Attachment Stability and Emotional and Behavioral Regulation

 

From Infancy to Preschool Age

Relations between attachment and child emotional and behavioral regulation were studied longitudinally in a sample of 223 children from urban, low_income families. Attachment in the Strange Situation at 12 and 18 months was scored using the infant classification system and at 24 months was scored using a preschool classification system. Only modest stability was found in attachment whether within or across classification systems, with the percentage of insecure attachments consistently increasing over time. Results indicated both concurrent and predictive associations with indices of child regulation based on observer ratings or maternal report. However, only the 24_month classification predicted maternal report of externalizing and internalizing behavior problems at age 3, with additional variance accounted for by selected measures of child emotional and behavior regulation from the same assessment. Attachment security (B) and atypical attachment classifications (D, A/C, and AD) appear to provide the most consistently useful information about child functioning. Results are discussed in terms of continuity and change from the perspective of developmental psychopathology.

Attachment Stability and Emotional and Behavioral Regulation

From Infancy to Preschool Age

Across the second year of life, children undergo a number of key developmental transitions, including the acquisition of symbolic functioning, the emergence of secondary or self_conscious emotions, new self_regulatory and coping skills, increasing locomotion and exploration of an expanded environment, and the start of the shift from sensorimotor to preoperational thinking (Bates, 1979; Campos et al., 1983; Kopp, 1982, 1992). These expanding abilities prompt higher expectations and demands from caregivers, but also more overt efforts to regulate child behavior. Concomitantly, by the end of this period, emotional and behavioral regulation__the management of emotional arousal and conscious control of behavior__begins what will be a very gradual shift from adult control to increasing child self_control (Cicchetti, Ganiban, & Barnett, 1991; Kopp, 1989). Differences in emotional and behavioral regulation during this period offer an early forecast of self_regulation differences__including problems in emotinal and behavioral regulation__apparent in the preschool and early school years, and appear to have their roots both in early_emerging physiological differences (Gunnar, 1990; Kagan, 1994) and in the quality of interactions between child and important caregivers, perhaps best captured at present in measures of infant attachment (Cassidy, 1994; Thompson, Flood, & Lundquist, 1995).

Distinctions between emotional reactivity and emotional regulation__the experience of arousal versus efforts to manage it__and between other aspects of child temperament (e.g., sociability, activity level) and behavioral regulation or control are often obscure. Actual strategies of self_management must often be inferred. To the extent that young children's expression of emotion and patterns of behavior are, in part, a function of both their own and their caregivers' control, they reflect evolving regulatory processes It goes without saying, however, that what is observed in children is also an indicator of physiologically_based differences in emotional arousal, behavioral style, and even cognitive skills and style. For the purpose of this study, different indices of emotional expression and behavioral style will be grouped under the rubric of "regulation," recognizing that observed emotion and behavior are partly products of the child's biological disposition, his or her emerging strategies of self_monitoring and self_management, and caregiver style of support and control.

ATTACHMENT AND EMOTIONAL AND BEHAVIORAL REGULATION

Conceptual arguments for links between attachment and various indices of emotional and behavioral regulation are persuasive: caregivers provide children with "emotional schemes" through modeling and interaction that become the foundation of their relationship, but that also socialize children's emotional experience and expression (Cassidy, 1994; Thompson, 1994). Securely attached children (B) are believed to experience a relationship that is responsive to and supportive of a wide range of emotional needs and signals. By having their anger, fear, and distress ameliorated by their caregiver, negative affect does not become overwhelming and disregulating, but can be regulated increasingly flexibly by the children themselves. Insecurely attached children, on the other hand, are believed to experience a relationship that evolves around more selective responses to emotional needs and signals, with less tolerance in some cases for emotional distress and neediness (avoidant [A] attachments) and less responsiveness in other cases to distress signals unless they have been amplified and/or exaggerated (resistant or ambivalent [C] attachments). Theoretically as a result, insecurely attached children become less flexible in regulating their emotional experiences, with underregulation or overregulation of affect and behavior emerging over time (Sroufe, 1983). In the most extreme case, a key emotional scheme in the relationship may be the disturbing experience of frightened or frightening behavior on the part of the caregiver (disorganized [D] attachments, referred to here as "atypical"), undermining child efforts to self_regulate in a coherent and flexible manner. Correlational data on mother_child interaction and attachment support these links to emotional and behavioral regulation (see Cassidy & Berlin, 1994, Cassidy & Kobak, 1988, and Lyons_Ruth, Bronfman, & Parsons, 1997).

On the other hand, physiological, genetically_based differences in emotional reactivity, inhibition, sociability, and attentional processes (DiLalla et al., 1990; Emde et al., 1992; Robinson, Kagan, Reznick, & Corley, 1992; see also Calkins, 1994; Goldsmith, Bradshaw, & Rieser_Danner, 1986) play a role both in the development of emotional and behavioral regulation and in the evolution of the attachment relationship. Patterns of attachment insecurity (although not insecurity itself) have been linked to behavioral differences in the newborn period (Grossmann et al., 1985; Waters, Vaughn, & Egeland, 1980) and infant emotionality across the first year (Belsky & Rovine, 1987; Frodi & Thompson, 1985). Differences in child temperament no doubt elicit different styles of caregiving, but also shape the child's experience of the same quality and style of care (Goldsmith & Alansky, 1987). At the same time, patterns of parental and child affect regulation probably have genetic as well as experiential linkages across generations.

From both conceptual standpoints__experiential and biological__then, one would anticipate associations between attachment and emotional and behavioral regulation, and associations have certainly been found (Cassidy, 1994; Fagot & Pears, 1996; Gunnar, Mangelsdorf, Larson, & Herstgaard, 1989; Lyons_Ruth, Alpern, & Repacholi, 1993; Thompson et al., 1995). However, relations are typically examined only within a single time frame, usually concurrently or predictively from a single attachment assessment. This is a limited developmental perspective since there is no opportunity to explore changes in attachment and whether these changes are, themselves, associated with emotional and behavioral regulation. The present study considers attachment at three points in time across the transition from infancy to preschool age__12, 18, and 24 months__and its relation to child regulation in the context of both stability and change in attachment classification.

DEVELOPMENTAL CHANGE IN ATTACHMENT

Crittenden (1992) argued that the transition from infancy to the preschool period involves concomitant elaboration and differentiation of internal working models of attachment. Bowlby (1969) and Ainsworth (Ainsworth, Blehar, Waters, & Wall, 1978) clearly enumerated age_related stages in the development of child_mother attachment; the last phase, called a "goal_corrected partnership," connotes a degree of shared responsibility in the relationship that is beyond the capabilities of an 18_month_old, whose cognitive, language, self regulatory, and motoric skills are too limited for this more sophisticated level of interpersonal negotiation.

Especially relevant, developmentally_based changes beginning around this time are (a) the advent of more complex language, which allows more varied attachment negotiations and richer, more direct communication about needs and (b) perspective_taking, which enables children to take into account another person's response, and may encourage them to hide their true feelings or thoughts in the service of maintaining tolerable attachment relationships. One system for coding attachment in children after 20 months of age is Crittenden's (1994) Preschool Assessment of Attachment (PAA). Crittenden (1992) described preschool attachment needs as revolving around the need to communicate rather than simply monitor, to have more control, and to be able to negotiate and plan with their caregivers. The PAA relies on the Strange Situation to elicit attachment behavior, but Crittenden's classification criteria are in keeping with her proposal that the preschool child is more sophisticated than the infant at evolving coping strategies, and that these strategies will both reflect a more complex interpretation of caregiver behavior and will include a greater diversity of attachment_related responses. Preschoolers have at their disposal a greater range of behaviors__including inhibition of and feigned affect and behavior__that can be used to achieve a particular goal; therefore attachment coding must address and focus on the function of behaviors rather than the presence or absence of a specific behavior.

