests that children’s representational models in part function to enable children to attend to the experience of others. Previous research has shown that children’s NSST representations are significantly associated with behavior with peers and teachers in preschool settings (Gullón-Rivera, 2013; Page and Bretherton, 2001; Page et al., 2011; Rydell et al., 2005; Vu and Howes, 2012), suggesting that representational models play important roles in relationships in social contexts outside the home.

Given that the hurricane stories protocol was modeled on the NSST protocol, a high degree of concordance in the children’s narratives might be expected (although in fact the presented story scenarios were structured very differently). That the actual concordance between the protocols was limited, however, to articulations of distress among people, on one hand, and ratings of attachment, exploration, and caregiving, on the other, indicates that the two protocols shared only a relatively precise domain of overlap, the nature of which could be interpreted to correspond to affective engagement in a play narrative task and the capacity to communicate this. The observed linkages, therefore, between attachment-related representations and articulation of distress experienced by others suggest that a core component of children’s representational models is the capacity for empathy for the experience of others. This point has been made in the attachment research literature based on observational studies demonstrating the predictive power of attachment security on children’s capacities for prosocial and empathic responsiveness to peers later in childhood (Sroufe, 2005). The present study extends this research to the representational realm in showing linkages between an attachment-based assessment and an assessment of children’s perceptions of the effects of a disaster on other people.

These results also suggest that children’s capacities to comprehend and verbally express understanding of an extremely frightening event vary positively with their social–emotional well-being, as assessed with the NSST. This would be consistent with theory indicating that such capacities are typically associated with children’s experience of mastery in coping with extreme distress. In the face of such distress, children who manage and regulate their reactions and do not become overwhelmed are more likely to possess understanding of the event and the ability to articulate this (Salmon and Bryant, 2002). Conversely, the absence of articulation of understanding of hurricane effects on people may indicate an avoidant process, and its association with relatively poorer NSST representations of attachment-related interactions would also be consistent with expectations.

An important corollary to this finding is that new narrative protocols should be adapted to reflect specific frightening circumstances to which children have been exposed. Researchers have already adapted story-stem protocols to reflect specific family or child characteristics (see, for example, Beresford et al., 2007), and there is a history in the clinical literature of using play scenarios of frightening circumstances such as a hospital visit to assess children’s perceptions and provide an avenue for them to communicate anxieties (Ellerton et al., 1985). Our study took this idea a step further in attempting to create representations of the circumstances of a major natural disaster to which children had been exposed, albeit in varying degrees. Following this approach, narrative protocols could be created to reflect other important social contexts in need of study from children’s perspectives, such as dangerous neighborhoods and severe injury.

With respect to our third hypothesis, teachers’ psychological distress as measured with a standardized assessment had no association with children’s narrative representations. At the same time, teachers’ psychological distress, as measured with the Hopkins inventory, was significantly inversely correlated in bi-variate analyses with children’s responses to the hurricane stories, specifically their demonstrations of general hurricane effects, effect on people, vocabulary, and scientific knowledge. This indicates that teachers’ emotional states may have influenced children’s knowledge and articulation of the circumstances associated with the disasters, possibly through limitations in teachers’ attention to this in the classroom.

While the Hopkins inventory was not associated with NSST representation, teachers’ reports of loss of property in the storms was. These intriguing findings open up several considerations, although they should be regarded with caution, due to the small sample size (only 14 teachers were involved) and modest effect sizes. It may be that teachers’ reports of loss of property captured a dimension of preoccupation and perhaps detachment or distress not captured in the standardized measure of psychological distress. Teachers were reported anecdotally by our project staff to be very preoccupied even while in the classroom with phone calls to insurance adjustors and other storm-related concerns. It may also be that teachers directly communicated to their students information about their property losses, perhaps thereby intensifying some children’s experience of instability and uncertainty. Furthermore, it is likely that the measure of teachers’ loss of property reflected not only the teachers’ personal experience but was also representative of broader community effects of the hurricanes, including of course the disruptions in school routines. These findings may be reflections, therefore, of broader community-level influences on children’s portrayals of attachment-related representations. At the same time, however, it is plausible that these findings provide limited evidence that the psychological well-being and/or availability of teachers may influence children’s representational models of caregiving. As such, they would be consistent with established research linking more developmentally sensitive classroom practices with lower stress levels in children (Burts et al., 1992), and with other recent studies that have found linkages between teacher–child relationship qualities and children’s representational models of caregiving (Vu and Howes, 2012). As noted, the findings presented here should be regarded with caution and are in need of replication.

