American Journal of ComnUnty Psychology, Vol. 15, No. 4 1987:

Primary Prevention of Behavior Problems in Mexican-American Children

[)ale L. Johnson and Todd Walker

UJniversity of Houston-University Park

~Behavior problems in children often appear early and some, especially those of the acting-out, aggressive type, tend to persist for many years, often continuing into adulthood (Loeber. 1982). Such behavior problems interfere with learning in school and are a matter of great concern for teachers, parents, and the children themselves. Attempts to treat these disorders have met with mixed results (Rutter & Garmezy, 1983). 11 is, therefore, a highly desirable goal to prevent the de\'elopment of behavior problems. There have been numerous primary prevention efforts, but most have been ineffective. Recently some efforts utilizing social problem solving training have

A primary prevention program, the Houston Parent-Child Development Center, directed towards infants and their parents, has effectively reduced the frequency of behavior problems for these children 5 to 8 years after the program’s completion. Teacher ratings showed significantly fewer acting-out, aggressive behaviors for program children. Ratings of classroom behaviors found program children significantly less hostile and more considerate than control children. In addition, program boys were less dependent than control boys. This appears to be the first primary prevention program to have demonstrated effectiveness in reducing behavior problems over such a long time.

Behavior problems in children often appear early and some, especially those of the acting-out, aggressive type, tend to persist for many years, often continuing into adulthood (Loeber, 1982). Such behavior problems interfere with learning in school and are a matter of great concern for teachers, parents, and the children themselves. Attempts to treat these disorders have met with mixed results (Rutter & Garmezy, 1983). It is, therefore, a highly desirable goal to prevent the development of behavior problems. There have been numerous primary prevention efforts, but most have been ineffective. Recently some efforts utilizing social problem solving training have demonstrated short-term effectiveness (Shure & Spivack, 1978). However, they have not always been successfully replicated (Rickel, Eshelman & Ligman, 1983). Schweinhart and Weikart (1983) obtained near-significant group differences (p < .06-.08) in behavior problems in a follow-up of Perry Preschool Project children. The absence of positive primary prevention effects has threatened the entire primary prevention enterprise, leading researchers and clinicians to conclude that treatment is the only viable response to children’s behavior problems. Before accepting that conclusion, we sought to test more intensive, early training procedures involving parents and children together. The oresent report is of an5 to 8 year follow-up of a primary prevention program that was conducted when children were from 1 to 3 years of age and without history of behavior problems when they entered the project.

The project, called the Houston Parent-Child Development Center (PCDC), was designed to promote social and intellectual competence in children from low-income Mexican-American families. It was one of three PCDCs established initially with Office of Economic Opportunity Support to develop effective interventions with low-income families. The other two original PCDCs, located in Birmingham and New Orleans, were designed differently and worked with other ethnic groups (Andrews et al., 1982). The Houston PCDC continues today as a nonresearch-oriented delivery system under the auspices of Head Start. The children were regarded as at risk for behavior problems because of their low socioeconomic and minority status.trhe goals for the program were broad but included preventing behavior problems and school failure (Johnson, 1975). The program required approximately 550 hours of participation over a 2-year period. The program las been described in detail in l2 volumes of curriculum materials and in summary form by Johnson, Kahn, and Leler (1976). In order to assure a igh level of participation, the first year of the program was in the home and was focused on rhe mother-child interaction. Mothers were visited by paraprofessionals 25 times for I V2-hour sessions during which they exchanged information about child development, parenting skills, and the use f the home as a learning environment. In order to involve fathers and sib- l1gs in the program, entire families participated in several weekend sessions 1 such issues as decision making in the home and family communication. during the second year, mother and child came to the project center four mornings a week to participate with other families in classes on child management, child cognitive development, family communication skills, d other topics related to family life. Mothers spent parr of their time in 'me management sessions and during this time the children were in a nursery school. Staff for the second year of the program included both professionals and paraprofessionals (Andrews et al. , 1982; Johnson, Breckenridge, & McGowan, 1984). A major short-term objective was 10 assist the mother in developing an affectionate relationship with her child This was done in the first year by helping mothers understand their infants. emotional states and developmental levels. In the second year, micro- teaching techniques were used in which mothers were videotaped while leaching their children or carrying out a child management activity. Tapes were viewed by individual mothers, and if they wished, by other program mothers, and positive feedback given.

