Revised version of School Paper-PCDC

A Follow-up Evaluation of the Houston Parent-Child Development Center: School Performance

The evaluation of compensatory early childhood education programs reached a peak with the publication of the results of the Consortium on Developmental Continuity (Darlington, Royce, Snipper, Murray, & Lazar, 1980; Lazar, Darlington, Murray, Royce, & Snipper, 1982; Consortium for Longitudinal Studies, 1983). These reports dealt with the long-term evaluation of the effectiveness of the twelve early childhood education programs that constituted the first generation of carefully planned interventions to promote the school competence of low-income and minority children. The results reported are highly complex and, because of problems inherent in the original research designs for many of the projects, remain equivocal on some issues. Overall, however, the programs were reported to have been successful in reducing the retention in grade of students and reducing the rate of referral for special educational services.

In 1970 a second generation of early childhood programs appeared. Their designs were based, in part, on what seemed to be the best features of the first generation programs. One of these was the Houston Parent-Child Development Center (PCDC), a two-year, parent-child education program for low-income Mexican American families. Goals for the program were developed through door-to-door surveys of families in Houston barrios about their goals for their children and suggestions for ways to attain the goals, focus groups of representatives of the community, consultation with local and national authorities on Mexican American families and child rearing, and a search of the relevant literature. The overarching goal that emerged was to promote the development of child competence through strengthening parental child rearing skills. It was theorized that parents would retain these skills and continue to provide a supportive and educationally stimulating environment for their children. Subordinate goals included optimizing later school performance and preventing the emergence of behavior problems. A major feature of the program was that parent and child would participate together, thus, the program is properly considered a parent-child education program. The development of the program is discussed in Johnson (1975).

The Parent-Child Development Center Program

Beginning at child age one, the first year of the program was in the home. There were 25 visits by a paraprofessional educator in which the mother was provided with information about how to understand her infant's behavior and how to stimulate cognitive, social and language development. The emphasis was on developing the mother's skills with her own resources for teaching her infant. For example, one session was on the value of children's folk songs for promoting child language development. Mothers were helped to recall songs learned in their own childhoods and to teach these to their children. Each in-home session began with a review of the preceding session and of homework assigned (e.g., keeping a record of all of the child's new words) and concluded with a new homework assignment.

In order to involve fathers, entire families attended several weekend sessions on decision-making and communication in families. These sessions used a human relations training format.

During the second year, mother and child attended the project center four mornings each week and participated in classes on child management, health and safety in the home, child cognitive and language development, and other related topics. These classes made use of a variety of educational procedures, but throughout, a mix of conceptual presentations and practice with feedback was used. On such topics as teaching concepts to young children, mothers videotaped their practice sessions with their children and played these tapes back to the other mothers for feedback. Most of the sessions were in a group setting and were designed to maximize discussion by the mothers. Mothers participated in English language classes if they wished. Fathers continued their participation through monthly evening meetings on topics of their own choosing. These included such matters as home-buying procedures, interacting with schools, and family planning.

The entire two-year program had a scheduled 550 hours of participant time; however, 400 hours was the actual mean time in program activities. Child illness was the main cause of missed sessions. All mothers participated in most of the first- and second-year programs including the English-language classes and about two-thirds of the families took part in the family weekend sessions on a regular basis. About three-fourths of the fathers were regular participants in the second year evening sessions. Fathers received no material incentives; however, participation was encouraged through discussions with a male staff member who had strong ties to the community.

A 12-volume set of curriculum materials was prepared by and for the staff. Cultural relevance for curriculum materials was assured by having bilingual Mexican-American staff members and consultants involved in all stages of curriculum and program development and implementation. Professionals provided supervision and developed curriculum materials. Paraprofessionals conducted most of the actual training. The program has been described in detail elsewhere (Andrews, Blumenthal, Ferguson, Johnson, Kahn, Malone, & Wallace, 1982).