The PAA utilizes one Secure (B) category and two integrated, insecure categories, Defended (A) and Coercive (C). It also includes three atypical classifications, Defended/ Coercive (AC), Anxious Depressed (AD), and Disorganized (D). Children classified in the secure category are open about their needs and feelings and negotiate directly with their caregiver about separation. They also share responsibility with others for emotional regulation. The coercive strategy represents an effort to control caregiver behavior through angry, threatening and/or coy (feigned immaturity or fearfulness) behavior. The defended strategy represents an effort to maintain physical proximity to the caregiver without emotional intimacy through inhibition of negative affect and/or false positive affect. Whereas some children classified as avoidant in infancy would expectably be coded as defended at 24 months (i.e., motoric avoidance develops into psychological avoidance, inhibition of affect), other infants with avoidant attachments might exhibit more coercive, threatening behaviors as preschoolers, precisely because their larger behavioral repertoire and increasing interactive needs result in more diverse coping strategies when their attachment needs are activated. Similarly, some securely attached infants may experiment with coy and coercive behavior as preschoolers and, depending on the responses of their caregiver, maintain a more coercive relationship or shift back to more flexible (i.e., secure) strategies. Crittenden's (1992) expectation is that, during the preschool years, membership in the C category will increase to about one third, so that each (integrated) classification is about equally represented, with notably fewer instances of unintegrated strategies (AC, AD, or D) observed.

The AC strategy represents an effort to cope with changing behavior on the part of the caregiver by using both coercive and defended strategies as the situation demands, whereas the AD classification involves inhibited affect, with sad or flat affect in the presence of an unresponsive caregiver and extreme distress in the absence of that caregiver. Finally, the rarely observed D classification is reserved for children who do not appear to use their behavior strategically either to elicit or respond to caregiver behavior or to soothe themselves. When children are in higher risk environments where greater independence can be, literally, dangerous (e.g., high_crime neighborhoods, unsafe housing, highly stressed parents), Crittenden (1998, personal communication) anticipates somewhat greater shifts to insecurity in the preschool years, particularly to the more extreme classifications of insecurity, whether integrated (A3, A4, C3, C4) or not (AC, AD, D).

In a sample of almost one hundred children from white, working_ and middle_class families, Fagot and Pears (1996) found more children insecure at 30 months using the Crittenden classification system (60%) than at 18 months using the Ainsworth system (49%, ABC only), with the only increase over time in the coercive (C) classification (36% at 18 months). Overall classification stability in that sample was 66%. A comparable sample of German children, assessed at 12 and 21 months, showed similar, but more extreme patterns (Rauh, Ziegenhain, Muller, & Wijnroks, in press). Sixty_two percent were scored insecure at 12 months using the Ainsworth/Main classification systems (ABCD), whereas 83% were scored insecure at 21 months using the Crittenden system (40% were C). Overall stability of security/insecurity was 68%; stability of classification was presumably lower. Children at 21 months were also scored using the Ainsworth/Main criteria, resulting in a decrease in the number of insecure children from 12 (62%) to 21 (48%) months. Based on attachment stability and change, in conjunction with observations of maternal interaction at 3 and 12 months, the authors concluded that both the Ainsworth/Main and the Crittenden systems were valid (and more discriminating than the Ainsworth system [ABC] alone) at 21 months.

ATTACHMENT STABILITY AND CHANGE

Attachment theory and research suggest that both stability and change in attachment patterns over time can be lawful processes associated with relevant parent/child characteristics and behavior (Egeland & Farber, 1984; Fagot & Pears, 1996; Vondra, Shaw, & Hommerding, in press; Wartner, Grossmann, Fremmer_Bombik, & Suess, 1995). Various studies have found that stability of attachment classification, when measured at 12 and 18 months, is positively related to socioeconomic status and negatively related to the presence of risk factors such as marital distress and dissolution, negative maternal personality attributes, and child maltreatment (Barnett, Ganiban, & Cicchetti, 1992; Egeland & Erickson, 1987; Egeland & Farber, 1984; Vondra, Hommerding, & Shaw, 1992). Although much of the research supporting hypotheses about stability (and occasionally change) in attachment security has been based on samples of low_risk, middle_class, mother_child dyads, a handful of studies using low_income and/or high_risk populations indicate, as noted, that changes in attachment security__particularly increases in disorganized or atypical ("D," "A/C," "Unclassifiable") attachments__are more characteristic of disadvantaged, high_risk, and/or maltreating populations than they are of low_risk, middle_class populations (Egeland & Sroufe, 1981; Lyons_Ruth et al., 1991; Spieker & Booth, 1985; Vaughn et al., 1979). In fact, recent longitudinal data indicate only moderate stability (46%_55%) in infancy in general (Belsky, Campbell, Cohn, & Moore, 1996), suggesting that the attachment measure__if not the attachment relationship__is more sensitive to changing circumstances than attachment theory has heretofore implied. Change in mother_child attachment classification clearly poses a challenge to the predictive power of early versus later assessments of attachment, although it does not preclude a meaningful role for attachment history in forecasting children's subsequent functioning (Sroufe, Egeland, & Kreutzer, 1990).

It is much rarer to find change in attachment pattern examined in relation to child emotional and behavioral regulation, due to small samples and/or attachment assessments at only one point in time. Egeland and Farber (1984) found that a shift from secure (B) to resistant (C) classification by 18 months was predicted by ratings of newborn emotional and behavioral irregularity and observed temperamental difficultness in feeding and play scenarios at 6 months. The opposite shift, from resistant to secure classification, was associated with higher developmental testing scores at 9 months (which tend to correlate highly with behavioral regulation during testing). The extent to which earlier attachment provides predictive power above and beyond later attachment has essentially been unexplored. However, Sroufe, Egeland, and Kreutzer (1990) offered preliminary data to support Bowlby's (1973, 1980) contention that developmental history must contribute to adaptation above and beyond contemporaneous circumstances.

The present investigation examines continuity and change in attachment across the second year of life in relation to concurrent and subsequent child emotional and behavioral regulation. The population under investigation is children from urban, low_income families at varying degrees of familial risk. Based on the literature reviewed, it was expected that:

1. Change in attachment would be more common in this low_income, higher stress population than in the white, middle_class population sampled in most longitudinal attachment studies, with an increasing percentage of insecure attachments over time. Change would be greater across scoring systems than within scoring systems.

2. Attachment classification would show concurrent and predictive links to child emotional and behavioral regulation.

3. Patterns of attachment change from infancy (12 and 18 months) to 24 months would help differentiate children in terms of their emotional and behavioral regulation.

4. Attachment at 12 and 18 months would provide some predictive power to child regulatory problems at age 3 over and above attachment at 24 months.

METHODS

SAMPLE

Two birth cohorts of children and their mothers from a longitudinal study of vulnerability and resiliency in childhood (Shaw, Owens, Vondra, Keenan, & Winslow, 1996; Vondra, Shaw, & Kevenides, 1995) provided the data for the current investigation. Two hundred twenty_three urban, low_income mothers (Cohort 1 n=103; Cohort 2 n=120) with infants between the ages of 5 and 11 months were recruited from Pittsburgh offices of the Women, Infant, and Children (WIC) Supplemental Nutrition Program administered by the County Health Department. Inclusion in the WIC Program requires low income (e.g., less than $22,385 for a family of four, less than $14,837 for a single mother and child in 1989, the first year of recruitment). Women who completed brief background questionnaires by phone or at the WIC office and took part in an initial lab visit with their infant at 12 months were included in the longitudinal investigation.

At the time of the first laboratory visit at 12 months infant age, 45% of mothers reported being married or living with a partner, 13% reported being separated or divorced, and 42% reported being single. Almost three quarters of the women (74%) reported having a high school education or less, 88% reported having a family income of less than $1,500 per month, 39% were of minority race (almost exclusively African American), and 20% were teenagers at the time of their first child's birth. Maternal age at the time of recruitment ranged from 16 to 39, with a mean of 25. Of the 223 infants, 54% were male, 45% were firstborn, and 13% were born more than two weeks prematurely. The date of the assessments at 12 and 18 months were corrected for infant gestational age.