Program evaluation results have been reported earlier (Andrews et al" 1982; Bridgeman, Blumenthal, & Andrews, 1981; Johnson et al" 1984), In general, for 3-year-old children, the PCDC had highly significant effects on the mothers and their children At the end of the program, mothers who had participated in the program. compared wilh randomly assigned controls, were more affectionate, used more appropriate praise, used less criticism, used less restrictive control, were more encouraging of child verbalization, provided a more stimulating home environment, and held more modern values. Program children had higher Bayley Mental Developmental Index scores at 2 years and higher Stanford-Binet IQs at 3 years.

A first follow-up evaluation of Ihe primary prevention effectiveness of the PCDC was carried out by interviewing 128 mothers of 4- to 7-year-old children about their child's behavior. A multivariate analysis of covariance (MANCOV A) showed main effects for group and sex as well as a Group x Sex interaction. Control boys were more destructive, overactive, negative attention-seeking, and less emotionally sensitive than children in the other three groups (Johnson & Breckenridge, 1982) Although strong program effects were found, the results were subject to possible bias because mothers, who were also project participants, provided the information about their children. To protect against observer bias a second follow-up was carried out, with teachers as the judges of behavior problems. II was expected that teacher ratings would support the results for mother reports; that is, control children were expected to show more acting-out, hostile behaviors than program children.

METHOD

The number of children in the follow-up study varied slightly as a function of the measure used. The Ns for each measure are shown in Tables 1 and II. At intake, when children were l-year-old, families had been assigned randomly to program and control groups as follows: A door-to-door search of Houston's barrios was conducted to identify low-income Mexican- American families with a l-year-old child who was healthy and had no apparent neurological problems. Eligible families were told about the project in droil. It was explained that there was a "program group" and a "child development research group" and that whether they were assigned to one or the other would depend on chance. They were asked if they would participate in either case. If the answer was affirmative, the family was randomly assigned. Controls received only assessment procedures. School data were collected when the children were in Grades 2 through 5, ages 8 to 11. The children had been in five different PCDC cohorts and represented a 59OJo sample of the total number of families in these cohorts. Others could not be located, largely because families had moved from the area. When families followed were compared with families not followed on family background characteristics of income, marital status, parental education, number of siblings, and language usage, no differences were found. There had been no differences between groups on these variables at intake. Thus, the original assignment of families was random and there was no evidence of differential attrition. Data for these analyses were collected at intake only. It may be of interest that although the recruitment procedure did not attempt to obtain a representative sample of low-income Mexican- American families in Houston, a comparison of demographic characteristics of the PCDC sample with census data for the areas included suggested that the PCDC families were typical of low-income families in the areas.

The number of control children followed is greater than that for program children because for two of the early cohorts two control groups were formed. One was to receive assessments plus medical examination and follow-up care if needed for the index child and the other control group was to receive assessments only. For a number of reasons it was not possible to carry out the planned medical examinations for the control group and, as the two control groups did not differ, they were combined. Thus, the number of control children was greater at intake. Dropout during the two program years was essentially the same, about 50%, for the program and control groups and was due almost entirely to families moving out of the project area. The attrition rate seems high, but it was consistent with attrition rates for programs reviewed by Gray and Wandersman (1980).

Fathers and mothers had had 7 years of education, there were 3.4 children in each family, per capita family income was $1,300, 91% of the households had a father present, all of the parents spoke Spanish, and about two-thirds were to some degree bilingual in Spanish and English. Background characteristics of project families have been presented in greater detail in Johnson et al. (1984).

Information on behavior problems was obtained with the AML, a brief, 11-item, screening instrument (Cowen et al., 1973). Teachers rated the frequency of certain classroom behaviors along a continuum ranging from I = never to 5 = most or all of the time. The AML consists of five items for A (aggressive, acting-out behaviors), five items for M (moody, withdrawn behaviors), and one for L (learning difficulties). Adequate test- retest and internal consistency reliabilities have been reported (Cowen et al., 1973). Satisfactory construct validity has also been reported (O'Brien & Bennett, 1983). Teachers did not know whether the children they were rating had been in the PCDC program or control group.

