Report on the study of the Home Inventory (preschool form) in a portuguese sample

1 Report on the study of the Home Inventory (preschool form) in a Portuguese sample Isabel Macedo Pinto Abreu-Lima Universidade do Porto Faculdade de Psicologia e Ciências da Educação SUMMARY The intent of this study was twofold: first and foremost we were interested to analyse the psychometric properties of the HOME Inventory (Caldwell & Bradley, 1984) in the cultural context of Portuguese families and their preschool aged children. In a second step, we focused on the relation between the quality of home environments and results obtained by children on a test of receptive vocabulary (Peabody Picture Vocabulary Test – Revised, Dunn & Dunn, 1981) and a measure of adaptive behaviour (Vineland Adaptive Behaviour Scales, Sparrow, Balla & Cicchetti, 1984). The study of the scale properties was conducted on a total of 407 Portuguese families, all of which participated on the European Child Care and Education Study (ECCE), that was initiated in 1992 by a group of European countries, in which Portugal was included. Its main purpose were to assess the diversity and quality of educational experiences of children from 3 to 6 years of age, in formal and family settings, and the impact of such experiences in their development and quality of life (ECCE Study Group, 1997). The validity of the Home inventory was estimated and a factor structure emerged (which was conceptually very identical to the original one). In a second step, the group of participants was reduced to 215 families, retaining only those that had complete information both on quality of home variables, family background variables and child outcome variables. Variables from the family’s demographic context were used to predict HOME total scores; socio-economic status appeared as the strongest predictor. The predictive value of the Home global score on child’s outcome measures was compared to other relevant variables, such as a socio- 2 economical index. HOME scores were found to be the most significant predictor when results on the cognitive, socialization and language domain were considered. METHOD. 1. Participants. A total of 407 families participated in the study. They came from two regions of Portugal, the North region (surrounding Porto, the second biggest city in the country) and the South region (around Lisbon, the capital). Although it is not possible to consider such a distribution as representative of the whole country, it is a fact the most of the Portuguese population concentrates in these regions, both located on the coast area. The interior of the country, mostly rural, has for the last 25 years been subjected to intense migrations, towards the urban centres and to the coast area. In each region, two zones were selected, metropolitan and non-metropolitan. In the four areas where participants were drawn from, shown in Table 1, live approximately 40% of the country population. Table 1 – Distribution of participant families by area of residence (region x zone) North South Total Metropolitan 159 134 293 Non-metropolitan 64 50 114 Total 223 184 407 All children (209 boys and 198 girls) were caucasian and had no diagnosed disorders or illness. Most of them (81.3%) attended some kind of early childhood center-based education programme. By the time the families were visited and interviewed, the children had between 50 and 63 months (4 years 2 months and 5 years 3 months), the mean age being 55.03 (4 years 7 months). 31 % of the families had only 3 one child, whereas 45% of the children had one sibling and 24% two or more siblings. Only 27 of these families were monoparental. 2. Measures Demographic data. Other information on family background variables were collected, using a structured interview, such as crowding ratio, mother’s age and number of hours mothers spent away from home daily. Peabody Picture Vocabulary Test – Revised (PPVT-R). The Peabody Vocabulary Test was administered to all the children, at home, providing a measure of the child receptive vocabulary. This measure was considered to be an estimate of the child’s verbal intelligence, and thus in the cognitive domain. Vineland Adaptive Behaviour Scales, (ECVA). The Vineland scales were used to assess Children’s adaptive behaviour. A global score was obtained as well as four scores/ domains (resulting from the four dimensions that stood out after factor analysis): Autonomy in daily life activities, Socialization, Language and knowledge and Motor development. EAS Temperament Scale for Parents (Buss & Plomin, 1984). Mothers