Earlier Evaluations of the PCDC Program

Families were assigned randomly to program or control groups. Annual assessment of program effectiveness was carried out using a wide range of procedures. At the time of program completion, it was demonstrated that the program was effective in enhancing mother warmth, use of praise, encouragement of child verbalization and provision of a cognitively stimulating home environment (Andrews et al., 1982). In addition, program children had higher Stanford-Binet IQs, but no significant differences were found on the Concept Familiarity Index (Bridgeman, Blumenthal & Andrews, 1981).

The Houston PCDC included most of the features that the Consortium later found to be characteristics of successful programs. These were beginning the program at an early age, high levels of parental involvement, inclusion of goals for parents as well as children, and a low adult-child teaching ratio. All of this suggests that the Houston PCDC should be successful in attaining its goal of optimizing child scholastic performance. This paper is a first report of the PCDC's school follow-up results.

Method

Subjects

The project neighborhoods selected were those that according to census data had the lowest family income, lowest adult level of education, and greatest number of Spanish surnames. Families were assigned initially to program or control groups as follows: A door-to-door survey was conducted to identify low-income Mexican American families with a one-year-old child who was healthy and had no apparent neurological problems. Eligible families were told about the project in detail. They were told that there was a "program group" and "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 random assignment was made. Control group families received no services, only annual assessment procedures.

School data were collected when the children were in the second through fifth grades and were from 8 to 11 years of age. The children had been in five different PCDC cohorts and represent a 56% sample of the total number of families in these cohorts. Other children could not be located, largely because families had moved from the area. The search process began with the Houston Independent School District (HISD), a very large metropolitan school system. When children were not located through HISD records the search was extended to the other 16 school districts in Harris County and to the parochial schools. Telephone directories were examined for parental addresses and neighbors of families who had moved were asked for information about the present location of families.

Statistical analyses were conducted to determine whether differential attrition had affected the comparability of groups at follow-up. A series of condition (program/control) by follow-up status (located/not located) ANOVAs were performed on the family background characteristics of parental age, education, and language usage, marital status, number of children, and family income. The statistical significance of the interaction term of each ANOVA was examined to determine if sample attrition had occurred differentially across condition and time. No statistically significant interactions were found. There had been no differences between groups on these variables at intake or for families located at the end of the program. Thus, the original random assignment of families to experimental condition yielded comparable groups and there was no evidence of differential attrition either at the end of the program or at the follow-up time. Data for these analyses were collected at intake only. See Table 1 for demographic characteristics of families who were located or not located for program and control groups.

----------------------------

Insert Table 1 about here

----------------------------

Although the number of program and control children was approximately equal at intake, more control children were available at the time of the follow-up owing to differences in criteria for follow-up inclusion. Completion of the two year program was required of program children, but all control children whose families completed intake data collection procedures were included even if end-of-program data collection was absent. For the follow-up, boys and girls were represented nearly equally. The Ns (see Tables 2 & 3) for the various school measures varied from 138 for grades and the Classroom Behavior Inventory (CBI; Schaefer & Edgerton, 1976) to 115 for Iowa Tests of Basic Skills (ITBS) (Lindquist & Hieronymus, 1956) achievement test scores. The number of children available for the ITBS was smaller because the ITBS was administered only to children judged by the teachers to be fluent in English, the language of the test. Enrollment in bilingual classes did not automatically preclude taking the ITBS. Children were located at 40 schools. In only three instances were two project children found in the same classroom.

Procedures

Grades were recorded for mathematics, reading, spelling, and language at the end of the school year. Letter grades were assigned a numerical code ranging from 1 to 5 in ascending order with a code of 1 representing an F and 5 an A. A total grade score was computed by summing the scores for the four individual subjects (range of the total grade score: 4 - 20).

