Table of Contents | Previous | Next |
HOW MUCH BETTER THAN EXPECTED? IMPROVING COGNITIVE OUTCOMES IN UTAH’S BEAR RIVER EARLY HEAD START
Lori A. Roggman, Lisa K. Boyce, Gina A. Cook and Andrea
D. Hart
Utah State University
What are the strongest early predictors of later cognitive skills? Can Early Head Start (EHS) buffer the effects of early risk indicators? Is the developmental trajectory of cognitive skills different for EHS children than comparison group children? Do EHS children do better than expected, based on predictions? And if they do better than expected, what aspects of EHS are related to how much better they do? These are some of the questions we asked as part of our local research with Utah’s Bear River Early Head Start.
The goal of Bear River EHS is to improve the developmental outcomes for infants and toddlers by helping low-income parents provide experiences infants and toddlers need during their early development. For children at risk because of poverty, EHS tries to help keep them on track developmentally so they make the same cognitive gains as children in more optimal circumstances. Families who applied and qualified for Bear River EHS were randomly assigned to either EHS or a comparison group, with children in EHS expected to do better developmentally because of the extra support provided by the program for them and their parents.
Of course in both the EHS and comparison groups, all of the infants and toddlers developed. The question is whether EHS children developed “more.” To answer that question is challenging. One challenge is that the population which is served by EHS programs varies widely in many ways: family background, reasons for the family’s low-income status, psychological and social functioning of the parents, and even the extent to which the family participates in the EHS services offered. These complex variations together form the context in which infants and toddlers develop and must be examined in order to assess developmental progress. An even greater challenge is that development in the first three years is rapid and variable with spurts and lulls common to all children. Also, during the first three years of life, developmental trajectories become increasingly differentiated for children in different environments.
In populations considered “at risk” for various reasons, there is a common pattern of early development. Except for those with relatively severe medical or developmental problems, test scores in the first year are typically about the same for infants in at-risk environments as they are for infants in low-risk environments. In the second and third years of life, however, the developmental trajectories begin to diverge for children in different environments (Egeland, Sroufe & Erickson, 1983; Egeland & Erickson, 1987; Gorman & Pollitt, 1992; Johnson, Diano, & Rosen, 1984; Rogan & Gladen, 1993; Villar, Smeriglio, Martorell, Brown & Klein, 1984). Children at risk because of poor nutrition, drug exposure, low socio-economic status, or poor parenting begin to fall behind; their cognitive test scores begin to decline compared to their peers. For this reason it is especially important to consider the complexities of early environments and to consider changes with time or age in addition to assessing intervention group differences on developmental test scores.
Despite all the variations in family context, in EHS participation, and in developmental trajectories, it was expected that those who had been randomly assigned to EHS would make more progress in cognitive skills than those who had not. Indeed, previous analyses indicated that by age two, EHS children’s cognitive skills were “better than expected,” and comparison group children’s were “worse than expected,” based on early predictors that included both family and child variables (Roggman, Boyce & Cook, 2001). To test whether EHS children at age three continued to do “better than expected,” as they had seemed to at age two, it was essential to look at interactive effects of EHS with developmental change over time.
By looking at the combination of developmental change in cognitive skills (comparing tests with earlier assessments) and intervention (comparing EHS to a comparison group) we can see a pattern of effects that takes into account both early risk factors and maturation in addition to differences in environmental support provided to the EHS group versus the comparison group. To see if the developmental path or trajectory for cognitive skills is different for children in EHS versus the comparison group, we included both age and intervention group in our data analyses. Our approach to statistical analysis is different from that used for the national cross-site study first because it considers both age and intervention together, and second because it includes early predictors from before families were enrolled in the EHS research study.
We have used several statistical methods to test the question of whether development is “better” for children in our local EHS group versus the comparison group. For each set of analyses, we used developmental measures at more than one age point, a grouping variable indicating whether the child’s family was in EHS or not, and in addition, a set of the strongest early predictors of children’s cognitive outcomes at age three.