PROCEDURES

Mothers were asked to bring their child to the University for observations and assessments at the child ages of 12, 18, and 24 months, and to schedule a home visit when their child was either 15 (Cohort 1) or 18 (Cohort 2) months old. Follow_up contact with mothers was completed when children were between 3 and 4 years by telephone and by mail. Of the 223 infants observed at 12 months, 165 were rated by their mothers at approximately 3 years of age on a behavior problem checklist. Indices of emotional and behavioral regulation collected between 12 and 24 months were used to compare those infants with and without follow_up data. Of 17 measures collected during infancy, only two showed differences, neither of which reached formal statistical significance. Children without follow_up ratings scored somewhat lower on activity/excitability during developmental testing in the home between 15 and 18 months (t[199]= 1.80, p < .10), but were rated as somewhat more fussy/difficult by their mothers at the 18_month lab visit (t[150]= _1.95, p < .10).

Lab assessments began with a free_play period for the child (when mothers completed questionnaires nearby with the examiner), followed by a series of interactive activities (examiner absent from the room), a rest/snack period in another room, and the Strange Situation. Mothers completed questionnaires with the examiner after the Strange Situation, while a babysitter entertained the child. The order of the Strange Situation and Free Play period was counterbalanced at 12 months only (all Cohort 1 children had the Strange Situation last, all Cohort 2 children had the Strange Situation first; no differences in the distribution of attachment classifications were found across cohorts). Each lab visit took approximately two hours to complete and was videotaped from a fixed camera on the wall and/or through a one_way mirror to facilitate behavioral coding.

Home visits began with a free play period and developmental assessment of the child. Following this, mothers completed an interview and questionnaires while a babysitter enter_tained the child. Observations of mother_child interaction throughout the visit were made. Examiners varied from visit to visit, although on occasion the same examiner conducted more than one visit for a given family. Across all pairs of lab and home visits, 87% (range 78_100%) were conducted by different examiners.

MEASURES

Attachment

Attachment security was assessed using the Strange Situation paradigm (Ainsworth & Wittig, 1969). At 12 and 18 months, attachment classification (A=Avoidant, B=Secure, C=Resistant, D=Disorganized) was coded using the Ainsworth (Ainsworth, Blehar, Waters, & Wall, 1978) and Main (Main & Solomon, 1990) criteria. The first author was trained to 100% reliability with A. Sroufe (5 cases), 85% reliability with M. Ward and B. Vaughn (20 cases), and 75% with D. Cicchetti (16 cases). Six graduate student raters, blind to other ratings of the mother and child, were trained to reliability by the first author and tested for interrater agreement using two different sets of attachment assessments, one set from the lab of J. Belsky and a second set from the lab of A. Sroufe. Interrater agreement on major classifications ranged from 80% to 100% with a mean of 83% for the test (non_study) assessments, and averaged 77% (20 cases) with the first author for a random set of study tapes. Tapes were scored by the first author and/or by at least two of the trained raters. In cases of disagreement, a third rater was used.

At 24 months, attachment classification (A=Defended, B=Secure, C=Coercive, A/C or AD=Atypical) was coded using the Preschool Attachment Assessment (PAA, Crittenden, 1994). Four graduate student raters were trained by P. Crittenden to 85% reliability using her system. Two of these raters had rated a subset of the 12 or 18 month tapes. To help ensure independence of ratings, 24_month attachment was always scored either by two independent raters or by a single rater blind to infant attachment classifications. Average interrater agreement on major classifications was 62% (27 cases) for a random set of cases from the study. In cases of disagreement, a third rater was used. Thirty (15%) of the 24_month attachments were scored by P. Crittenden.

Child Emotional and Behavioral Regulation

During each lab visit (12, 18, and 24 months), children were observed with their mothers during a series of four teaching tasks. Activities were selected that were too difficult for children of that age to complete independently, and thus required maternal assistance (e.g., stack a set of plastic rings on a pole at 12 months, work a lock_key toy at 18 months, complete a puzzle at 24 months). In each case, mothers were instructed to work on the task for 3 minutes and then, at a knock, to put away the current task and start the next one. Ratings of child positive and negative affect and task_oriented (attentive, effortful) behavior were completed by two trained raters at 12 and 24 months (blind to attachment classifications), using simple three_point scales (0=none, 1=low, 2=high). Percent exact agreement was 75% for positive affect, 86% for negative affect and 84% for task orientation using a random set of 32 cases from the study. Because emotional and behavioral regulation in toddlerhood is still a joint function of child and caregiver, child affect and persistence can be viewed as a reflection of how effectively the dyadic system is regulating the child.

Global ratings of child emotional and behavior regulation were made from observations of the entire 24_month lab visit using the Early Coping Inventory (Zeitlin & Williamson, 1988). Designed for use with young children between the ages of 4 and 36 months, the inventory assesses the age_ and situation_appropriate effectiveness of a wide range of behaviors considered important to early coping efforts. The ECI consists of 48 items relecting temperament, sensory processing, motor control, and socioemotional factors. Scores on the ECI differentiate young children with mental or physical disabilities from non_disabled peers (Williamson, Zeitlin, & Szczepanski, 1989; Zeitlin & Williamson, 1990). A single trained rater, blind to attachment classification and other emotional and behavioral regulation ratings, completed 36 of the 48 five_point behavior ratings on the basis of behavior throughout the 24_month visit (excluded items concerned sensory and motor handicaps not seen in the sample). Ratings were factor analyzed to create internally consistent global ratings of regulation. A four_factor solution emerged, accounting for 52% of the variance in scores. The first factor, termed "assertive," consisted of 8 items and loaded heaviest on items "Child expresses likes and dislikes," Child expresses a range of feelings," and "Child has an energy level that is forceful and vigorous." The second factor, termed "adaptable," consisted of 9 items and loaded heaviest on items "Child adapts to changes in environment," "Child accepts substitute people or objects when necessary," "Child finds a way of handling a new or difficult situation," and "Child demonstrates ability to self comfort." The third factor, termed "sociable," consisted of 7 items and loaded heaviest on items "Child gives warmth and affection to others," "Child accepts warmth and support from familiar persons," and "Child maintains visual attention to people and objects." The fourth and final factor, termed "competent exploration," consisted of 7 items and loaded heaviest on items "Child initiates exploration of own body or objects using a variety of strategies," "Child demonstrates persistence during activities," and "Child completes self_initiated activity." On a random subset of 12 cases, agreement within 5 points on the summed scales ranged from 83% to 100%, with correlations ranging from .78 to .92.

During the home visit, trained examiners administered the Bayley Scales of Infant Development and its accompanying Infant Behavior Record (IBR, Bayley, 1969). The IBR consists of a series five_ and nine_point rating scales (13 in all) evaluating social and object orientation during standardized developmental testing. Scale scores were entered into a factor analysis to create internally consistent global ratings of emotional and behavioral regulation. A three_factor solution emerged, accounting for 99% of the variance. The first factor, termed "mastery motivated," consisted of 5 items and loaded heaviest on items "Persistence in goal_directed effort" and "tendency to persist in attending to any one object, person, or activity." The second factor, termed "active/reactive," consisted of 3 items and loaded heaviest on items "Amount of gross bodily movement" and "The ease with which a child is stimulated to react in general, excitability." The third and final factor, termed "comfortable," consisted of 4 items and loaded heaviest on items "Tenseness of body" and "Degree of happiness." Interrater reliability was assessed using a small number (N=8) of pilot subjects at the end of training, prior to data collection. Agreement on scores based on these three factors varied tremendously, with good to excellent reliability on the mastery motivation (intraclass correlation = .72) and comfort (intraclass correlation = .92) scores, but no reliability on the activity/reactivity score (intraclass correlation = _.06), due to extremely restricted variability across pilot children on this factor. Examiners making IBR ratings during the home visit were blind to attachment classifications and the other behavioral ratings described above.

Maternal Ratings

At each lab visit, mothers completed the Difficultness scale of the Infant Characteristics Questionnaire (ICQ, Bates, Freeland, & Lounsbury, 1979). This scale had the strongest psychometric characteristics in the validation study of the ICQ, with moderate correlations between both maternal and paternal report and observer ratings on the same measure. Slightly different versions are available for 13_ (nine items) and 24_month (seven items) infants (used at the 12_ and 24_month lab visits, respectively). At the 18_month visit, items from both versions were used. Scores from the measure have been shown to predict behavior problems at preschool age (Bates, Maslin, & Frankel, 1985), and its selection reflected our interest in the early expression and regulation of negative affect as an early indicator of developmental vulnerability.