rated their children on the EAS Temperament Survey. After factor analysis of the data on the sample, four dimensions of temperament were retained for each child, respectively Energy, Sociability, Introversion and Emotionality 1.1. Socio-economic status Data on some of the family background variables such as mother and father’s education and occupation and income per capita were used in the calculation of a socioeconomic index1. Information on these five variables allowed us to compute an estimate of the socioeconomic status (SES) of each family, by attributing a numeric value to categories within each variable and adding them up. Obviously, families in which parents had more years of schooling, more specialized occupations and higher incomes, received higher rankings. 1 This index was quite similar to the Hollingshead index for computing SES. 4 RESULTS a. Psychometric characteristics of the HOME inventory for families of Portuguese preschool age children. Table 2 displays the percentage of the tested families that got credit for each item, which we can assume to reflect its “difficulty level”. Around 33% of the items were quoted positively in 81% or more of the families. Item 19 received the maximum amount of credit, whereas item 37 received the minimum. Table 2 – Percentage of families that passed each item in HOME Item % Item % Item % Item % Item % 1 41.8 12 59.5 23 75.2 34 64.4 45 74. 4 2 44 13 56.8 24 77.6 35 43.2 46 34.6 3 34.4 14 74.4 25 92.4 36 83.8 47 81.8 4 59 15 88.9 26 83.3 37 16.2 48 56.8 5 83.8 16 87.7 27 85.7 38 47.4 49 26.5 6 34.4 17 84.3 28 86.7 39 38.8 50 85 7 59.5 18 69.8 29 86 40 38.6 51 77.6 8 63.1 19 92.9 30 40 41 59 52 65.1 9 25.6 20 62.4 31 58.7 42 64.4 53 89.4 10 53.8 21 80.8 32 26 43 75.2 54 89.9 11 30.7 22 84 33 53.6 44 86.7 55 71.5 Point biserial correlation coefficients were calculated to analyse the discriminative power of the items, and are presented in Table 3. With the exception of items 18, 50 and 51, all items showed a positive correlation with the global score. This anomalous situation led us to exclude the three items referred from further analysis. 5 Table 3 – Point biserial correlation (rpbis) of each item with total score Item rpbis Item rpbis Item rpbis Item rpbis Item rpbis 1 .62 12 .70 23 .63 34 .44 45 .43 2 .68 13 .37 24 .51 35 .35 46 .57 3 .58 14 .52 25 .78 36 .42 47 .33 4 .63 15 .72 26 .66 37 .34 48 .56 5 .84 16 .59 27 .57 3 8 .28 49 .36 6 .68 17 .58 28 .69 39 .16 50 (a) 7 .66 18 (a) 29 .65 40 .40 51 (a) 8 .70 19 .89 30 .31 41 .40 52 .31 9 .50 20 .05 31 .41 42 .32 53 .45 10 .41 21 .65 32 .18 43 .66 54 .34 11 .46 22 .50 33 .33 44 .60 55 .36 (a) Items excluded Tetrachoric correlations among the 52 dichotomous items and the corresponding matrix were estimated with STATISTICA. Over such a matrix, a principal component analysis was conducted, followed by varimax rotation using an eigenvalue cut-off of 1.0. The analysis produced six main factors and several other factors composed of a small number of items, so we limited the analysis to six factors, which accounted for 70.4% of the variance of the results. The six factors and the items that compose them, as well as the item loadings and communality values are presented in Table 4. 6 Table 4 – HOME: Item loadings for the six factors2 Itens h2 F1 F2 F3 F4 F5 F6 8. Dez livros visíveis na casa .882 .926 2. Três ou mais puzzles .871 .907 4. Brinquedos que p ermitem expressão livre .902 .891 7. Dez livros infantis .875 .882 43. Tem instrumentos de música .786 .867 5. Brinquedos que exigem movimentos precisos .847 .861 23. Casa tem 9 m2 por pessoa .824 ,845 3. Gira disco s, gravador e 5 discos infantis .795 .843 24. Divisões não superlotadas .796 .841 19. Edifício parece seguro .809 .839 21. Interior da casa não é escuro .841 .834 12. Brinquedos que ensinam nomes de animais .772 .822 15. Mãe utiliza gramática e dicção correctas .764 .816 6. Brinquedos que ensinam números .781 .814 48. Mãe usa frases complexas .679 .763 1. Brinquedos que ensinam cores, tamanhos, formas .748 .757 45. Fez viagens de mais de 40 km. .640 .732 25.Casa limpa e arrumada .753 .726 46.Cr. levada a um museu .669 .622 9. Família lê jornal .686 .620 49. Trabalhos manuais da cr. expostos .737 .611 44. Cr. levada a passear .490 .549 16. Mãe estimula crian ça a falar e tem tempo p. ouvir .758 .533 10. Família assina uma revista .527 .494 29. Mãe receptiva quando c. fala .808 .865 28. Mãe responde verbalmente à cr. .868 .859 27. Mãe conversa com cr. 2 vezes durante entrevi sta .794 .835 32. Mãe ajuda a cr. a mostrar habilidade .858 .825 17. Mãe transmite sentimentos