The ITBS consists of seven subtests and a Composite score. The subtests are Vocabulary, Reading, Language, Spelling, Arithmetic Concepts, Arithmetic Problem Solving, and Arithmetic Total. Percentile scores were collected from the student files because neither raw scores nor standard scores were available. The percentiles were converted to z-scores prior to submitting them to statistical analyses.

The Classroom Behavior Inventory (Schaefer, 1975) is a 32-item inventory which includes scales for Hostility, Considerateness, Task Orientation, Distractibility, Extraversion, Introversion, Independence, Dependence, and Intelligent Behavior. Teachers were instructed to rate each child on a 4-point scale ranging from 1 (Not at all like) to 4 (Very much like) for each item. The author reported median interrater reliability coefficients ranging from .42 for Introversion to .68 for Hostility. The CBI was completed by teachers at the end of the school year. At the same time, teachers were interviewed about their intention to promote or retain the child, the presence of behavior problems, referral to special classes, whether the classroom was bilingual or monolingual English, and the number of contacts the teacher had with the parents during the year. Information was also obtained from school records as to whether the child had ever been retained in grade or had ever been in a bilingual program.

Results

The data collected on school grades, ITBS, and CBI were analyzed using a series of group by gender ANOVAs. Because cell sizes were unequal, the order of entry of the factors was alternated in order to test the main effect of each factor while controlling for any confounding effect of the other factor. This procedure was recommended by Applebaum and Cramer (1974). Gender was included in the ANOVAs because this was found to be important in an earlier PCDC follow-up (Johnson & Breckenridge, 1985) and because gender differences commonly appear in analyses of school grades and less consistently for achievement test scores (Keeves, 1985).

Grades

The means and standard deviations for the individual grades and the total grade score appear in
Table 2.

_____________________________________

Insert Table 2 about here

____________________________________

No differences were found between the program and control groups for any of the individual subject grades or for the total grade score. In addition, none of the group by gender interactions were significant.

Gender differences, favoring girls, were found for the total grade score (F(1,133) = 5.21, p < .025), and for the subjects of Reading (F(1,133) = 4.46, p < .04), Language (F(1,133) = 8.30, p < .005), and Spelling (F(1,133) = 3.91, p < .05).

Achievement Tests

The ITBS percentile scores were converted to z-scores for all statistical analyses. In order to facilitate the interpretability of the achievement scores, the means and standard deviations of the nontransformed data are shown in Table 3.

______________________________

Insert Table 3 about here

_____________________________

Utilizing the group by gender ANOVAs, significant group differences favoring the program group were found on three of the ITBS verbal scales and on the Composite score
(F(1, 107) = 5.54, p < .02). The mean percentile on the Composite score was 57.7 for the program group and 46.4 for the controls. In addition, the program group scored higher on the Vocabulary (F(1, 107) = 7.40, p < .01), Reading (F(1, 107) = 6.72, p < .02), and Language (F(1, 107) = 5.7, p < .02) scales. The group main effect was not significant for Spelling or any of the Arithmetic scales.

Four of the scales revealed gender differences. Girls scored higher than boys on the Composite score (F(1, 107) = 4.29, p < .05) and three of the individual scales: Language (F(1, 107) = 8.23, p < .005), Spelling (F(1, 107) = 4.26, p < .05, and Arithmetic Total (F(1, 107) = 5.68, p < .02).

No significant group by gender interactions were found.

Classroom Behavior Inventory

Table 4 lists the means and standard deviations for the CBI scales. The two-way ANOVAs performed with the CBI revealed a program effect for one of the scales, Hostility (F(1, 134) = 7.67, p < .006). Students in the control group were rated as being more hostile than those in the program group.

____________________________

Insert Table 4 about here

____________________________

Gender differences were found for three of the scales, Distractibility, Hostility, and Consideration. Boys were perceived as being more distractible (F(1, 134) = 14.93, p < .001), more hostile (F(1, 134) = 6.94, p < .01), and less considerate (F(1, 134) = 12.47, p < .001) than girls.