As part of the 3_year follow_up contact, mothers completed the Child Behavior Checklist for Ages 2 to 3(Achenbach, 1992) by mail. This 100_item questionnaire generates two broad_band factors, externalizing (problems with underregulating affect and behavior) and internalizing behavior problems (problems with overregulating affect and behavior), for both boys and girls. Test_retest reliability is reported to be .87 by the author.

RESULTS

RELATIONS AMONG INDICES OF INFANT REGULATION

As a preliminary step, patterns of association were examined among the 16 indices of emotional and behavioral regulation that were collected in infancy. Within a particular task/ activity, correlations were generally moderate (but highly significant), ranging from an average of .21 (ratings during 12_month teaching tasks) to .55 (maternal ratings of difficultness at 12, 18, and 24 months). Correlations across activities (and time) were notably less consistent, but present in piecemeal fashion. Factor analysis of the 16 scores indicated a three_factor solution, which accounted for 88% of the variance. Factor 1 consisted of all observational ratings at 24 months (on the Early Coping Inventory and during the teaching tasks, rated by independent observers). Factor 2 incorporated the three maternal ratings of difficultness (at 12, 18, and 24 months), and Factor 3 represented examiner ratings of behavior during developmental testing (at 15 or 18 months). The 12_month teaching task ratings did not load on any of these factors and also had the lowest intercorrelations. Individual scores with the most frequent correlations across measures were the "mastery motivation" (IBR) score during developmental testing (the only observer rating associated__negatively__with maternal ratings of difficultness) and the four "coping" scores from the Early Coping Inventory. Given that regulation indices were collected during different activities at different ages, these general results are not surprising.

CLASSIFICATION DISTRIBUTIONS

The distribution of attachment classifications at each age appears in Table 1. As predicted, the percentage of secure (B) attachments consistently declined and the percentage of insecure attachments increased both across time (12, 18, 24 months) and classification system (Ainsworth to Crittenden). Within the insecure category, the percentage of A attachments consistently increased; the percentage of children with C or atypical attachments tended to be low at each age, and no consistent patterns were discernible. The change to a different classification system at 24 months, however, was accompanied by an increase in the percentage of C attachments and a decrease in the percentage of atypical attachments.

_______________________________

Insert Table 1 about here

_______________________________

Attachment stability data appear in Tables 2 and 3. It should be noted that criteria for stability changed between infancy (12 to 18 months), when exact classification match was used, and preschool age (12/18 to 24 months), when at least one classification match was used (a match between either 12 and 24 months or 18 and 24 months). Thus, the criterion for a match across classification systems was more liberal than that within the infant classification system. Perfect matches across all three ages were quite rare: 20 BBB (10% of the sample), 7 AAA (4%), 3 CCC (1.5%) and 1 DD_Atypical (<1%). With this caveat, the overall rates of stability were identical from 12 to 18 months (45% stability) and from 12/18 to 24 months (45% stability).

 _______________________________________

Insert Tables 2 and 3 about here

_________________________________________

Seventy four percent of the sample had the same classification at two or more points in time, with 24% showing stability in the infant system but a change to the preschool system. Twenty six percent of the sample changed classification at each of the three time points. Stability of classification was more common among B and D attachments from 12 to 18 months, but was more common (for those who did not change at each assessment) among A and C attachments across classification systems from 12/18 to 24 months.

DIFFERENCES AMONG THE STABLE GROUPS

Selected children (38% of the sample) with "stable" A, "stable" B, or "stable" C attachments were compared by analysis of variance (only two children had stable atypical attachments, too few to be included in analyses). "Stable" attachment, once again, indicates a classification match between either 12 or 18 months and 24 months. Cases designated as stable B with an A or D classification coded at the second infant assessment were considered anomolous in their combination of security and insecurity and were excluded from analyses. Mean differences between the three stable groups__subsequently tested using Bonferonni post_hoc comparisons__appear in Table 4.

______________________________

Insert Table 4 about here

_______________________________

Observers

No differences were found among the groups on behavior during developmental testing in the home. Child behavioral regulation ratings during mother_child teaching tasks, however, distinguished the children with stable C attachments from those with stable B attachments. ANOVA results indicated that the stable C group was rated as less task oriented at both 12 (F[2,68]=4.40, p < .05) and 24 months (F[2,71] = 13.37, p < .001), but not different in affect. Child behavioral and emotional regulation during the 24_month lab visit showed differences on three of the four (Early Coping Inventory) scores. The C group scored lower than either of the other groups on adaptive behavior (F[2,70] = 19.88, p < .001), and the B group scored higher than the C group on sociability (F[2,67] = 3.61, p < .05) and competent exploration (F[2,69] = 9.46, p < .001). No differences were noted for assertiveness.

Mothers

Mothers reported differences in perceived difficultness among the three stable attachment groups only at 12 months (F[2,72] = 6.48, p < .01). Children in the stable C group were rated as significantly more difficult than children in the stable A group. At age 3, children with a stable secure attachment history scored lower than either the stable A or stable C groups on mothers' report of externalizing behavior problems (B_group T_score mean = 43.6; A_group = 52.5; C_group = 53.0; F[2,65] = 7.62, p < .01). No differences emerged for internalizing behavior problems.

DIFFERENCES ASSOCIATED WITH TRAJECTORIES TOWARD INSECURITY

Enough children changed from stable security in infancy to insecurity at 24 months (BBC or BBA) to contrast children who showed a stable pattern across classification systems with those who changed to insecurity with the preschool classification system at 24 months. Results are reported for the B/C contrast and the B/A contrast separately.

BBB versus BBC versus CCC

Observers. The three groups scored similarly on the Bayley Scales (MDI) during the home visit. During the mother_child teaching tasks at 12 and 24 months, the BBC_attachment change group scored between the stable B and stable C groups. On task orientation at 24 months, all three groups were significantly different from each other, with the stable B group highest and the stable C group lowest (F[2,66] = 13.63, p < .001). Also during the 24_month lab visit, the BBC_change group again scored between the stable B and stable C groups on overall behavioral and emotional regulation, with the stable C group lowest on adaptive behavior (F[2,68] = 12.27, p < .001), sociability (F[2,65] = 4.25, p < .05), and competent exploration (F[2,68] = 13.27, p < .001), significantly lower than the BBC_change group on both adaptive behavior and competent exploration.

Mothers. Consistent differences emerged in terms of mothers' perceptions of child difficultness, with the stable C group scoring highest and the BBC_change group scoring lowest on perceived difficultness at 12 (F[2,68] = 3.25, p < .05), 18 (F[2,45] = 5.72, p < .01), and to some extent 24 months (F[2,67] = 2.93, p < .10). In each case, the stable B group scored in between and was not significantly different from either of the extreme groups. At age 3, the children in the stable B group scored significantly lower than those in the stable C group on mothers' report of externalizing behavior problems, with the BBC_change group scoring in between (BBC_group T_score mean = 49.4; F[2,57] = 6.03, p < .01). No differences emerged for internalizing behavior problems.

BBB versus BBA versus AAA

Observers. As was the case with the stable group comparison, few child behavior differences emerged between B attachment groups and A attachment groups. Observers saw no significant differences among any of the three groups in behavior during developmental testing in the home, or during the mother_child teaching tasks at 12 or 24 months. However, the BBA_attachment change group (X[14] = 110.2) scored significantly higher than the stable A group (X[30] = 101.0) on the Bayley Mental Development Index (F[2,70] = 3.91, p < .05), with the stable B group scoring in between the two extremes (X[29] = 107.5). During the 12_ and 24_month lab visit, the BBA_change group scored midway between the stable A and stable B groups, with means for the two extreme groups only significantly different on competent exploration (F[2,72] = 4.72, p < .05) at 24 months.

Mothers. Mothers reported no differences among these three groups in terms of perceived difficultness, whether at 12, 18, or 24 months. At age 3, the children in the stable B group scored significantly lower than boththose in the stable A group and in the BBA_change group on mothers' report of externalizing behavior problems (BBA_group T_score mean = 52.1; F[2,64] = 8.42, p < .001). No differences emerged for internalizing behavior problems.