Limitations

The study has several limitations. The measures of children’s exposure to the hurricanes, as reported by parents, and the ratings of the hurricane stories protocol are not standardized and hence are open to criticism concerning their validity. Further research particularly with the rating scheme used for the hurricane stories is needed. While some classroom-level effects were found, it is possible that the study did not measure other teacher/classroom-level effects which may have influenced these effects. Again, the small sample sizes, especially for teachers, underscore the need for replication.

Conclusion

The circumstances of this classroom-based study of children who were exposed, in varying degrees, to the hurricanes of 2005 provided an opportunity to study potential influences on narrative enactments from a new perspective. To add to our knowledge of the composition of children’s narratives, more study of potential external influences is needed. This study is also unique in that its circumstances also allowed for the development of a distinct though similar narrative protocol with its own coding system. Associations across these protocols provide some initial evidence that representational capacities for empathic understanding of the distress of others can be meaningfully assessed at the representational level. Such capacities may have important implications for children’s broader personal adjustment and social functioning.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the National Science Foundation, Grant # 0555387, and the Louisiana Board of Regents.

References

American Psychiatric Association (2000) Diagnostic and Statistical Manual of Mental Disorders. 4th ed., text rev. Washington, DC: American Psychiatric Association.

Beresford

Robinson

Holmberg

. (2007) Story stem responses of preschoolers with mood disturbances. Attachment & Human Development 9: 255–270.

Bettmann

Lundahl

(2007) Tell me a story: a review of narrative assessments for preschoolers. Child & Adolescent Social Work Journal 24: 455–475.

Bowlby

(1973) Attachment and Loss: Separation, vol. 2. New York: Basic Books.

Bowlby

(1982) Attachment and Loss: Attachment, vol. 1. New York: Basic Books.

Bowlby

(1988) A Secure Base. New York: Basic Books.

Bretherton

Gullón-Rivera

Page

. (2013) Children’s attachment-related self-worth: a multi-method investigation of postdivorce preschoolers’ relationships with their mothers and peers. Attachment & Human Development 15: 25–49.

Bretherton

Oppenheim

Buchsbaum

. (1990) The MacArthur Story-Stem Battery. Unpublished manual, University of Wisconsin–Madison, Madison, WI.

Bretherton

Ridgeway

Cassidy

(1990) Assessing the internal working models of the attachment relationship: an attachment story completion task for 3-year-olds. In: Greenberg

Cicchetti

Cummings

(eds) Attachment in the Preschool Years: Theory, Research, and Intervention. Chicago, IL: University of Chicago Press, pp. 273–308.

10.

Buchanan

Casbergue

Baumgartner

(2009) Consequences for classroom environments and school personnel: evaluating Katrina’s effect on schools and system response. In: Kilmer

Gil-Rivas

Tedeschi

. (eds) Meeting the Needs of Children, Families, and Communities Post-Disaster: Lessons Learned from Hurricane Katrina and Its Aftermath. Washington, DC: American Psychological Association, pp. 117–139.

11.

Buchsbaum

Toth

Clyman

. (1992) The use of a narrative story stem technique with maltreated children: implications for theory and practice. Development and Psychopathology 4: 603–625.

12.

Burts

Hart

Charlesworth

. (1992) Observed activities and stress behaviors of children in developmentally appropriate and inappropriate kindergarten classrooms. Early Childhood Research Quarterly 7: 297–318.

13.

Buyse

Verschueren

Doumen

(2011) Preschoolers’ attachment to mother and risk for adjustment problems in kindergarten: can teachers make a difference? Social Development 20: 33–50.

14.

Coates

Gaensbauer

(2009) Event trauma in early childhood: symptoms, assessment, and intervention. Child and Adolescent Psychiatric Clinics of North America 18: 611–626.