The Task Orientation and Dependency scales showed significant group by gender interactions. Post hoc pairwise comparisons examining group differences analyzed by gender were performed to investigate further the effects of the factors on Task Orientation and Dependency. Program boys were rated as significantly less dependent (t(64) = 2.14, p < .04) than control boys. No differences in Dependency were found for girls. In addition, no group differences were revealed for Task Orientation for either gender.

Retention in Grade and Special Class Participation

Chi square analyses were conducted to determine whether differences existed in retention in grade and referral to special resources (See Table 5). Although on virtually all items, program children seem to be doing better in school, none of the differences were statistically significant. There were no gender differences on these measures.

______________________________

Insert Table 5 about here

_____________________________

Differential participation in bilingual classes was also examined using chi square. Significantly fewer program children were reported to be attending bilingual classes at the time of the teacher interview (X² (1) = 6.60, p < .01). Only 14% of the program children were in bilingual classes compared with 36% of the control group children. There was no difference in bilingual class attendance in kindergarten or first grade. In addition, no difference was found between groups in percentage of children who were English dominant at age three.

Groups did not differ in the number of teacher-reported contacts with parents during the school year. Means were 6.6 for the program families and 6.4 for families in the control group.

Discussion

Grade point averages were used in this program evaluation because they seemed to offer a solution to the problem that is inherent in the use of achievement test scores in the evaluation of bilingual children: the results may be influenced by the English language competence of the children. Grades are not as subject to that influence, that is if the teachers are bilingual and sensitive to the child's linguistic status, but we discovered in the process of data collection that grades present other problems. We found that in some instances children were given rather high grades even though they were doing poor work in school. The teachers were using the grades to reward "trying" with children who they believed were not actually capable of better performance. Under these circumstances the absence of group differences is not surprising. Nevertheless, grades are one indicator of school performance and not entirely whimsical. We obtained a correlation of .52 between total grade score and ITBS composite score and did find significant gender differences in grades favoring girls on language-related subjects.

Significant differences favoring the program group were found on the ITBS Composite score and on all of the verbal achievement test scores except spelling. No group differences were found on measures of arithmetic achievement. These results represent partial goal attainment for the PCDC since enhanced school competence was a program goal, but it was expected that program superiority would be found in all areas of school achievement, not just in the verbal area. Why the program had effects on verbal, but not on arithmetic achievement is not obvious and considering the broad scope of the program the search for an answer may be meaningless. However, it is possible that the skills demonstrated by program mothers in encouraging their children's verbal behaviors at the end of the program (Andrews et al., 1982) may have lasting effects, or more likely, the mothers may have continued to practice these skills after completing the program with a resulting emphasis on language behaviors by their children.

Although the PCDC and the Consortium projects are all early childhood programs, they also vary greatly. The major differences are in the age of the child at the time of intervention (most of the Consortium children were older) and whether or not parents were the focus of training (Parents were involved in a major way only in the Gordon, Levenstein, and Weikart programs). Given these differences, it is perhaps most appropriate to compare the results of the PCDC with those reported for the Gordon Parent Education Infant and Toddler Program (Jester & Guinagh, 1983). This program was similar to the PCDC in that it also included infants, and parents were involved. Their follow-up when the children were in the 2nd to 4th grades found significant group differences on the Metropolitan Achievement Test for Reading, Math Concepts, and Math Problem Solving. They did not find differences on the Classroom Behavior Inventory. The Perry Preschool Program (Schweinhart & Weikart, 1983) also resembles the PCDC in that both home visiting and group training programs were used and the mothers were involved. However, the children did not begin the program until they were two. The follow-up achievement test results showed no group differences through age 11. At age 14, significant differences were obtained on the California Achievement Test. The PCDC results seem to be at least as favorable as those obtained for the two most comparable Consortium programs.