Other Attachment Change Groups

Contrast analyses were run incorporating other, relatively small, subsets of children with specific kinds of attachment change from 12 to 24 months (e.g., A_>C, D_>C, C_>A). Few consistent patterns of differences emerged in these contrasts, whether the result of power problems, measurement error, or simply the absence of differences. The numbers of children who changed from stable insecurity to security at 24 months were too small for statistical analysis (AAB=2, CCB=0, CAB=2, DDB=4).

CORRELATES OF ATTACHMENT TYPE

To gain a clearer picture of how attachment classification from 12 to 24 months relates to behavioral and emotional regulation, four continuous scores were created representing frequency of each attachment classification, A, B, C, and D/atypical. These scores represent the number of times, across three assessments (range = 0_3), that a child's attachment was scored in each classification. Thus, a child whose attachment was scored a B at 12 and 24 months, and an A at 18 months, would receive a 1 on the A score, a 2 on the B score, and a 0 on the C and D scores. These scores were then correlated with each of the indices of emotional and behavioral regulation used as dependent measures. Results appear in Table 5.

________________________________

Insert Table 5 about here

________________________________

In general, correlations were primarily found for the B and D scores. Frequency of B classification from 12 to 24 months was modestly associated with greater concurrent behavioral and emotional regulation and fewer externalizing behavior problems at age 3. Frequency of D/atypical classification was modestly associated with poorer regulation and more externalizing behavior problems. Frequency of C classification was associated almost exclusively with poorer regulation during the 24_month lab visit, and frequency of A classification was essentially unrelated to measures of behavioral and emotional regulation.

INFANT (12/18 MONTHS) VERSUS PRESCHOOL (24 MONTHS) PREDICTION

To examine whether infant (12_18 month) attachment classification provided any unique predictive power to age 3 behavior problems above and beyond 24_month attachment, two regression analyses were conducted. Externalizing and internalizing behavior problems reported by mothers were each predicted in a hierarchical regression by scores for 24_month attachment and for combined 12_ and 18_month attachment. Results appear in Table 6. In both cases, the equation using 24_month scores only was significant and accounted for between 7% and 9% of the variance in behavior problems (Externalizing T_score: F[3,161] = 6.86, p < .001, R2 = .11; Internalizing T_score: F[3,161] = 4.33, p < .01, R2 = .08). Externalizing problems were significantly predicted by codes for each insecure classification (A, C, atypical); internalizing problems were predicted only by the atypical classification. Therefore, an atypical attachment at 24 months was associated with greater internalizing and externalizing problems reported by mother at age 3, whereas a defended or coercive relationship was associated only with higher reported externalizing problems.

________________________________

Insert Table 6 about here

________________________________

Adding in 12_ and 18_month attachment classification gave no additional predictive power for behavior problems at 3. None of the insecurity codes in infancy contributed significantly to the regression equation once 24_month attachment was entered, resulting in an overall equation of less significance. Thus, 24_month (Crittenden) attachment significantly predicted mothers' report of both externalizing and internalizing behavior problems at 3, and infant (Ainsworth) attachment history__often different__did not predict above and beyond that. Univariate ANOVAs using attachment classification at each assessment separately indicated that associations with behavior problem scores were confined exclusively to 24_month classifications (mean Externalizing T_scores: B= 45, A= 53, C= 51, Atypical= 53, F[3,161] = 6.86, p < .001; mean Internalizing T_scores: B= 48, A= 50, C= 50, Atypical= 60, F[3,161] = 4.33, p < .01). Mean differences in later problems with regulating behavior were related only to attachment as it was scored at 24 months, not scores at 12 or 18 months, supporting the findings of the regression equation.

24_MONTH REGULATION VERSUS ATTACHMENT PREDICTION

To examine whether 24_month child emotional and behavioral regulation scores provided any unique predictive power to age 3 behavior problems above and beyond 24_month attachment, two regression analyses were conducted. Externalizing and internalizing behavior problems reported by mothers were each predicted in a hierarchical regression by scores for 24_month attachment (dummy coded) and for 24_month regulation scores. Zero_order correlations indicated that only two of the four regulation scores__sociable (r[162] = _.28, p < .001) and competent exploration (r[164] = _.19, p < .05)__significantly correlated with externalizing problems at age 3. Only competent exploration significantly correlated with internalizing problems (r[164] = _.17, p < .05). Therefore, only these two regulation scores were used in the regressions. Results appear in Table 7. In both cases, adding in the two regulation scores improved prediction of later behavior problem ratings, with equations now accounting for between 11% and 17% of the variance (Externalizing T_score: F[5,146] = 5.90, p < .001, R2 = .17; Internalizing T_score: F[5,146] = 3.45, p < .01, R2 = .11). Unique prediction to Externalizing T_scores was provided by 24_month A classification, C classification, D classification, and sociability. Unique prediction to Internalizing T_scores was provided by 24_month D classification and competent exploration.

________________________________

Insert Table 7 about here

________________________________

In summary, the predictive power of 24_month attachment was enhanced by adding information about the child's ratings on emotional and behavioral regulation over the course of the same lab visit, but was not improved by information about the child's prior attachment history. Insecure patterns of attachment, as they were scored using a preschool attachment system, predicted maternal report of child problems in behavioral regulation at age 3. Independently scored indices of the child's overall effectiveness in regulating him or herself during the same visit provided complementary information, whereas attachment in infancy did not.

DISCUSSION

ATTACHMENT STABILITY AND CHANGE

Results indicate modest stability in attachment from 12 to 18 months and from 12/18 to 24 months, using a sample of children from urban, low_income families at varying degrees of familial risk for developmental problems and delays. Rates of 45% within the infant coding system (exact matches from 12 to 18 months) and 45% across coding systems (at least one match from 12/18 to 24 months) are lower than those for smaller, primarily white, middle_class infant samples (75% on average, see Belsky et al., 1996) although at least one, larger and more recent investigation documented rates of between 46% and 55% (Belsky et al., 1996). They are, however, more comparable with rates (30% to 64%) found in higher risk and/or maltreating infant samples (Barnett et al., 1992; Egeland & Farber, 1984; Lyons_Ruth et al., 1991). Even higher rates of stability from infancy to preschool (Howes & Hamilton, 1992, 72%) and school age (Main & Cassidy, 1988, 84%; Wartner et al., 1994, 82%) in white, middle_class samples are also balanced by lower rates in unpublished studies using infancy/preschool samples at medical risk (Beckwith & Rodning, 1991, 42%; Goldberg et al., 1997, 42%), and with a study of white, primarily middle_class children using the Ainsworth coding system at 18 months and the Crittenden system at 30 months (Fagot & Pears, 1996, 66%). Stability is, of course, higher (69% from 12/18 to 24 months) when security/insecurity is the focus, rather than classification. This rate almost exactly replicates the stability of security/insecurity (68%) in the Rauh et al. (in press) study examining attachment change from 12 months, using the Ainsworth/Main system, to 21 months, using the Crittenden system, in a sample of middle_class and lower middle_class infants in Germany. Stability of attachment, in other words, though discernible, is far from universal, particularly among samples of children at some developmental risk. Less stability and supportiveness of family life and parental functioning are hallmarks of developmental risk for children and have clearly been associated with discontinuity in attachment classification over time (see Lamb, Thompson, Gardner, & Charnov, 1985). The change in coding systems from 12/18 to 24 months encompasses a change in emphasis and in criteria for making attachment classifications. It is important, then, to consider its contribution to the instability found. At the same time, it is imortant to recognize that stability was not the norm for infants in this sample, even within a single coding scheme.

Although overall stability rates were roughly comparable (given different stability criteria) from 12 to 18 months and from 12/18 to 24 months, patterns were somewhat different. Consistent patterns over time were a decrease in the number of secure attachments, as expected, and an increase in the number of A (Avoidant, Defended) attachments. As Crittenden (1992) predicted, however, approximately one third of the 24_month attachments were scored as secure (B=26%), one third as defended (A=33%), and one third as coercive (C=37%). An increase in the percentage of C attachments from infancy to the preschool period was predicted by Crittenden (1992), is reflected in her PAA scoring system (1994), and has been found by other investigators using the PAA (Fagot & Pears, 1996; Rauh et al., in press; Teti, Gelfand, Messinger, & Isabella, 1995). Crittenden (1992) argued that the discovery by preschool_age children of coy behavior as a mechanism to handle angry feelings becomes integrated into the coercive strategy, particularly for children with resistant attachments during infancy. But because it is explored to a greater or lesser extent by all preschool children, the result is what might be considered an overrepresentation of coercive preschool attachment classifications. In any case, the increase in insecure attachment classifications from 12 to 18 to 24 months is consistent with other data on attachment among samples at developmental risk.