Child behavior was also assessed by teachers using the Classroom Behavior Inventory (CBI). The 32-item version of the CBI was used (Schaefer & Edgerton, 1976). Teachers rated children on five-point scales which were selected to describe both positive and negative classroom behaviors. Eight scales were used: Hostility, Con:siderateness, Extroversion, Introversion, Task-orientation, Distractibility, Intelligent Behavior, and Dependency. Schaefer (1975) has reported inter-rater reliabilities in the .60s for the CBI when completed by teachers who had had contact with children in open classrooms. The Task-Orientation, Distractibility, and Intelligent Behavior factors were related to achievement test scores, but the other factors were not. Schaefer has not reported relevant validity data for the Hostility and Considerateness factors.

Teachers were asked whether the child had been referred for special help for emotional or behavioral problems during the school year or whether the child was receiving such help at the time of the interview.

Results

Means and standard deviations for all AML items are shown in Table I. Separate Group x Sex multivariate analyses of variance (MANOV A) were run for A and M items. A Group x Sex ANOV A was used for the L item. In each analysis the unique effects of the fact.ors were obtained by alternating the order in which the factors were entered, as recommended by Applebaum and Cramer (1974). This was necessary because the cell sizes used in the analyses were unequal.

A significant, F(5, 123) = 2.873, p < .03, MANOV A was found for group on A items, indicating more acting-out problems for control children. Follow-up univariate tests indicated the control children were more impulsive (p < .004), obstinate (p < .007), restless (p < .004), disruptive (p < .019), and more often involved in fights (p < .007). The MANOV A for sex was also significant, F(5, 123) = 4.787, p < .001. Boy! were more often involved in fights (p < .001) and were more restless (p < .001), disruptive (p < .001), obstinate (p < .020), and impulsive (p < .001). The Group x Sex interaction was not significant, F(5, 123) =0.574 p < .75.

Differences were not as great on the M items, although all rating means favored program children. The group MANOV A approached significance, F(5, 124) = 2.140, p < .08. The MANOV A for sex was not significant, F(5, 124) = 1.664, p < .150, nor was the Group x Sex interaction,F(5,124)=1.747,p<.13.

The data distributions violated the MANOV A homogeneity of variance assumption: and were positively skewed. The effect of these viola- tions on the results is not known (Harris, 1975). The data were reanalyzed utilizing two methods of data transformation-square root and base 10 logarithm-but the data remained skewed. The analysis on the untransformed data was the most conservative and was therefore retained for this report. This method of reporting the results is consistent with that followed by other researchers using the AML (Cowen et al., 1973; O'Brien & Bennett, 1983). There were no significant differences for the L item.

MANOV A was also used in the analysis of the CBI results for scales that were highly correlated. The means and standard deviations are shown in Table II. There were no problems of non normal distribution for these data. The Hostility and Considerateness scales showed significant group differences, F(2, 133) = 4.01, P < .02, and sex differences, F(2, 133) = 6.30, P < .002. Program children were rated as being more Considerate and less Hostile. On the Extroversion and Introversion scales there were no significant differences. On Task-orientation and Distractibility a significant MANOVA for sex was found, F(2, 133) = 7.91, P < .001, with girls being more Task-oriented and less Distractible. ANOVAs were used for the analyses of the results for Intelligent Behavior and Dependency. There were no significant findings for the former and for the latter, only the Group x Sex interaction was significant, F(I, 135) = 5.44, P < .03. In the tests for simple effects within sex, control boys were significantly more dependent than program boy& (p < .05). The girls did not differ significantly.

Teachers reported that one program child and four control children, all boys, had been referred for special help for emotional or behavioral problems. Although the control children were four times as often referred for special help, the overall number of children referred was so small that the group Differences were not significant.

DISCUSSION

The results obtained from two measures of child behavior completed by teachers are in general agreement that children who were in the PCDC program presented fewer acting-out, aggressive behavior problems than control children. These results are also consistent with those found for mother reports of child behaviors several years earlier (Johnson & Bieckenridge, 1982).

Differences between groups on moody, withdrawn behaviors approached but did not achieve significance. These.results support the findings for acting-out, aggressive behaviors and suggest that a general reduction in behavior problems had occurred. There were no differences in teacher-reported learning problems. This result is in accord with a result obtained in an analysis of other Houston PCDC school data in which no differences in school grades were found (Johnson & Walker, 1985). However, this analysis did find significant differences in school achievement test results, with program children obtaining significantly higher Iowa Test of Basic Skills Composite scores.