Method
Our EHS local research project included 201 mothers (103 EHS group, 98 comparison group) who were either pregnant at the time of application or had infants less than10 months old. To meet program requirements, over 90 percent were low income as defined by federal poverty guidelines, and most families (97 percent) received some sort of public assistance such as Medicaid, food stamps, and WIC. Most children were Caucasian (82 percent, 11 percent Latino, 7 percent other). Their mothers were mostly married or living with a partner (73 percent), over the age of 19 (75 percent; mean age = 22.9), had at least a high school education (65 percent), and were not working (79 percent). Family size at enrollment ranged from zero to seven children.
The developmental outcome that is the focus of this study is cognitive skills. Cognitive skills were assessed using the Bayley Scales of Infant Development at 14, 24, and 36 months. These data were collected as part of the national study. Additional data for this study included early measures of parent functioning expected to be related to children’s development. These data were from interviews with mothers before random assignment to EHS or a comparison group. In addition, data were collected assessing the quantity and quality of services to the families in EHS.
Maternal interviews included questions about family characteristics (e.g., education, employment, income, ethnicity, marital status, family size). These interviews also included questions adapted from questionnaire scales developed for measuring various aspects of psychosocial functioning. The scales used for these analyses included those measuring maternal depression, social support, and attitudes about close relationships. The measure of maternal depression was The Center for Epidemiological Studies Depression Scale (CES-D; Radloff, 1977). Questions from the CES-D ask how often in the past week the individual has had emotions and thoughts associated either positively or negatively with depression, such as “I felt happy,” or “I thought my life had been a failure.” Reliability on this measure has been reported as a coefficient alpha of .92 (Radloff, 1977). Social support was assessed using items from the Family Crisis Oriented Personal Evaluation scale (F-COPES) of family coping related to the use of social support from friends, neighbors, and relatives (McCubbin & Patterson, 1982). Questions were modified from the F-COPES to use less difficult language and to ask how frequently the mother was likely to use these coping strategies. Parents were asked how often they sought support when there was a problem, for example, how often they “talk about a problem with neighbors” or “seek advice from relatives.” Reliability on this measure has been reported as a coefficient alpha of .83 (McCubbin & Patterson, 1982). Attitudes about close relationships were measured by the Adult Attachment Style scale, providing an avoidance index in addition to an overall insecurity score. Questions include “I find it difficult to trust others completely” and “I’m comfortable having others depend on me” (Simpson, Rholes & Nelligan, 1992). Although the scale was originally developed to assess orientation toward romantic relationships, the few items that refer to “partner” were revised to refer to “people close to me” to include relationships with family members or close friends. Reliability on this measure has been reported as a coefficient alpha of .81 (Simpson, et al., 1992).
In addition to maternal interview and child testing data, data for this study also included indicators of the quantity and quality of EHS services. Program staff provided a tally of the number of home visits and group activities in which each family participated. In addition, videotaped home visits were coded using observational measures. Trained coders rated parent engagement from 1 (unengaged) to 6 (highly engaged) using an established scale (McBride and Peterson, 1997). Coders also rated the effectiveness of home visitor facilitation of parent-child interaction during home visits. Coders used a 5-point coding scheme developed with program staff, with 1 representing no home visitor facilitation or overly intrusive and directive behavior and 5 representing effective facilitation and responsiveness. A second coder independently coded 13 of the home visit videotapes (22 percent of the total 58 tapes), and inter-rater agreement was the same for both scales used in the analyses, 88 percent, Kappa = .75.
Our analyses involved a series of steps to assess whether or not EHS children were performing better than expected on cognitive skills tests. First, we explored possible early risk indicators by calculating correlations between early measures and later cognitive outcomes. Second, we tested the statistical interaction of development and intervention in a repeated measures analysis of variance testing age by group interactions, with age point as a within subjects variable, EHS versus comparison group as a between subjects variable, and selected early predictors as covariates. Third, we developed regression models of early predictors of later outcomes, assessed “better than expected” outcomes by examining the residuals (differences between predicted and actual scores), and compared the residuals for EHS versus the comparison group. This approach to analysis of longitudinal data has been used successfully in previous studies of constructs similar to those of interest in the proposed research (Pianta, Sroufe & Egeland, 1989; Roggman, Hart & Jump 1996). Finally, we explored correlates of the residuals to see what the strongest predictors were of children doing “better than expected.”