positivos à cr. ou acerca dela .698 .766 30. Mãe elogia qualidades da cr .730 .765 31. Mãe acaricia ou dá beijos durante a entrevista .744 .730 40. Mãe apresenta a criança .612 .575 14. Mãe ensina à criança bons modos .624 .555 2 Although the Portuguese version of the items is presented, numbers correspond to the original items 7 Table 4 – HOME: Item loadings for the six factors Itens h2 F1 F2 F3 F4 F5 F6 36. Cr. estimulada a aprender números .745 .813 35. Cr. estimulada a aprender relaç. espaciais .733 .801 11. Cr. estimulada a aprender as formas .700 .788 33. Cr. estimulada a aprender cores .591 .691 34. Cr. estimulada a aprender lenga - lengas .540 .648 13. Cr. est imulada a aprender o alfabeto .504 .601 37. Cr. estimulada a aprender a ler palavras .518 .578 26. Mãe abraça a criança algum tempo por dia .613 .497 53. Mãe não utiliza repressão física na visita .850 .805 54. Mãe não bate na cr. durante visita .932 .803 52. Mãe não repreende cr. na visita .790 .757 38. Cr. habituada a respeitar horário na alimentação .605 .629 22. Zona que rodeia a casa é agradável .854 .627 47. Cr. habituada a arr umar os brinquedos .691 .621 20. Zona de brincadeiras ao ar livre é segura .554 .567 39. TV utilizada de forma criteriosa .385 .476 55. Não mais do que um castigo na semana .690 .654 42. Cr. pode bater na mãe sem severa represália .633 .609 41. Cr. pode exprimir sentimentos negativos .629 .562 Valores - próprios ( eigenvalues ) 16.97 6.51 6.33 3.73 3.13 2.06 % de Variância 30.9 11.8 11.5 6.8 5.7 3.8 The first factor accounted for the largest amount of the common variance (31%) and was called Support for Development. It drew from several existing subscales, namely Learning Stimulation, Language Stimulation, Physical Environment and Variety of Experience. It refers to play materials and household equipment that turn the home into a stimulating and challenging environment, as well as the parent’s ability to organize and make available cultural experiences to the child. The interactional dimension appeared in the second factor that accounted for 11.8% of the common variance and grouped most of the observational items, particularly those that expressed a warmth and positive affective relation with child. It corresponded closely to subscale Pride, Affection and Warmth, and was named Positive Interaction. The third factor was responsible for 11.5% of the variance and most of the items reflected a deliberate effort from the adult to cognitively stimulate the child. It 8 overlaped with the original Stimulation of Academic Behaviour scale and was named Academic Stimulation accordingly. The remaining three factors - Absence of Hostility, Structure and Permissive Discipline, appeared to be less robust, in what concerned the percentage of variance they accounted for (6.8, 5.7 and 3.8%, respectively) although conceptually their meaning seemed quite relevant. The items included in the subscales Absence of Hostility and Permissive discipline indicated that the adult was able to model appropriate emotional control and that child was not often punished or beaten for inappropriate behaviour. The items grouped under Structure reflected the existence of rules and routines of family organization that seemed important aspects of family life. Reliability Internal consistency of the results was calculated for the whole scale and for each subscale, using the Kuder-Richardson (20 formula) and are presented on Table 5. The whole scale presented good internal consistency, as well as the first subscale. Table 5 – Internal consistency ( Kuder-Richardson coefficients) for the Home scale and subscales Subscales Number of items K - R (20) 1. Support for development. 24 .95 2. Positive interaction 9 .87 3. Academic stimulation 8 .83 4. Absence of hostility 3 .93 5. Structure 5 .61 6. Permissive discipline 3 .74 HOME – total score 52 .94 Intercorrelations among subscale scores are displayed in Table 6. The coefficients varied from negligible to strong. 9 Table 6 – Intercorrelations among Home subscales 1 2 3 4 5 6 1. Support for development. 2. Positive interaction .38*** 3. Academic stimulation .42*** .28*** 4. Absence of hostility .21*** .14** .05 5. Structure .23*** .17** .20*** .08 6. Permissive discipline .27*** .33*** .12* .21*** .06 HOME – total score .90*** .63*** .61*** .32** .40*** .44*** * p<.05 **p<.01 ***p<.001 Finally, the mean and standard-deviation for each sub-scale and for the whole scale, as well as minimum and maximum values, were calculated and are presented in Table 7. Table 7 – Mean, Standard deviation, maximum and minimum values for HOME Inventory sub-scales and total scale N of items Mean S - D. Mín. Máx. 1. Support for development. 