The PCDC is a comprehensive program involving families in many activities, providing a broad range of information and including specific training in child management and cognitive and social stimulation. It also serves as a social ~ort program. It is not possible to point to any particular aspect of the program as the key primary prevention element, but there are some clues to the sources of the program's effectiveness. Breckenridge (1980) used videotaped mother-child interaction sessions at child age 2 to predict child behavior problems at 4 to 7 years of age. He found that maternal high criticism, low affectionateness, low use of praise, and rigid control were related to follow-up measures, based on interviews with the mothers, of child high destructiveness, high attention-seeking, and high restlessness. It has been shown that the program has had a significant effect on these same maternal variables. Thus, it appears that the program mothers' greater affectionateness, praise, appropriate control, and possibly, encouragement of child verbalization, have either had early effects on the children, which have persisted through the years, or that these early maternal behaviors were continued and have had concurrent effects on the children. Data are not now available to resolve the issue of the source of preventive effects, but another follow-up of the PCDC families is underway which includes an investigation of the ongoing parent-chi1d affective relationship. It will provide information bearing on this issue. Quite likely, however, the effects have been carried forward in a transactional manner with both mother and child affected by the program and having continuing effects on each other .

Although there have been few studies that have attempted to predict child behavior problems from early parent-chi1d interactions, two have reported results that are relevant to the present study. Klein and Durfee (1979) found that secure attachment at age 1 predicted social compliance in. the preschool years. Lewis, Feiring, McGuffog, and Jaskir (1984) reported that insecure attachment at age 1 was predictive of behavior problems at age 6, for boys but not for girls. Although attachment was not measured explicitly, our videotaped mother-child interaction measures of affectionateness, praise, criticism, and control are comparable measures of the mother-child affectional relationship and appear to have had similar effects. A warm, secure early relationship that continues undisturbed is expected to promote trusting, caring relationships with other people. In the school situation, this was perhaps evidenced in more positive and compliant social relationships with other children and teachers. The finding that program boys were less dependent than control boys implies that this harmonious relationship was not at the expense of child autonomy.

Statistically significant differences between groups were obtained, but a question remains as to the clinical significance of the differences. One way to assess this is to examine the clinical experience of this sample of children. Only five of the entire group of children had actually been referred for help with emotional or behavioral problems and four of the five were making use of the referral. These five children had mean AML combined scores of 33.20, a considerably larger score than the mean AML of 18.65 for the entire group. Using the lowest AAML total score from the referred groups, whaich was 23, as a cutoff score to identify "potential referrals." 5 program children and24 control children were identified. The group difference was significant, X²(1,N = 139) = 4.96, p < .03. This result suggests that if referral resources had been readily available, and teyt were not, that another 29 children would have been referred. Further, PCDC children would have required referral less often. Support for this rationale may be found in a study by Durlak (1978) of children identified by teachers using the AML for inclusion in two types of school-based interventions for problem behaviors. The mean pre-intervention AML total scores for the entire sample was 34.01, a score that is remarkably similar to the mean score obtained for referred children in the present study, 33.2. Although resolution of the issue of clinical significance requires extensive clinical assessment of children in the project, and this was not done, the analysis of referrals des suggest that the PCDC program has had effects that are not only statistically significant, but clinically significant in that the number of children showing problematic behaviors as perceived by teachers was reduced.

The results suggest that other early childhood programs, especially those designed to have a strong, positive impact on the parent-child relationship, should be examined for possible primary prevention effects on behavior problems.

REFERENCES

Andrews, S. R., Blumenthal, J. B., Johnson, D. L., Kahn, A. J., Ferguson, C. J., Lassiter, T. M., Malone, P., & Wallace, D. B. (1982). The skills of mothering: A study of Parent- Child Development Centers. Monographs of the Society for Research in Child Development,. 47(6, Serial No. !98).

Applebaum. M. I., & Cramer, E. M. (1974). Some problems in the nonorthogonaJ analysis or variance. Psychological Bulletin, 8/. 335-343.

Breckenridge, J. N. (1980). Predicting child behavior problems from early mother-child interaction. Unpublished master's thesis, University or Houston. Houston.

