Development Of A Parenting Assessment Tool For Chinese Parents Of Infants/toddlers

ObjectivesIt is well recognized that the period from birth to two years of age is a "critical window" for the promotion of optimal growth, health, neuropsychological and behavioral development. The parenting quality of infants and toddlers is one of the most important potential contributors to early childhood development and is crucial in influencing a range of health and social outcomes across the lifecourse. There was not too much valid and reliable scales for assessing parenting of infants/toddlers, even less within China. Researchers in China often borrowed well developed parenting scale from abroad, diretedly in research. Limitations in the current domestic parenting scales were as the following:1) Parenting differs across cultures and by socioeconomic status. It is not always suitable to assess parents of China using items in the western developed scales.2) Existing parenting assessment tool are usually measuring an isolated angle of parenting and can hardly show the whole picture.The present study aimed at1)developing a parenting assessment tool (PAT) for parents of children aged zero to2years to assess parenting quality within the Chinese socio-cultural context,2)exploring a much better method between factor-analytic methods of scale development and the derivation of conceptually designed subscales,3)examining a priori hypothesized relationships between parenting and children’s outcomes.MethodsThe development of the parenting assessment tool for Chinese parents of infants/toddlers fell into three basic stages. Stage1was item development, or the generation of individual items. Through a thorough review of the literature, experts consultation and focus group discussion, we developed the theoretical definition of the construct of parenting of infants and toddlers. The definition was then used as a guide for the development of items. Then the deductively generated items were subjected to a sorting process that served as a pretest, permitting the deletion of items that were deemed to be conceptually inconsistent.Stage2was scale development, or the manner in which items are combined to form scales. Data for this stage(n=284) were gathered from5cities. Two approaches were used to determine PAT subscales. The first approach was exploratory factor analysis. Another was deductive approach, aiming at developing subscales that were conceptually meaningful. Items were grouped together based on theoretical connectivity rather than statistical analyses. Firstly18experts developed10dimensions of PAT. Then the experts evaluated the degree to which each item was representative of the parenting dimension. Items that were not catigorized within a dimension by at least two-thirds of the judges were removed. This process resulted in the retention of48items.Stage3was scale evaluation, or the psychometric examination of the new measure. In order to confirm the properties of two scales, a second data collection was conducted(n=607) in ten cities. Confirmatory factor analysis was used to evaluate the latent structure of two scales and to modify the models when necessary. Then we assessed the psychometric properties of the scales including Cronbach’s alpha coefficient, split-half reliability, test-retest reliability, content validity, construct validity, and criterion-related validity. To examine the relationship between parenting and children’s outcomes, partial correlations were calculated between PAT subscales scores and children’s development scores.ResultsBased on a review of the literature, experts consultation and focus group discussion,78items were developed. Twenty-eight items were deleted that were deemed to be conceptually inconsistent.Factor analysis revealed9factors with Eigenvalues greater than1.0. By the scree test, the scree plot of EFA suggests that a7-factor model might be most suitable in this dataset based on a substantial decrease in Eigenvalues. Based on this result, we proposed a7-factor model named PAT I. Results of the confirmatory factor analysis suggested that one additional item should be deleted from the scale. After the removal of this item, the7-factor model provided good fit.Experts developed10dimensions of PAT. Factor9and factor10has an alpha below a minimal acceptable level (0.50), so these two factors were not included in this study. Based on this result, we proposed a8-factor model named PAT Ⅱ. Results of the confirmatory factor analysis suggested that two additional items should be deleted from the scale. After the removal of these two items, the8-factor model provided good fit.Cronbach’s alpha coefficient of these two models was0.818and0.845respectively, Guttman split-half reliability was0.752and0.816respectively, test-retest reliability was0.951and0.918respectively, correlation coefficients between total and factors were from0.525to0.769and from0.561to0.704respectively. With respect to criterion-related validity, most subscales of PAT Ⅱ and less than half subscales of PAT Ⅰ were positively and significantly related to children’s specific development.ConclusionBoth scales showed good psychometric properties and fitted the data well. Some items of PAT Ⅰ seemed not conceptively consistent with its scale derived based only on the results of factor analysis. Simply because items load on the same factor does not mean that they necessarily measure the same theoretical construct. PAT Ⅱ was more closely related to children’s development outcomes, so it could be used as an effective tool to assess parenting quality of infants and toddlers in China. Compared to factor-analytic methods, the application of deductive approach would help derive conceptually designed subscales, hence the criterion-related validity is improved as well.