RELATIONS TO EMOTIONAL AND BEHAVIORAL REGULATION

As predicted, attachment classification was associated with both concurrent and subsequent ratings of child emotional and behavioral regulation. This was most apparent for 24_month attachment classification (Crittenden system) and for children with stable attachment classifications from 12/18 to 24 months. In general, the strongest and most frequent associations were related to the secure and atypical classifications, although each 24_month classification provided unique information about mothers' report of behavior problems __problems in regulation__at age 3. Security across all assessments was associated with better emotional and behavioral regulation rated concurrently by observers, and with fewer reported externalizing behavior problems at 3. The converse was true for atypicality (D at 12/18 months, AC or AD at 24 months). Type C classifications, particularly at 24 months, were associated with poorer regulation, but almost exclusively among the (concurrent) 24_month measures. Interestingly, children with Type A attachments were generally indistinguishable from the others, with the exception of age 3 behavior problems, when 24_month A classification (as well as C and atypical classifications) predicted more externalizing problems reported by mother, regardless of prior (infant) attachment classification.

Specific patterns of change provided less information about child emotional and behavioral regulation than did information about most frequently observed attachment classification, or even just 24_month attachment classification. Repetitive attachment themes and most recent attachment classification offered the most insights about a child's effectiveness in regulating behavior and affect, including problems in their regulation. This makes both methodological and developmental sense. On the one hand, patterns of behavior observed repeatedly across time are probably measured with less error and/or capture more generalizable styles of functioning than those that are specific to a single point of assessment. On the other hand, development is change, so measures of comparable psychometric strength capturing behavior patterns later in development are likely to be more "up to date" about development and its ecology than those capturing patterns earlier in development.

Generally speaking, children who changed from a secure infant classification to an insecure 24_month classification scored between the stable secure and stable insecure groups, but did not score significantly different from either. This pattern supports the notion that attachment history does, in fact, provide meaningful information about child functioning (Sroufe et al., 1990). But the relative lack of significant differences for the change group also highlights the modest, or classification_specific, effect sizes of the group differences. By age 3, infant (12/18 month) attachment classification no longer provided additional predictive information, at least to maternally reported behavior problems. Any insecure 24_month classification predicted externalizing problems, and the atypical classification significantly predicted internalizing problems, but prior attachment history (12/18 months) added no predictive power and, in fact, was not associated with maternal report of behavior problems. In contrast, independent observer ratings of overall child emotional and behavior regulation (lower sociability, less competent exploration) during the 24_month lab visit provided unique predictive power, both to externalizing and internalizing scores. Thus, knowing a child's attachment at 24 months__using a preschool attachment classificatory system__and knowing a child's emotional and behavior regulation during the same assessment, independently and cumulatively predicted behavior problems one to two years later. Knowing a child's infant attachment history did not.

Results contribute to the growing body of literature on relations between attachment and emotional and behavioral regulation. Although relations have been found for children from all backgrounds, especially when regulation is assessed during interactions with caregivers and peers, (Ainsworth et al., 1978; Howes, Hamilton, & Matheson, 1994; Sroufe, 1983; see Belsky & Isabella, 1988), prediction from infant attachment to behavior problems__problems in regulatory functioning__in the preschool years has been less robust (Bates & Bayles, 1988; Fagot & Kavanagh, 1990), and seems to be restricted somewhat to samples already at familial/developmental risk and to the disorganized infant attachment classification (Lyons_Ruth, Easterbrooks, & Cibelli, 1997; Shaw et al., 1998). More robust are associations between preschool and early school age measures of attachment (the "controlling" classification in the Cassidy, Marvin, et al. [1991] preschool scoring system and in the Main, Kaplan, & Cassidy [1985] age 6 scoring system) and concurrent externalizing behavior problems (Greenberg, Speltz, DeKleyn, & Endriga, 1991; Solomon, George, & DeJong, 1995). Thus, associations tend to be strongest when assessed concurrently and/or among samples at risk, using atypical attachment classifications.

But prediction from early preschool age to later behavior problems has now been demonstrated in two very different samples of children using the Preschool Assessment of Attachment (Crittenden, 1994). In the present investigation, prediction from 24 months to externalizing problems reported by mothers at age 3 was found foreach attachment classification, and prediction to internalizing problems was found for atypical classifications, in a sample of urban, low_income children. Using a sample of 90 mostly white, middle_class children at 30 months, Fagot and Pears (1996) found concurrent relations to maternal ratings of aggressive behavior and prediction to teacher ratings of both internalizing and external_izing behavior problems (T_Scores at or above 60) at age 7 for the coercive classification.

Two common themes emerge from these findings. First, attachment classification is related to affective and behavioral regulation in social contexts, with concurrent relations most easily attained. Second, prediction from attachment to ratings of behavior problems relies strongly on disorganized classifications in infancy and controlling, coercive, or atypical classifications in the preschool period. A growing body of evidence relates these classifications to familial risk: the disorganized (or A/C) classification in infancy (Carlson, Cicchetti, Barnett, & Braunwald, 1989; Cicchetti & Barnett, 1991; Lyons_Ruth et al., 1991; O'Connor, Sigman, & Brill, 1987; Spieker & Booth, 1988; Teti et al., 1995) the coercive and atypical (Crittenden, 1988; Radke_Yarrow, Cummings, Kuczynski, & Chapman, 1985; Teti et al., 1995) classifications during the preschool years. Whether these attachment classifications are a mediator between family risk and child behavior problems, or whether they are simply a marker for family risk associated with child behavior problems, remains to be examined (Greenberg, Speltz, & DeKlyen, 1993).

There are theoretical and empirical connections between attachment disorganization in infancy and the "controlling" classification in the preschool and school years (Main & Cassidy, 1988; Wartner et al., 1994), but there are also arguments for maintaining distinctions among preschool_age children showing different patterns of controlling interactions with their caregiver (Crittenden, 1992). It may be that the infant disorganized and preschool controlling classifications help identify children who are, in general, at greater familial (and developmental) risk, but that a more fine_tuned understanding of interactional dynamics and their relation to specific patterns of developmental psychopathology will require the kinds of distinctions Crittenden (1992, 1994) makes in her preschool classifications. In this investigation, infant attachment history, including the disorganized classification, did not provide information about early reported behavior problems above and beyond 24_month (insecure) classification, nor did it discriminate, as 24_month atypical attachment did, between children with higher versus lower internalizing scores.

The fact that contemporaneous indices of child coping strategies, more specifically, child social orientation (sociability) and competence in exploration, did provide unique prediction to maternal report of behavior problems indicates, as one might expect, that attachment does not, in and of itself, capture all the variance in child behavior and interactions relevant to problems in reulation at a later date. Indices of emotional and behavioral regulation that are less specific to the attachment relationship may reflect temperamental differences (e.g., sociability, inhibition, activity level) in concert with other interactional and familial effects (e.g., parental scaffolding of task behavior, punitiveness, behavioral monitoring) that additively or interactively foster more problem behavior in the preschool or school period.

In summary, the data from this investigation suggest that attachment classification and attachment history provide useful information about child emotional and behavioral regulation. Generally, however, infant attachment appears most useful in understanding contemporaneous patterns of regulation, with knowledge about what are viewed as problems in regulation best forecast by attachment that is assessed later in time and/or by a different classification system. This makes inherent developmental sense, since change in family circumstances and change in child capabilities should be associated with alterations both in the attachment model, attachment behavior, and strategies for regulating emotion and behavior. The fact that some continuity is discernable supports theoretical work on the nature of the attachment internal working model, on the importance of genetic endowment/ temperament, and on developmental psychopathology. The degree of discontinuity that was present testifies to the importance of understanding how changing circumstances and capabilities are translated into individual and relationship functioning, both at a conceptual and methodological level.