A surprise in the PCDC evaluation is that the children were doing so well in elementary school. Although the program children were doing better than the controls on some measures, both groups were performing rather well considering their low socioeconomic status. The overall percentile level for the Iowa Composite was 50.1; precisely at the normative mid-point. This average level of success is also reflected in their C+ overall grade average. To check on whether project children were performing at a level typical of all children in the area of Houston included for the PCDC a survey was done of representative schools. The eight elementary schools having the highest proportion of project children were selected. ITBS Composite scores for three years and three grades were pooled. The overall mean Composite percentile was 48, slightly below that found for the project children. The eight schools selected are located in working class neighborhoods, but it is unlikely that the majority of families would meet poverty guidelines. Virtually all project families had incomes below the poverty level at the beginning of the PCDC project. Thus, it appears that the control group children were doing about as expected for a low socioeconomic group and program children were functioning at above the expected level on the ITBS.

The percentage of children ever retained in grade was also lower for the PCDC than for programs in the Consortium using a near-random design. Their median percentages were 25.8 for program children and 36.6 for control children. The differences probably reflect differences in local school system promotion practices, but may indicate that the PCDC project children overall were doing better in school.

This project demonstrated that a parent-child education program can have long-lasting effects on child competence which are manifest in school performance. It was assumed at the beginning of the project that this would be the case and that the lasting effects would be a function of parental involvement in the program. It was also assumed that parents would, as a result of the intervention, continue to be effective teachers of their children. We now know from the Consortium reports that early childhood programs that do not involve parents also may have long-lasting positive effects on school performance.

Parent education may, nevertheless, have other important consequences and PCDC results not reported here seem to support this. For example, Guzman and Johnson (1984) found that PCDC program children offered more "work" responses when presented with difficult tasks than control children. Johnson and Breckenridge (1982) found that PCDC children were less likely to present behavior problems as reported by mothers. Teachers reported significantly fewer behavior problems for program than for control children (Johnson & Walker, 1987).

These results suggest that the PCDC has had broad influences on social and academic competence. The CBI results reported here indicate that the PCDC youngsters are viewed by teachers as less hostile in the classroom and, therefore, quite likely viewed as more desirable students. The combination of social acceptability and strong academic performance for the program children suggests a general competence. We have some evidence that a source of this competence is in the development of an early harmonious parent-child relationship, one that provides a secure base for the child from which he or she can range to explore the environment around. Breckenridge (1980) found that child behavior problems including resistiveness, destructiveness, and high activity in the preschool years were related to low maternal warmth and high maternal criticism at child age two. Also working with PCDC data, Carrillo (1988) obtained results indicating that maternal warmth and lack of criticism at child age three predicted problem behaviors in elementary school. Other researchers (e.g., Lewis, Feiring, McGuffog, & Jaskir, 1984) have reported similar results. McGowan and Johnson (1984) have reported a causal model for school achievement test scores that suggests that a stimulating early home environment together with affectionate and encouraging parent-child relationships are conducive to school success. The results of the present study suggest that these family background factors, enhanced by participation in the PCDC, may also have contributed to greater school achievement.

Although the compensatory early childhood education programs developed in the 1960s and 1970s included in their goals the eventual enhancement of school performance, the results have been mixed (Consortium, 1983). With the positive results of the present follow-up, conducted 5 to 8 years after a two-year parent-child education program, the Houston PCDC joins comparable Consortium programs in demonstrating long-term effects on school performance. Nevertheless, generalizations to other early childhood programs can be made only with caution: Other consortium programs did not have long-term effects on school performance, and the Houston-PCDC results must be considered in the context of a relatively low follow-up recovery rate. More controlled research is needed before we can be confident about the effects of early childhood programs with low-income families.

References

Andrews, S.R., Blumenthal, J.B., Ferguson, C.J., Johnson, D.L., Kahn, A.J., Lasater, T.M., Malone, P.E., & Wallace, D.B. (1982). The skills of mothering: A study of the Parent-Child Development Centers. Monographs of the Society for Research in Child Development, 47, (6, Serial No. 198).