Brid8eman, B., Blumenthal. J. B" & Andrews. S. R. (1981). Parenl-Child Development Centers: Final evaluation report. Unpublished manuscript Educational Testing Service. Princeton, NJ.

Cowen, E., Dorr, D., Clarfield, S., Kreling, B., McWilliams, S. A., Pokracki, R., Pratt, D. M., Terrell, D., & Wilson. A. (1973). The AML: .a quick-screening device for early identification or school maladjustment. American Journal OfCOmnluniry Psychology, /, 12-35

Durlak, J A. (1978, August). Behavioral versus relationship preventive intervention for high-risk school children. Paper presented at the meeting or the American Psychological Association, Toronto.

Gray, S.. & Wandersman, L. (1980). The methodology of home-based intervention studies. Child Developmenl, 5/, 993-1009.

Harris, R. J. (1975). A primer of multlivariate statistics. New York. Academic Press. Johnson, D. L. (1975). The development or a program for parent-child education among Mexican-Americans in Texas. In B. Z. Friedlander, a. M. Sterritt, & a. E. Kirk (Eds.), Exceptional infant (Vol. 3). New York- Brunner/Mazel.

Johnson, D. L., & Breckenridge, J. N. (1982). The Houston Parent-Child Development Center and primary prevention of behavior problems in young children. American Journal of Community Psychology, /0, 305-316.

Johnson, D. L., Breckenridge, J. N., & McGowan. R. J. (1984). Home environment and early mental development in Mexican-American children. In A. W. Gottfried (Ed.), Home environment and early menial development (pp. 151-195). New York: Academic Press.

Johnson, D. L., Kahn. A. J., & Leler, H. (1976). Houston Parent-Child Development Center, Final Report (ERIC Document Reproduction Service No. ED 135 459)

Johnson. D. L., & Walker, T. (1985, April). A follow-up evaluation of the Houston Parent- Child Development Center: School performance. Paper presented at the meeting or the American Educational Research Association, Chicago.

Klein, R. P., & Durfee, J. T. (1979). Prediction or preschool social behavior from social-emotional development at one year. Child Psychiatry and Human Development, 9, 145-151.

Lewis, M., Feiring, C., McGuffog, C., & Jaskir, J. (1984). Predicting psychopathology from early social relations. Child Development, 55, 123-136.

Loeber. R. (1982). The stability or antisocial and delinquent child behavior: A review. Child Development, 53, 1431-1446.

O'Brien, P., & Bennett. B. (1983, April). The construct validity. of the AML as a quick screening measure of children's behavioral adjustment. Paper presented at the meeting or the Society for Research in Child Development, Detroit.

Rickel, A. U., Eshelman, A. K., & Loigman, a. A. (1983). Social problem-solving training: A follow-up study or cognitive and behavioral effects. Journal of Abnormal Child Psvchology, 1/, 15-28.

Rutter, M, & aarmezy, N. (1983). Developmental psychology and psychopathology in children. In P. H Mussen. (Ed.), Handbook of child ps}.chology (VoI. 4. pp. 775-911). New York: Wiley.

Schaefer. E. S. (1975, April). Major replicated dimensions of adjusrment and achievement: Cross-seclional and longitudinal research. Paper presented at the meeting or the American Educational Research Association. Washington, DC.

Schaefer, E. S., & Edgerton, M. (197 Classroom Behaviar /nventory. Unpublished manuscript.

Schweinhart, L J.. & Weikart. D. 1983) The effects of the Perry Preschool Program on youths through age 15-A summary. In Consortium for Longitudinal Studies. As the twig is bent: lasting effects of preschool programs (pp 71-101). Hillsdale, NJ. Lawrence Erlbaum Associates.

Shure, M B., & Spivack. .a. (1978). Problem-solving techniques in child rearing:. San Francisco: .Jossey-Bass.

Acknowledgements

We thank the administrators and teachers of the Houston Independent School District and Joe Carbonari. Eugene Doughtle. Lynn Rehm, and Richard Rozelle for their thoughtful comments on the manuscript. The Parent-Chi1d Development Center was originally supported by grants from the Office 0f Economic Opportunity, Office of Child Development and the Agency for Children, Youth, and Families.

