Dimensions of Early Childhood Behavior Problems: Stability and Predictors of Change from 18 to 30 Months.

Kristin S. Mathiesen [1,3]

Ann Sanson [2]

The prevalence, structure, stability, and predictors of change in early behavior problems were examined in a population-based sample of Norwegian children at 18 and 30 months of age (N = 750). A clear factor structure involving four dimensions emerged at both assessment times: Two factors were characterized by externalizing behaviors and were labeled Social Adjustment and Overactive-Inattentive; one factor tapped internalizing problems and was labeled Emotional Adjustment; and the fourth, related to general immaturity, was labeled Regulation. Specific patterns of child and family risk factors were associated with stability and change over the two time points for each factor. Children with stable problems had the most problematic characteristics on all significant predictors, followed by children with problems at one, but not both, time points. The data suggest that it is possible to identify risk factors for stable problems at 18 months, allowing some prediction of those children whose problems will persist o ver early childhood. Since specific risk factors emerged for specific types of behavior problems, the results may provide some much-needed guidance to early intervention efforts.

KEY WORDS: Preschool children; behavior problems, temperament, risk factors.

It is generally accepted that from 12 to 20% of children exhibit behavior problems during the early developmental period (Dunn, 1988). Normative data from instruments measuring behavior problems are available for children from 2 to 3 years of age (Achenbach, Edelbrock, & Howell, 1987; Prior, Smart, Sanson, Pedlow, & Oberklaid, 1992; Richman, Stevenson, & Graham, 1982; Thomson et al., 1996). However, the assessment and prevalence of behavior problems among very young children (1-3 years) have been only scantily addressed (Koot, Van Den Oord, Verhulst, & Boomsma, 1997). Children's behaviors often are believed to be diffuse and undifferentiated before 2 to 3 years of age compared with behaviors at later ages (Buss & Plomin, 1984; McGuire & Richman, 1986). The developmental nature of several problem behaviors makes it difficult to differentiate them from normal challenging behaviors at the earliest ages. Data on age-related changes suggest that many annoying or difficult behaviors (as defined by adult caregivers) are age appropriate, reflecting developmental change or age-related conflict or frustration (Campbell, 1995). As a consequence, the majority of the large community-based and longitudinal studies have started the assessment of behavior problems when the children were 3 years of age or older without trying to identify problem behaviors at earlier ages. Therefore, knowledge about problem identification and persistence among children in the youngest age groups is still limited. Besides its theoretical importance, such knowledge is essential as a basis for well-targeted and effective preventive interventions.

Although most problem behaviors among toddlers do not form clear-cut diagnostic categories, data obtained from factor analysis of items used in various scales have provided evidence for a multidimensional structure of disorders. The studies have varied, however, in the number of factors or dimensions derived to describe problem behavior. Campbell (1995) compared results across different measuring instruments and concluded that problems in young children generally load on two broadband factors: externalizing, which encompasses problems with attention, self-regulation, aggression, and noncompliance; and internalizing; which includes depression, withdrawal, and anxiety. In addition, there is evidence of more fine-grained distinctions between dimensions within these two broad bands.

Similar factors also have been found in the few studies of behavior problems among children between 2 and 3 years old. These studies have mainly used the Child Behavior Checklist for Ages 2-3 (CBCL/2-3; Achenbach, 1992; Achenbach et al., 1987). A comparison of the factor structure between American and Dutch samples at these ages supported the validity of differentiating the problem behaviors into an internalizing grouping (constituting withdrawal and anxiety) and an externalizing grouping (reflecting overactive behavior, aggression, and oppositional behavior). Sleep problems were found to represent a separate syndrome at this age (Koot et al., 1997). Although these American and Dutch studies increase knowledge of behavior problems in early childhood substantially, the validity of differentiating young children's problem behavior into syndromes at these ages would be further supported if similar factors were found when different instruments assessed behavior problems.

As a parent-report screening instrument, the CBCL/2-3 scale is long, containing 99 questions about behavioral and emotional problems. The Behavioral Checklist (BCL; Richman, 1977) contains a relatively short list of items (19), and the fuller descriptions in the wording of the questions about behavior problems may increase parents' comfort in answering them. The validity and internal consistency of the BCL have been demonstrated through various methodologies, including observations of children, comparisons between clinical and nonclinical populations, and comparisons with another screening questionnaire (Boyle & Jones, 1985; Koot & Verhulst, 1991; McGuire & Richman, 1986; Richman, Stevenson, & Graham, 1975). Several researchers have used the instrument to detect preschoolers at risk for behavior problems (Earls & Joung, 1987; Richman et al., 1982; Stallard, 1993; Thomson et al., 1996). The main limitation of the BCL compared to the CBCL scale is that the BCL does not cover all of the domains that are measure d through the CBCL.

The factor structure of the 19 items in the BCL was examined recently in a large British study (Sonuga-Barke, Thompson, Stevenson, & Viney, 1997). Analysis of the BCL scores completed by 1,047 parents of 3-year-old British children revealed six factors representing poor social adjustment, poor emotional adjustment, overactive-inattentive, sleep problems, eating problems, and soiling. This study argues against the undifferentiated nature of problems in early childhood. In addition, the study supports the idea that problems of a clinically significant level are likely to be comorbid since the children classified into the group with severe problems (10.4% of the sample) were showing problems in more than one area. It is not known, however, whether the BCL will detect similar maladjustment dimensions among younger (1- to 3-year-old) children.