REFERENCES

Achenbach, T.M. (1992). Manual for the Child Behavior Checklist/2_3 and 1992 Profile. Burlington, VT: Department of Psychiatry, University of Vermont.

Ainsworth, M.D.S., Blehar, M.C., Waters, E., & Wall, S. (1978). Patterns of attachment. Hillsdale, NJ: Lawrence Erlbaum.

Ainsworth, M.D.S., & Wittig, B.A. (1969). Attachment and the exploratory behavior of one_year_olds in a strange situation. In B.M. Foss (Ed.), Determinants of infant behavior (Vol. 4). London: Methuen.

Barnett, D., Ganiban, J., & Cicchetti, D. (1992, May). Temperament and behavior of youngsters with disorganized attachments: A longitudinal study. Paper presented at the International Conference on Infant Studies, Miami Beach.

Bates, E. (1979). The emergence of symbols: Cognition and communication in infancy. New York: Academic Press.

Bates, J.E. & Bayles, K. (1988). Attachment and the development of Behavior Problems. In J. Belsky & T. Nezworski, (Eds.), Clinical implications of attachment (pp. 253_299). Hillsdale, NJ: Lawrence Erlbaum.

Bates, J.E., Freeland, C.A., & Lounsbury, M.L. (1979). Measurement of infant difficultness. Child Development, 50, 794_803.

Bates, J.E., Maslin, C.A., & Frankel, K.A. (1985). Attachment security, mother_child interaction, and temperament as predictors of behavior_problem ratings at age three years. In I. Bretherton & E. Waters (Eds.), Growing points of attachment theory and research. Monographs of the Society for Research in Child Development,50 (1_2), 167_193.

Bayley, N. (1969). Manual for the Bayley Scales of Infant Development. New York: The Psychological Corporation.

Beckwith, L. & Rodning, C. (1991, April). Stability in attachment from 13 to 36 months. Paper presented at the Biennial Meeting of the Society for Research in Child Development, Seattle.

Belsky, J., Campbell, S.B., Cohn, J.F., & Moore, G. (1996). Instability of infant_parent attachment security. Developmental Psychology, 32, 921_924.

Belsky, J. & Rovine, M. (1987). Temperament and attachment security in the Strange Situation: An empirical rapprochement. Child Development, 58, 787_795.

Bowlby, J. (1969). Attachment and Loss. Vol. 1. Attachment. New York: Basic Books.

Bowlby, J. (1973). Separation. New York: Basic Books.

Bowlby, J. (1980). Attachment and Loss. Vol. 3. Loss. New York: Basic Books.

Calkins, S.D. (1994). Origins and outcomes of individual differences in emotion regulation. In N.A. Fox (Ed.), The development of emotion regulation, Monographs of the Society for Research in Child Development, 59 (2_3), 53_72.

Carlson, V., Cicchetti, D., Barnett, D., & Braunwald, K. (1989). Disorganized/ disoriented attachment relationships in maltreated infants. Developmental Psychology, 25, 525_531.

Campos, J.J., Caplovitz, K.B., Lamb, M.E., Goldsmith, H.H., & Stenberg, C. (1983). Socioemotional development. In M.M. Haith & J.J. Campos (Eds.), Handbook of child psychology: Vol. 3. Infancy and developmental psychobiology (pp. 783_915). New York: John Wiley.

Cassidy, J. (1994). Emotion regulation: Influences of attachment relationships. In N.A. Fox (Ed.), The development of emotion regulation, Monographs of the Society for Research in Child Development, 59 (2_3), 228_249.

Cassidy, J. & Berlin, L.J. (1994). The insecure/ambivalent pattern of attachment: Theory and research. Child Development, 65, 971_991.

Cassidy, J. & Kobak, R.R. (1988). Avoidance and its relation to other defensive processes. In J. Belsky & T. Nezworski (Eds.), Clinical implications of attachment (pp. 300_323). Hillsdale, NJ: Lawrence Erlbaum.

Cassidy, J., Marvin, R.S., & the MacArthur Working Group on Attachment (1991). Attachment organization in preschool children: Coding guidelines. Unpublished manuscript, University of Virginia, Charlottesville.

Cicchetti, D., & Barnett, D. (1991). Attachment organization in maltreated preschoolers. Development and Psychopathology, 3, 387_411.

Cicchetti, D., Ganiban, J., & Barnett, D. (1991). Contributions from the study of high_risk populations to understanding the development of emotion regulation. In J. Garber & K.A. Dodge (Eds.), The development of emotion regulation and dysregulation. New York: Cambridge University Press.

Crittenden, P.M. (1988). Relationships at risk. In J. Belsky & T. Nezworski (Eds.), Clinical implications of attachment (pp. 136_174). Hillsdale, NJ: Lawrence Erlbaum.

Crittenden, P.M. (1992). Quality of attachment in the preschool years. Development and Psychopathology,4, 209_241.

Crittenden, P.M. (1994). The Preschool Assessment of Attachment coding manual. Miami, FL: Family Relations Institute.

DiLalla, L.F., Thompson, L.A., Plomin, R., Phillips, K., Fagan, J.F., Haith, M.M., Cyphers, L.H., & Fulker, D.W. (1990). Infant predictors of preschool and adult IQ: A study of infant twins and their parents. Developmental Psychology, 26, 759_769.

Egeland, B. & Farber, E. (1984). Infant_mother attachment: Factors related to its development and change over time. Child Development, 55, 753_771.

Egeland, B., & Sroufe, L.A. (1981). Attachment and early maltreatment. Child Development, 52, 44_52.

Emde, R.N., Plomin, R., Robinson, J., Corley, R., DeFries, J., Fulker, D.W., Reznick, J.S., Campos, J., Kagan, J., & Zahn_Waxler, C. (1992). Temperament, emotion, and cognition at fourteen months: The MacArthur Longitudinal Twin Study. Child Development, 63, 1437_1455.

Egeland, B. & Erickson, M.F. (1987). Psychologically unavailable caregiving. In M.R. Brassard, R. Germain, & S.N. Hart (Eds.), Psychological maltreatment of children and youth (pp. 110_120). New York: Pergamon Press.

Fagot, B.I., & Kavanagh, K. (1990). The prediction of antisocial behavior from avoidant attachment classifications. Child Development, 61, 864_873.

Fagot, B.I., & Pears, K.C. (1996). Changes in attachment during the third year: Consequences and predictions. Development and Psychopathology, 8, 325_344.

Goldberg, S., Washington, J., Myhal, N., Janus, M., Simmons, R.J., MacLusky, I., & Fowler, R.S. Stability and change in attachment from infancy to preschool. Unpublished manuscript, Hospital for Sick Children and University of Toronto, Toronto, Ontario.

Goldsmith, H.H., & Alansky, J. (1987). Maternal and infant temperamental predictors of attachment: A meta_analytic review. Journal of Consulting and Clinical Psychology, 55, 805_816.

Goldsmith, H.H., Bradshaw, D.L., & Rieser_Danner, L.A. (1986). Temperamental dimensions as potential developmental influences on attachment. In J.V. Lerner & R.M. Lerner (Eds.), New Directions for Child Development: Temperament and psychosocial interactions in infancy and childhood (pp. 5_34). San Francisco: Jossey_Bass.

Greenberg, M.T., Speltz, M.L., & DeKlyen, M. (1993). The role of attachment in the early development of disruptive behavior problems. Development and Psychopathology, 5, 191_213.

Greenberg, M.T., Speltz, M.L., DeKlyen, M., & Endriga, M.C. (1991). Attachment security in preschoolers with and without externalizing problems: A replication. Development and Psychopathology, 3, 413_430.

Grossmann, K., Grossmann, K.E., Spangler, G., Suess, G., & Unzer, L. (1985). Maternal sensitivity and newborns' orientation responses as related to quality of attachment in Northern Germany. In I. Bretherton and E. Waters (Eds.), Growing points of attachment theory and research. Monographs of the Society for Research in Child Development, 50, 233_256.

Gunnar, M., Mangelsdorf, S., Larson, M., & Herstgaard, L. (1989). Attachment, temperament, and adrenocortical activity in infancy: A study of psychoendocrine regulation. Developmental Psychology, 25, 355_363.