All correspondence should be sent to Dale L. Johnson, Department of Psychology, University of Houston, 4800 Calhoun, Houston, Texas 77204-5341

_______________________________________________________________________________________________

Table 1

Means and Standard Deviations for AML Scores of Program and Control Children

_______________________________________________________________________________________________

Program (N = 51) Control (N = 88)

AML Items Boys (n = 26) Girls (n = 25) Total Boys (n = 41) Girls ( n = 47) Total

_______________________________________________________________________________________________

Acting-Out

Fights 1.72 1.35 1.55 2.40 1.58 1.99

(0.68) (0.57) (0.65) (1.17) (0.85) (1.10)

Restless 1.88 1.43 1.67 2.63 1.72 2.15

(1.01) (0.66) (0.88) (1.13) (0.88) (109)

Disrupts 1.64 1.26 1.46 2.28 1.49 1.90

(0.64) (0.54) (0.62) (1.45) (0.83) (1.24)

Obstinate 1.32 1.17 1.25 1.88 1.42 1.64

(0.63) (0.49) (0.56) (1.07) (0.70) (0.92)

Impulsive 1.76 1.26 1.51 2.38 1.72 2.02

(0.97) (0.54) (0.82) (1.15) (0.98) (1.10)

Moody, Withdrawn

Coaxed 1.44 1.29 1.37 1.50 1.35 1.45

(1.12) (0.86) (0.99) (0.75) (0.69) (0.76)

Unhappy 1.32 1.63 1.46 1.73 1.49 1.61

(0.56) (0.58) (0.58) (0.88) (0.55) (0.73)

Becomes Sick 1.24 1.25 1.25 1.40 1.21 1.31

(0.83) (0.68) (0.75) (0.71) (0.67) (0.71)

Feels Hurt 1.72 1.83 1.78 2.08 1.95 2.02

(0.84) (1.01) (0.92) (0.94) (0.90) (0.91)

Moody 1.60 1.21 1.41 2.03 1.67 1.84

(0.96) (0.41) (0.76) (0.89) (0.84) (0.87)

Learning Difficulty 1.88 1.73 1.98 2.50 1.77 2.17

(1.05) (1.10) (1.25) (1.18) (1.05) (1.24)

_______________________________________________________________________________________________

Multivariate F(5, 123) = 2.873, p < .03 (Group); (5, 123) = 4.787, p < .001 (Sex).

_______________________________________________________________________________________________

Table 2

Classroom Behavior Inventory Means and Standard Deviations

_______________________________________________________________________________________________

Program (N = 51) Control (N = 88)

CBI Items Boys (n = 26) Girls (n = 25) Total Boys (n = 41) Girls ( n = 47) Total

_______________________________________________________________________________________________

Extroversion 18.41 17.21 17.89 18.59 19.38 19.01

(3.71) (5.12) (4.40) (3.49) (3.93) (3.73)

Introversion 5.54 5.88 5.70 5.88 6.21 6.06

(2.44) (3.27) (2.85) (2.50) (3.11) (2.83)

Task Oriented 16.65 17.16 16.90 15.07 19.34 17.35

(6.22) (7.05) (6.58) (5.52) (4.29) (5.32)

Distractible 7.73 6.16 6.96 8.68 6.34 7.43

(3.37) (3.39) (3.62) (2.85) (2.57) (2.93)

Hostility 4.85 4.20 4.53 6.68 5.09 5.82

(1.97) (2.66) (2.34) (3.19) (2.79) (3.06)

Considerate 19.35 20.20 19.77 16.88 20.66 18.90

(4.19) (5.89) (5.06) (4.61) (3.49) (4.45)

Intelligent 16.31 15.00 15.67 14.81 16.36 15.64

(5.14) (6.21) (5.67) (4.42) (4.48) (4.50)

Dependency 5.31 6.04 5.67 6.71 5.15 5.88

(2.62) (3.99) (3.35) (2.60) (2.21) (2.51)

_______________________________________________________________________________________________

Task Oriented, Distractible: Multivariate F(2, 133) = 7.91, p < .001 (Sex)

Hostility, Considerate: Multivariate F(2, 133) = 4.01, p < .02 (Group); Multivariate F(12, 133) = 6.30, p < .002 (Sex).

Dependency: Univariate F(1, 135) = 5.44, p < .03 (Group x Sex)

_______________________________________________________________________________________________