15.

Cordón

Pipe

Sayfan

. (2004) Memory for traumatic experiences in early childhood. Developmental Review 24: 101–132.

16.

Cummings

Schermerhorn

Keller

. (2008) Parental depressive symptoms, children’s representations of family relationships, and child adjustment. Social Development 17: 278–305.

17.

Dubois-Comtois

Moss

(2008) Beyond the dyad: do family interactions influence children’s attachment representations in middle childhood? Attachment & Human Development 10(4): 415–431.

18.

Ellerton

Caty

Ritchie

(1985) Helping young children master intrusive procedures through play. Children’s Health Care 13: 167–173.

19.

Fivush

Nelson

(2006) Parent–child reminiscing locates the self in the past. British Journal of Developmental Psychology 24: 235–251.

20.

Fivush

Sales

(2006) Coping, attachment, and mother-child narratives of stressful events. Merrill-Palmer Quarterly 52: 125–150.

21.

Gabe

Falk

McCarty

. (2005) Hurricane Katrina: Social-Demographic Characteristics of Impacted Areas. Washington, DC: Congressional Research Service, The Library of Congress.

22.

Goodman

Aber

Berlin

. (1998) The relations between maternal behaviors and urban preschool children’s internal working models of attachment security. Infant Mental Health Journal 19(4): 378–393.

23.

Goodman

Aber

(2010) Predictors of representational aggression in preschool children of low-income urban African-American adolescent mothers. Infant Mental Health Journal 31: 33–57.

24.

Granot

Mayseless

(2001) Attachment security and adjustment to school in middle childhood. International Journal of Behavioral Development 25(6): 530–541.

25.

Graumann

Houston

Lawrimore

. (2006) Hurricane Katrina: A climatological perspective. National Climactic Data Center. National Oceanic and Atmospheric Administration, Asheville, NC, August.

26.

Gullón-Rivera

(2013) Puerto Rican kindergartners’ self-worth as coded from the Attachment Story Completion Task: correlated with other self-evaluation measures and ratings of child behavior toward mothers and peers. Attachment & Human Development 15: 1–23.

27.

Heller

Boris

Hinshaw-Fuselier

. (2006) Reactive attachment disorder in maltreated twins follow-up: from 18 months to 8 years. Attachment & Human Development 8: 63–86.

28.

Hodges

Steele

Hillman

. (2000) Effects of abuse on attachment representations: narrative assessments of abused children. Journal of Child Psychotherapy 26(3): 433–455.

29.

Kent

(2006) 2005 Louisiana Hurricane Impact Atlas, vol. 1. Louisiana Geographic Information Center. Available at: http://lagic.lsu.edu/lgisc/publications/2005/LGISC-PUB-20051116-00_2005_HURRICANE_ATLAS.pdf

30.

Knabb

Brown

Rhome

(2011) Tropical cyclone report: Hurricane Rita. National Oceanic and Atmospheric Administration, National Hurricane Center, Miami, FL, September. Available at: http://www.nhc.noaa.gov/pdf/TCR-AL182005_Rita.pdf

31.

LaGreca

Silverman

Vernberg

. (1996) Symptoms of posttraumatic stress in children after Hurricane Andrew: a prospective study. Journal of Consulting and Clinical Psychology 64: 712–723.

32.

Laible

(2006) Maternal emotional expressiveness and attachment security: links to representations of relationships and social behavior. Merrill-Palmer Quarterly 52: 645–670.

33.

Louisiana Recovery Authority (2006) Initial quarterly report, June. Available at: http://lra.louisiana.gov/assets/docs/searchable/QuarterlyReports/Feb2006Qtreport.pdf

34.

MacFie

Cicchetti

Toth

(2001) The development of dissociation in maltreated preschool-aged children. Development and Psychopathology 13: 233–254.

35.

Main

Kaplan

Cassidy

(1985) Security in infancy, childhood and adulthood: a move to the level of representation. In: Bretherton

Waters

(eds) Growing Points of Attachment Theory and Research (Monographs of the Society for Research in Child Development, 50(1–2, Serial no. 209)). Chicago, IL: University of Chicago Press, pp. 42–66.