Results and Discussion
What are the strongest early predictors of later cognitive skills? Expected predictors of later developmental outcomes were examined. The strongest predictors of poor cognitive skills at 36 months were measures of cognitive skills at earlier ages, 14 and 24 months, r = .48, p < .001; r = .67, p < .001. Of course, other aspects of the early environment may also affect poor cognitive development. Risk factors were examined that were expected to predict cognitive outcomes among the toddlers in this sample. Indicators of poor parental functioning that predicted poorer later cognitive skills included low maternal education, r = .29, p < .01, high maternal insecurity--specifically avoidance in close relationships, r = -.30, p < .01, and infrequent family use of social support, r = .26, p < .01.
Is the development of cognitive skills any different for EHS than comparison group children? To answer this question, we used a repeated measures analysis of variance with time of measurement as a within-subjects variable and program versus control group as a between-subjects variable. Our analyses also included covariates based on the strongest earlier predictors: maternal education, insecurity, and social support. We tested the statistical interaction of age and group to see if change over time was different for children in the EHS program group versus the comparison group. Results of between-group repeated measures (by age) analyses of variance showed that for cognitive skills scores, there were statistically significant interactions between age and group, F (2, 83) = 3.68, p < .05. What this means is that for cognitive skills, age changes were different for those in EHS than for those in the comparison group. Simple effects tests were used to test age changes within each group, the EHS group and the comparison group. Across the three age points, EHS toddlers maintained stable standardized test scores that did not change significantly with age, while comparison group toddlers, similar to others in poverty, began to lose ground as indicated by statistically significant decreases in their standardized cognitive skill scores (simple effects test for comparison group, F (2, 163) = 6.2, p < .01). Figure 1 shows the different trajectory for the EHS children versus the comparison group.
Do EHS children do better than expected, based on early predictors? To answer these questions, we compared children’s actual test scores with their predicted ones. The predictions were based on regression models using the strongest early predictors. We used the earlier assessment along with early risk predictors to predict later outcomes. The strongest earlier predictor is, of course, an earlier measure of the same thing. We included the earliest measure of cognitive skills, the Bayley at 14 months, in all regression models. Low maternal education at enrollment was the most persistent predictor of low scores on later cognitive skills assessments so we included maternal education level in all models as well. To get the best prediction, we also included maternal avoidance in close relationships and use of social support to predict later cognitive skills at 36 months. The resulting regression model explained about a third of the variance in cognitive skills scores, R = .60, Adj.R2 = .33, F (4, 89) = 12.5, p < .001.
Based on this predictive model, we examined the residuals, the differences between the predicted scores and children’s actual scores, to see if children were doing better or worse than expected in the cognitive domain. The greater the distance the actual score was above the predicted score, the more a particular child was doing better than expected; the greater the distance the actual score was below the predicted score, the more a particular child was doing worse than expected.
The EHS children were, on the average, doing better than expected; the comparison group were, on the average, doing worse than expected. The group difference in these residuals was statistically significant, t (90) = 2.1, p < .05.
|
[D] |
EHS children do better than expected in cognitive development. What aspects of EHS are related to how much better they do? Measures of the quantity and quality of EHS services, for those who received at least 6 months of services, were examined in relation to the size of residuals, that is, to how much better the cognitive skills scores were than expected based on early predictors. The strongest correlate of the residuals was the rating of parent engagement during observed home visits, r = .37, p < .05. Additional variables were, in turn, related to parent engagement during home visits. These included the effectiveness of home visitors in facilitating parent-child interaction during home visits, r = .53, p < .001, total number of group activities attended, r = .30, p < .05, and lack of maternal avoidance in close relationships, r = -.43, p < .01. Mothers who were more engaged in home visits were thus more trusting and responsive to close relationships, more likely to participate in other program activities, and more likely to have more facilitative home visitors.