Applebaum, M. I., & Cramer, E. M. (1974). Some problems in the nonorthogonal analysis of variance. Psychological Bulletin, 81, 335-343.

Breckenridge, J. N. (1980). Predicting child behavior problems from early mother-child interaction. Unpublished masters thesis, University of Houston.

Bridgeman, B., Blumenthal, J.B., & Andrews, S.R. (1981). Parent-Child Evaluation Center: Final Evaluation Report. Educational Testing Service, Princeton, NJ.

Carrillo, J. (1988). Predicting behavior problems in elementary school from early mother-child interactions. Unpublished master's thesis. University of Houston.

Consortium for Longitudinal Studies (1983). As the twig is bent: Lasting effects of preschool programs. Hillsdale, NJ: Erlbaum.

Darlington, R.B., Royce, J.M., Snipper, A.S., Murray, H.W., & Lazar, I. (1980). Preschool programs and later school competence of children from low-income families. Science, 208, 202-204.

Guzman, A.M., & Johnson, D.L. (1984). The responses of Mexican-American children to cognitive demands in a testing situation. Hispanic Journal of the Behavioral Sciences, 6, 261-275.

Jester, R.E., & Guinagh, B.J. (1983). The Gordon Parent Education, Infant and Toddler Program. In Consortium for Longitudinal Studies, As the twig is bent: Lasting effects of preschool programs (pp. 103-132). Hillsdale, NJ: Erlbaum.

Johnson, D. L. (1975). The development of a program for parent-child education among Mexican-Americans in Texas. In B. Z. Friedlander, G. M. Sterritt, & G. E. Kirk (Eds.), Exceptional infant: Vol. 3 (pp. 374-398). 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, 10, 305-316.

Johnson, D.L., & Walker, T. (1987). The primary prevention of behavior problems in Mexican-American children. American Journal of Community Psychology, 15, 375-385.

Keeves, J. P. (1985). Sex differences in ability and achievement. In T. Husen & T. N. Postlewaite (Eds.) International encyclopedia of education (Vol. 8, pp. 4539-4544. London: Pergamon Press.

Lazar, I., Darlington, R., Murray, H., Royce, J., & Snipper, A. (1982). Lasting effects of early education: A report from the consortium for longitudinal studies. Monograph of the Society for Research in Child Development, 47, (2-3, Serial No. 195).

Lazar, I., Hubbell, V.R., Murray, H., Rosche, M., & Royce, J. (1977). The persistence of preschool effects. DHEW Publication No. (OHDS) 78-30130. Washington, DC: U.S. Government Printing Office.

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

Lindquist, E. F., & Hieronymus, A. N. (1956). Iowa Test of Basic Skills. Boston: Houghton Mifflin.

McGowan, R.J., & Johnson, D.L. (1984). The mother-child relationship and other antecedents of academic performance: A causal analysis. Hispanic Journal of the Behavioral Sciences, 6, 205-224.

Schaefer, E.S. (1975, April). Major replicated dimensions of adjustment and achievement: Cross-cultural, cross-sectional, and longitudinal research. Paper presented at the annual meeting of the American Educational Research Association, Washington, DC.

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

Table 1

Demographic Characteristics of Families

Program Control

Not Not

Followed Followed Followed Followed

Variable

____________________________________________________________

n 50 40 87 114

Years of Education

Father 7.3 7.0 6.9 7.4

Mother 7.4 7.6 7.7 7.2

Age

Father 32.3 30.2 31.0 30.2

Mother 29.1 27.3 27.7 26.8

Number of Children 3.7 3.4 3.2 3.4

Per Capita

Monthly Income $1291 $1458 $1219 $1325

Percent Married 94.3 95.7 90.6 93.1

Percent Mothers

Prefer Spanish 52.4 53.1 50.6 53.3