24 15.83 4.92 2 24 2. Positive interaction 9 5.90 2.08 0 9 3. Academic stimulation 8 4.49 1.85 0 8 4. Absence of hostility 3 2.47 .83 0 3 5. Structure 5 3.18 1.20 0 5 6. Permissive discipline 3 2.00 1.03 0 3 HOME – total score 52 33.87 8.05 10 50 16 then 29 years). Mean HOME scores from families whose mothers were aged in between (29-34 years) where not significantly different from the other two groups. The family’s place of residence also showed some relation with quality of home environment. Children from the south of the country had higher mean HOME scores than their peers from the north; families in urban areas also tended to receive higher scores than those from non-urban areas. The contrast north/south was a significant predictor of HOME scores, as referred. This is in accordance with the Portuguese tendency for an asymmetrical distribution of resources throughout the country, a trend that has been aggravated in the last years and that calls for immediate action from the part of policy-makers. Socio-economic status was the variable that revealed the strongest correlation with HOME scores. It aggregated information on mother and father educational level, mother and father occupation status and family income, added up to form a continuous variable. It was quite evident from our data that children with higher socio-economic status had higher HOME scores. Our data have also shown (although numbers have note been presented here) that dispersion of the results on the HOME scores was higher on the group of children with lower SES, which led us to believe that the relation amongst the two variables might be stronger in the higher SES groups. In other words, families from low socio-economic backgrounds do not necessarily receive the lowest scores on the HOME, and may be able to develop stimulating and challenging environments for their children. This interesting result undoubtedly deserves further investigation. Finally, it is important to refer that although demographic data predicted half of the HOME variance, another half of the variance was unaccounted for. Other family variables, apart from those commonly related with SES, are certainly involved in the determination of the quality of home environments Question 4. To what extent do scores on the HOME agree with other measures of parenting or the home environment? 17 At this point we do not have much information on this issue. However some people from our group are conducting research on Engagement, and have used the 0-2 version. Data are being analysed and we will soon be able to send some of the results. Question 5. How are HOME scores related to measures of child outcomes/well being/development? The HOME was used as an independent variable, together with other variables from the ecological context of families, to predict the results obtained by children on two different outcome measures: the level of receptive vocabulary and the adaptive behaviour. Whereas the first is a traditional domain of psychological evaluation, assessed through a standardized norm-referenced test– the PPVT-R , the second presents different characteristics, being a less traditional area of assessment where the competence of child is evaluated through the mother and regarding the child’s competence in non-standardized situations. The HOME was a significant predictor in the case of the cognitive measures and it showed to be the best predictor in the outcome measures Socialization and Language and knowledge. However it did not predict the child’s competence in the Autonomy in Daily life activities domain as well as in the Motor domain, where children’s characteristics like age and sex seemed to be in motion. It is relevant to note that the HOME total score was always a better predictor of children’s outcome measures then the socio-economic index used in this study, which, as explained, was a continuous variable aggregating information from five different sources. It was also interesting to find out that significant correlations between children’s results and HOME scores were only to be found in the group of families from low socio-economic background. So, on one side, the HOME score seems to be the most powerful single predictor of child development. On the other side, it seems obvious that the HOME inventory constitutes a particularly useful instrument in the case of families from low socio-economic backgrounds or at risk for other socio-cultural reasons, either for assessment or for intervention purposes. 