Summary
In summary, the developmental trajectory is better for children in EHS compared to the comparison group. Early risk factors of poor maternal education, maternal avoidance, and infrequent family use of social support appeared to be buffered by the EHS experience. While cognitive skills scores declined for the comparison group, they did not for the EHS children. For children from low-income families in northern Utah and southern Idaho, those who had been enrolled in Bear River EHS had better than expected outcomes in the cognitive domain. They did better on cognitive tests than expected, maintaining age appropriate progress in their cognitive skills in spite of early test scores and early risk factors. In contrast, toddlers in the comparison group did not show similar progress in the cognitive domain; they did not maintain age appropriate cognitive skills. The advantage gained by EHS children was evidently due to the level of engagement of their mothers during the EHS home visits, engagement that was related to more involvement in other EHS activities, more facilitative home visitors, and less maternal avoidance.
By examining both age changes and intervention, our results indicate a different developmental trajectory for EHS toddlers versus the comparison group. Even though the average group differences in Bayley scores are not large clinically, the EHS group is maintaining their trajectory during an age period when children with similar risk factors typically begin to decline. This difference in trajectories is especially important for an at-risk group whose developmental trajectories, with increasing age and exposure to risk factors, would be expected to diverge substantially from those children in more optimal environments.
References
Egeland, B., Sroufe, L. A., & Erickson, M. F. (1983). Developmental consequences of different patterns of maltreatment. Child Abuse and Neglect, 7, 459-469.
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.
Gorman K. S., & Pollitt E. (1992). Relationship between weight and body proportionality at birth, growth during the first year of life and cognitive development at 36, 48, and 60 months. Infant Behavior & Development, 15, 279-96.
Johnson, H.L., Diano, A., & Rosen, T.S. (1984). 24-month neurobehavioral follow-up of children of methadone-maintained mothers. Infant Behavior & Development,7, 115-123.
McBride, S.L., & Peterson, C.A. (1997). Homebased interventions with families of children with disabilities: Who is doing what? Topics in Early Childhood Special Education, 17, 209-233.
McCubbin, H. I., & Patterson, J. M. (1982). Family adaptation to crisis. In H. McCubbin, A. Cauble, & J. Patterson (Eds.), Family Stress, Coping, and Social Support. Springfield, IL: Thomas.
Pianta, R. C., Sroufe, L. A., & Egeland, B. (1989). Continuity and discontinuity in maternal sensitivity at 6, 24, and 42 months in a high-risk sample. Child Development, 60, 481-487.
Radloff, L. S. (1977). The CES-D Scale: A self-report depression scale for research in the general population. Applied Psychological Measurement, 1, 385-401.
Rogan W.J., & Gladen B.C. (1993). Breast-feeding and cognitive development. Early Human Development, 31, 181-93.
Roggman, L. A., Boyce, L. K., Cook, G. A., (2001). Keeping kids on track: Interactive effects of age and intervention. In Mathematical Policy Research & Head Start Bureau, Early Head Start Interim Report to Congress. Washington, DC: DHHS, ACF, ACYF.
Roggman, L. A., Hart, A. D., & Jump, V. (1996, April).
Attachment in relation to parenting stress and temperament: Longitudinal
effects from 10 to 18 months. International
Conference on Infant Studies, Providence, RI.
Simpson, J., Rholes, W., & Nelligan, J. (1992). Support seeking and support giving within couples in an anxiety-provoking situation: The role of attachment styles. Journal of Personality and Social Psychology, 62, 434-446.
Villar J., Smeriglio V., Martorell R., Brown C.H., Klein R.E. (1984). Heterogeneous growth and mental development of intrauterine growth-retarded infants during the first 3 years of life. Pediatrics, 74, 783-791.
| Table of Contents | Previous | Next |