Gunnar, M. (1990). The psychobiology of infant temperament. In J. Columbo & J. Fagan (Eds.), Individual differences in infancy. Hillsdale, NJ: Lawrence Erlbaum.

Howes, C. & Hamilton, C.E. (1992). Children's relationships with childcare teachers: Stability and concordance with parental attachments. Child Development, 63, 867_878.

Howes, C., Hamilton, C.E., & Matheson, C.C. (1994). Children's relationships with peers: Differential associations with aspects of the teacher_child relationship. Child Development, 65, 253_263.

Kagan, J. (1994). On the nature of emotion. In N.A. Fox (Ed.), The development of emotion regulation,Monographs of the Society for Research in Child Development, 59 (2_3), 7_24.

Kopp, C.B. (1982). Antecedents of self_regulation: A developmental perspective. Developmental Psychology, 18, 199_214.

Kopp, C.B. (1989). Regulation of distress and negative emotions: A developmental view. Developmental Psychology, 25, 343_354.

Kopp, C.B. (1992). Emotional distress and control in young children. In N. Eisenberg & R.A. Fabes (Eds.),Emotion and its regulation in early development (New Directions in Child Development, No. 55). San Francisco: Jossey_Bass.

Lamb, M., Thompson, R., Gardner, W., & Charnov, E. (1985). Infant_mother attachment. Hillsdale, NJ: Lawrence Erlbaum.

Lyons_Ruth, K., Alpern, L., & Repacholi, B. (1993). Disorganized infant attachment classification and maternal psychosocial problems as predictors of hostile_aggressive behavior in the preschool classroom. Child Development, 64, 572_585.

Lyons_Ruth, K., Bronfman, E., & Parsons, E. (1997). Maternal atypical responsiveness and disorganized infant attachment patterns. Unpublished manuscript, Department of Psychiatry, Cambridge Hospital.

Lyons_Ruth, K., Easterbrooks, M.A., & Cibelli, C.D. (1997). Infant attachment strategies, infant mental lag, and maternal depressive symptoms: Predictors of internalizing and externalizing problems at age 7. Developmental Psychology, 33, 681_692.

Lyons_Ruth, K., Repacholi, B., McLeod, S., & Silva, E. (1991). Disorganized attachment behavior in infancy: Short_term stability, maternal and infant correlates, and risk_related subtypes. Development and Psychopathology, 3, 377_396.

Main, M., Kaplan, N., & Cassidy, J. (1985). Security in infancy, childhood, and adulthood: A move to the level of representation. In I. Bretherton & E. Waters (Eds.), Growing point of attachment theory and research,Monographs of the Society for Research in Child Development, 50 (Serial No. 209), 66_104.

Main, M., & Cassidy, J. (1988). Categories of response to reunion with the parent at age six: Predicted from attachment classifications and stable over a one_month period. Developmental Psychology, 24, 415_426.

Main, M., & Solomon, J. (1990). Procedures for identifying infants as disorganized_disoriented during the Ainsworth Strange Situation. In M. Greenberg, D. Cicchetti, & E.M. Cummings (Eds.), Attachment in the preschool years: Theory, research, and intervention. Chicago: University of Chicago Press.

O'Connor, M.J., Sigman, M., & Brill, N. (1987). Disorganization of attachment in relation to maternal alcohol consumption. Journal of Consulting and Clinical Psychology, 55, 831_836.

Radke_Yarrow, M., Cummings, E.M., Kuczynski, L., & Chapman, M. (1985). Patterns of attachment in two_ and three_year_olds in normal families and families with parental depression. Child Development, 56, 884_893.

Rauh, H., Ziegenhain, U., Muller, B., & Wijnroks, L. (in press). Stability and change in infant_mother attachment in the second year of life: Relations to parenting quality and varying degrees of daycare experience. In P. Crittenden (Ed.), The organization of attachment relationships: Maturation, culture, and context. New York: Cambridge University Press.

Robinson, J.L., Kagan, J., Reznick, J.S., & Corley, R. (1992). The heritability of inhibited and uninhibited behavior: A twin study. Developmental Psychology, 28, 1030_1037.

Spieker, S.J. & Booth, C.L. (1988). Maternal antecedents of attachment quality. In J. Belsky & T. Nezworski (Eds.), Clinical implications of attachment (pp. 95_135). Hillsdale, NJ: Lawrence Erlbaum.

Shaw, D.S., Owens, E.B., Vondra, J.I., Keenan, K., & Winslow, E.B. (1996). Early risk factors and pathways in the development of early disruptive behavior problems. Development and Psychopathology, 8, 679_699.

Shaw, D.S., Winslow, E., Owens, E.B., Vondra, J.I., Cohn, J., & Bell, R. (1998). The development of early externalizing problems among children from low_income families: A transformational perspective. Journal of Abnormal Child Psychology, 26, 95_107.

Solomon, J., George, C., & DeJong, A. (1995). Children classified as controlling at age six: Evidence of disorganized representational strategies and aggression at home and at school. Development and Psychopathology, 7, 447_463.

Sroufe, L.A. (1983). Infant_caregiver attachment and patterns of adaptation in preschool: The roots of maladaptation and competence. In M. Perlmutter (Ed.), Minnesota symposium in child psychology (Vol. 16, pp. 41_81). Hillsdale, NJ: Lawrence Erlbaum.

Sroufe, L.A., Egeland, B., & Kreutzer, T. (1990). The fate of early experience following developmental change: Longitudinal approaches to individual adaptation in childhood. Child Development, 61, 1363_1373.

Teti, D.M., Gelfand, D.M., Messinger, D.S., & Isabella, R. (1995). Maternal depression and the quality of early attachment: An examination of infants, preschoolers, and their mothers, Developmental Psychology, 31, 364_376.

Thompson, R.A. (1994). Emotion regulation: A theme in search of definition. In N.A. Fox (Ed.), The development of emotion regulation, Monographs of the Society for Research in Child Development, 59 (2_3), 25_52.

Thompson, R.A., Flood, M.F., & Lundquist, L. (1995). Emotional regulation: Its relations to attachment and developmental psychopathology. In D. Cicchetti & S.L. Toth (Eds.), Rochester Symposium on Developmental Psychopathology: Emotion, cognition, and representation (pp. 261_299). Rochester, NY: University of Rochester Press.

Thompson, R.A., Lamb, M., & Estes, D. (1982). Stability of infant_mother attachment and its relationship to changing life circumstances in an unselected middle_class sample. Child Development, 53, 144_148.

Vaughn, B., Egeland, B., Sroufe, L.A., & Waters, E. (1979). Individual differences in infant_mother attachment at twelve and eighteen months: Stability and change in families under stress. Child Development, 50, 971_975.

Vondra, J.I., Hommerding, K.D., & Shaw, D.S. (1992, May). Infant attachment stability in a low_income sample. Paper presented at the International Conference on Infant Studies, Miami Beach.

Vondra, J.I., Shaw, D.S., & Hommerding, K.D. (in press). Stability and change in attachment in a low_income sample. In J.I. Vondra & D. Barnett (Eds.), Atypical attachment in infancy and early childhood among children at developmental risk, Monographs of the Society for Research in Child Development.

Vondra, J.I., Shaw, D.S., & Kevenides, M.C. (1995). Predicting infant attachment classification from multiple, contemporaneous measures of maternal care. Infant Behavior and Development, 18, 415_425.

Wartner, U.W., Grossmann, K., Fremmer_Bombik, E., & Suess, G. (1995). Attachment patterns at age 6 in South Germany: Predictability from infancy and implications for preschool behavior. Child Development, 66, 69_79.

Waters, E. (1978). The reliability and stability of individual differences in infant_mother attachment. Child Development, 49, 483_494.

Williamson, G.G., Zeitlin, S., & Szczepanski, M. (1989). Coping behavior: Implications for disbled infants and toddles. Infant Mental Health Journal, 10, 3_13.

Zeitlin, S., & Williamson, G.G. (1990). Coping characteristics of disable and nondisabled young children. American Journal of Orthopsychiatry, 60, 404_411.

Zeitlin, S., Williamson, G.G., & Szczepanski, M. (1988). Early Coping Inventory. Bensenville, IL: Scholastic Testing Service, Inc.