18 Appendix A The following tables contain results of multiple regression analysis on all outcome measures: Receptive vocabulary (PPVT-R), Adaptive Behaviour (VABS), and the four adaptive behaviour domains – Autonomy in daily activities, Socialization, Language and knowledge and Motor development Table A – Multiple regression for PPVT-R, using child and ecological context variables as predictors Predictors β t sr ² R² F(9,205) Energy - .051 - .739 .00 .220 6.413*** Sociability .089 1.321 .01 Emocionality - .003 - .041 .00 Introversion .012 .184 .00 Age - .061 - .978 .00 Sex - .067 - 1.074 .01 SES .141 1.555 .01 Preschool attendance (1=yes, 0=No) -.006 -.095 .00 HOME .336 3.938*** .07 +p<.10 * p<.05 **p<.01 ***p<.001 Table B - Multiple regression for VABS, using child and ecological context variables as predictors. Predictors β t sr² R² F(9,198) Energy .101 1.440 .01 .234 6.708*** Sociabi lity .230 3.375** .05 Emocionality -.088 - 1.346 .01 Introversion .100 1.512 .01 Age .293 4.636*** .10 Sex -.179 - 2.838** .04 SES -.002 - .064 .00 Preschool attendance (1=yes, 0=No) -.084 -1.256 .01 HOME .253 2.944** .04 +p<.10 * p<.05 **p<.01 ***p<.001 19 Table C - Multiple regression for Autonomy, using child and ecological context variables as predictors Predictors β t sr² R² F(9,205) Energy .135 1.907 + .02 .178 4.949*** Sociability .096 1.379 .01 Emocionality -.058 - .881 .00 Introversion .039 .584 .00 Age .152 2.360* .03 Sex -.327 - 5.101*** .12 SES -.036 - .391 .00 Preschool attendance (1=yes, 0=No) -.123 -1.814+ .02 HOME .096 1.098 .01 +p<.10 * p<.05 **p<.01 ***p<.001 Table D - Multiple regression for Socialization, using child and ecological context variables as predictors Predictors β t sr ² R² F(9,205) Energy - .013 - .184 .00 .157 4 .240*** Sociability .245 3.480** .06 Emocionality -.072 - 1.067 .01 Introversion .084 1.225 .01 Age .100 1.530 .01 Sex -.124 - 1.912 + .02 SES -.096 - 1.015 .00 Preschool attendance (1=yes, 0=No) .001 .015 .00 HOME .318 3.579*** .06 +p<.10 * p<.05 **p<.01 ***p<.001 20 Table E - Multiple regression for Language and knowledge, using child and ecological context variables as predictors Predictors β t sr² R² F(9,199) Energy .109 1.614 .01 .280 8.590*** Sociability .219 3.322** .05 Emocionality -.093 - 1.314 .01 Introversion .048 .755 .00 Age .281 4.596*** .10 Sex -.118 - 1.939 + .02 SES .111 1.254 .01 Preschool attendance (1=yes, 0=No) -.144 -2.229* .02 HOME .307 3.690*** .06 +p<.10 * p<.05 **p<.01 ***p<.001 Table F - Multiple regression for Motor Development, using child and ecological context variables as predictors Predictors β t sr ² R² F(9,205) Energy .054 .744 .00 .130 3.409** Sociability .088 1.229 .01 Emocionality -.061 - .899 .00 Introversion .070 1.002 .00 Age .296 4.480*** .10 Sex .105 1.600 .01 SES -.057 - .595 .00 Preschool attendance (1=yes, 0=No) .138 1.976* .02 HOME - .038 - .427 .00 +p<.10 * p<.05 **p<.01 ***p<.001 21 Bibliographic references Bradley, R. H., Mundfrom, D. J., Whiteside, L., Casey, P.H., & Barrett, K. (1994). A factor analytic study of the infant-toddler and early childhood versions of the HOME Inventory administered to white, black, and Hispanic American parents of children born preterm. Child Development, 65, 880-888. Bronfenbrenner, U. & Morris, P. A. (1998). The ecology of developmental processes. In W. Damon & R. Lerner (Eds.), Handbook of Child Psychology, Vol. 1 (pp. 993-1027). New York: John Wiley. Buss, A. H. , & Plomin, R. (1984). Temperament: early developing personality traits. Hillsdale, N. J. : Lawrence Erlbaum Ass., Publ. Caldwell, B. & Bradley, R. (1984). Home Observation for Measurement of the Environment (revised edition). Little Rock, A.: University of Arkansas at Little Rock. Dunn, L. M. & Dunn, L. M. (1981). PPVT - Peabody Picture Vocabulary Test - Revised. Manual for Forms L and M. Circle Pines: American Guidance Service. ECCE-Study Group (1997). European child care and education study. Cross national analysis of the quality and effects of early childhood programmes on children’s development. Final report for work package #1 submitted to European Union DG XII: Science, Research and Development. RTD Action: Targeted Socio-Economic Research. Palacios, J., Lera, M.-J., & Moreno, M. C. (1994). Evaluación de los contextos familiares y extrafamiliares en los años preescolares: escalas HOME y ECERS. Infancia Y Aprendizaje, 66, 71-85. Sparrow, S. S., Balla, D. A., & Cicchetti, D. V. (1984). Vineland Adaptive Behaviour Scales. Survey Form Manual. Circle Pines, MN: American Guidance Service. Este relatório foi escrito em Julho de 2004