If behavior problems in the early years were usually transitory, the need for accurate assessment would be less vital. However, the stability of behavior problems usually is found to be high from 3 to 4 years on and to increase with increasing age (Hemphill, 1996; McConaughy, Stanger, & Achenbach, 1992; Verhulst & Koot, 1992; Verhulst & van der Ende, 1995). In her review of recent research in the field, Campbell (1995) noted that follow-up studies of preschool children identified as having behavior problems at ages 3 or 4 years generally report a high probability (around 50%) that the children will continue to show difficulties throughout the elementary school years. The stability of behavior problems seems to vary according to the types of problem that are studied. Early-emerging attentional and discipline problems often are found to be most persistent (Hemphill, 1996; Richman et al., 1982; Sanson & Prior, 1998; Verhulst & van der Ende, 1995), although relatively high stability is found for both externalizi ng and internalizing problem behavior in samples from the general population (McConaughy et al., 1992; Sanson, Pedlow, Cann, Prior, & Oberklaid, 1996; Verhulst & Koot, 1992). Such findings point to the value of early intervention. However, despite there being a great practical need for knowledge about the stability of behavior problems from the second year of life, population-based studies have only rarely examined the stability of problem behaviors from such young ages.

The persistence of problems seems to reflect more than just continuity in child behavior. A substantial proportion of the children with behavior problems at an early point of time "recover" or return to a normative level of functioning, but others develop problems later in the preschool years, and others remain in the maladjusted group (Campbell, 1995; Prior et al., 1992). The emergence and maintenance of behavior problems appears to be explained best by a combination of environmental and intrinsic child factors. A direct connection between maternal depression and behavior problems in children is well and consistently documented (Cummings & Davies, 1994; Rutter, 1990; Webster-Stratton & Hammond, 1988). It is further documented that the mothers of children showing problems during the preschool period are more likely to have low social status and to have been exposed to more stressful life events during the past year than other mothers (Campbell, Pierce, March, & Ewing, 1991; McGee, Patrige, Williams, & Silva, 1991; Richman et al., 1982).

Marital conflict appears to have significant effect on children's behavioral problems (Cummings & Davies, 1994; Dadds & Powell, 1991). The influence of the parents' social networks on parenting and child development also has been documented (Crnic, Greenberg, & Slough, 1986). Support available to the mothers is frequently available to the infants as well, so that, for example, the negative impact of unresponsive mothering on child development may be buffered by involved grandparents, fathers, or friends (Werner, 1994). However, it is likely that children differ in their reaction to similar environmental factors because of individual characteristics. It is documented that intrinsic child factors such as temperament characterized by negative emotionality and shyness (Kagan, Snidman, & Arkus, 1995; Oberklaid, Sanson, Pedlow, & Prior, 1993; Rutter, 1987; Sanson & Rothbart, 1995) and physical health problems contribute to problem status among children.

One study that has examined stability and change in behavior problems from early childhood is the Australian Temperament Project (ATP; Prior, Sanson, & Oberklaid, 1989). In that longitudinal study of a community sample, Prior et al. (1992) examined the differences between stable and transient behavior problems of children and the association between group assignment and different sets of risk factors. The children were classified into stable and transient groups according to four ratings of overall behavior problems between 2 and 6 years. Prior et al. found group differences to be linear rather than categorical, with children with stable behavior problems showing more difficult temperament and more aggressive behavior in the 2- to 4-year age period than those who had "recovered." The children with transient problems had more problematic characteristics than the children without behavior problems at all of the four time periods. Maternal psychological health and stress factors in the family were also more adv erse for the stable group. This study did not differentiate between types of behavior problems. Since other studies have shown that the effects of intrinsic and environmental factors on child behavior vary according to the type of problem behavior (Hirshfeld et al., 1992; Rothbart & Bates, 1998; Sanson & Prior, 1998), it is worthwhile to consider predictors of stability and change in more differentiated types of problem behavior.

Theoretically, one might expect temperamental activity and emotionality to lead particularly to externalizing problems, and low sociability and high shyness to lead to internalizing problems. Empirically, there is some evidence of such specific correlates, despite the fact that internalizing and externalizing scores are often substantially correlated (Bates, 1990; Caspi, Henry, McGee, Moffitt, & Silva, 1995; Kagan et al., 1995). Different environmental factors seem to be specifically related to different maladjustment dimensions as well. Externalizing problem behaviors are shown to be associated with factors such as power-assertive parenting, depressed mothers, and family experience of adversity and stress (Hemphill, 1996). Negative self-attributions, avoidant coping styles, and maternal anxiety tend to discriminate children with anxiety and/or depression from normal controls (Hirshfeld et al., 1992; Rapee, 1997). Given arguments about differential developmental pathways to internalizing and externalizing pr oblems, it seems to be important to examine predictors of stability and change separately for these dimensions.

The current study is a part of a larger, population-based, and longitudinal project designed to investigate the influence of environmental risk and protective factors, child temperament, and maternal personality factors on mental health problems among children as well as their mothers. The study sought to identify dimensions of behavior problems in this population of young children, examine their stability from 18 to 30 months, and determine the predictors of stability and change. Specifically, we aimed to test the following hypotheses:

1. …