36.

Oppenheim

Emde

Warren

(1997) Children’s narrative representations of mothers: their development and associations with child and mother adaptation. Child Development 68(1): 127–138.

37.

Page

(2001) The social meaning of children’s narratives: a review of the attachment-based narrative story stem technique. Child & Adolescent Social Work Journal 18(3): 171–187.

38.

Page

(2007) Coding manual for the narrative story stem technique. Unpublished manual, Louisiana State University, Baton Rouge, LA.

39.

Page

Bretherton

(2001) Mother- and father-child attachment themes in the story completions of pre-schoolers from post-divorce families: do they predict relationships with peers and teachers? Attachment & Human Development 3(1): 1–29.

40.

Page

Boris

Heller

. (2011) Narrative story-stems with high risk six year-olds: differential associations with mother- and teacher-reported psycho-social adjustment. Attachment & Human Development 13: 359–380.

41.

Pianta

(1994) Patterns of relationships between children and kindergarten teachers. Journal of School Psychology 32: 15–31.

42.

Pynoos

Goenjian

Steinberg

(1998) Children and disasters: a developmental approach to posttraumatic stress disorder in children and adolescents. Psychiatry and Clinical Neurosciences 52: 129–138.

43.

Rosenblum

Dayton

Muzik

(2009) Infant social and emotional development: emerging competence in a relational context. In: Zeanah

(ed.) Handbook of Infant Mental Health. 3rd ed. New York: Guilford Press, pp. 80–103.

44.

Rydell

Bohlin

Thorell

(2005) Representations of attachment to parents and shyness as predictors of children’s relationships with teachers and peer competence in preschool. Attachment & Human Development 7: 187–204.

45.

Salmon

Bryant

(2002) Posttraumatic stress disorder in children: the influence of developmental factors. Clinical Psychology Review 22: 163–188.

46.

Schechter

Zygmunt

Coates

. (2007) Caregiver traumatization adversely impacts young children’s mental representations on the MacArthur Story-Stem Battery. Attachment & Human Development 9(3): 187–205.

47.

Scheeringa

Zeanah

(2008) Reconsideration of harm’s way: onsets and comorbidity patterns of disorders in preschool children and their caregivers following Hurricane Katrina. Journal of Clinical Child & Adolescent Psychology 37: 508–518.

48.

Sroufe

(2005) Attachment and development: a prospective, longitudinal study from birth to adulthood. Attachment & Human Development 7: 349–367.

49.

Stern

(2008) The Serge Lebovici distinguished lecture, World Association for Infant Mental Health: the clinical relevance of infancy: a progress report. Infant Mental Health Journal 29: 177–188.

50.

Sylva

(1994) School influences on children’s development. Journal of Child Psychology and Psychiatry 35: 135–170.

51.

Toth

Maughan

Todd Manly

. (2002) The relative efficacy of two interventions in altering maltreated preschool children’s representational models: implications for attachment theory. Development and Psychopathology 14: 877–908.

52.

Howes

(2012) Examining Mexican-heritage children’s representations of relationships with mothers and teachers in preschool. Journal of Research in Childhood Education 26: 91–106.

53.

Warren

Emde

Sroufe

(2000) Internal representations: predicting anxiety from children’s play narratives. Journal of the American Academy of Child & Adolescent Psychiatry 39(1): 100–107.

54.

Watamura

Phillips

Morrissey (2011) Double jeopardy: poorer social-emotional outcomes for children in the NICHD SECCYD experiencing home and child-care environments that confer risk. Child Development 82: 48–65.

55.

Winokur

Rickels

. (1984) Symptoms of emotional distress in a family planning service: stability over a four-week period. The British Journal of Psychiatry 144: 395–399.

56.

Woolgar

(1999) Projective doll play methodologies for preschool children. Child Psychology & Psychiatry Review 4: 126–134.

57.

Zahn-Waxler

Cole

Richardson

. (1994) Social problem solving in disruptive preschool children: reactions to hypothetical situations of conflict and distress. Merrill-Palmer Quarterly 40: 98–119.