The psychometric properties of eHCI was evaluated in a representative sample of children aged 3–4 years from all districts of Shanghai. Results of the present study suggest the eHCI is psychometrically sound for Chinese children.
In term of reliability indicators, the α coefficient indicates good internal consistency sufficient for group comparison other than the domain of physical (26). The physical subscale was designed to understand children’s disability, health status and behavior. The four items in the subscale are: "Is this child frequently sickly? ", "Does this child have good hygiene i.e. always wash their hands after toileting?", "Does this child have any disabilities/special needs?" ,"Does this child have a regular diet?", are not strongly correlated with each other. Perhaps indicating that these physical factors act mainly as independent characteristics rather than as a scale of physical development. An ICC above 0.75 is considered as excellent (27). The result of our reliability analysis suggested that eHCI had good internal consistency and temporal stability. However, the inter-rater agreement in the present analysis was more variable, with subscales related to Literacy Numeracy showing excellent consistency between scores rated by parents and teachers, and the others showing greater heterogeneity in responses. Since the items in Numeracy Concepts, Reading, and Writing are relatively objective indicators, it is reasonable that scores in those aspects were more consistent between parents and teachers. These results are consistent with the reported reliability of other measures of child development and the reasons for inconsistent are likely to be related to parent, teacher and child factors as well as context (for example; parental knowledge of child development, parent literacy levels, parental engagement in the school system, teacher qualifications and knowledge of development, teachers experience across different socioeconomic settings, child behavior being different in the school compared to home due to shyness or other factors). The paired t-test results suggested that teacher scored higher than parents for the same children. For example, the items in cultural spiritual are "Does this child talk politely?", and "Is this child good to his or her parents?". It may be because children act differently in kindergarten than at home. It also may be because parents expected too much of their children. We can’t draw a conclusion without deeper exploration of the reason behind the disagreement. In the future, when using the eHCI or other measures of child development it will be important to distinguish the raters prior to scores being compared across different populations.
The results of confirmatory factor analysis supported the underlying structure of the eHCI. The model fit demonstrated that the extracted factors from all items are capable of assessing the different developmental domains in Chinese children. All but five items have high factor loadings. Those five are reverse coded question: kick, bite or hit adults or other children; impatient; need constant reminding to finish something off; get easily distracted from a task; frequently sickly. Even though the factor loadings of the reverse coded items were lower than expected, it may be important to keep the items worded in a negative fashion. There is evidence to suggest that respondents get into a pattern of response and reversing the direction of a question requires deeper thinking, however others in survey methodology would recommend keeping all survey items in the same direction for simplicity and to reduce confusion (17, 28). This may be something worth exploring further with future use of the eHCI.
The eHCI showed significant correlations with other metrics covering different domains of child development, such as Age & Stages Questionnaire (gross motor, fine motor, communication, problem-solving, social-personal), Strengths and Difficulties Questionnaire (psychosocial well-being), etc. However, those metrics are inclined to screen the individual with high-risk of development. The eHCI was designed to monitor the comprehensive abilities of children in population level. As such, we wouldn’t have expected correlations larger than what was found.
The discriminant validity of eHCI with demographic characteristics was also presented in the results. The eHCI scores of girls were significantly more than that of boys, consistent with the conclusion of other studies that girls mature earlier than boys (29). The results also suggest that higher eHCI scores appeared in the groups with higher socioeconomic status, in keeping with prior research (30). The significant association between eHCI scores and demographic characteristics verified that eHCI could detect the development heterogeneity of different populations.
This study has several limitations that deserve mention. First, although the eHCI was proved to be a feasible and comprehensive tool for identifying the developmental level of Chinese children, the overall sample was not representative of the national population, even though children from migrant workers from rural areas in Shanghai were included within this sample. Second, although the eHCI could be applied as an instrument for monitoring and to compare the status of early childhood development in different populations worldwide for its cross-culture design, it is not meant to replace traditional screening or diagnostic tools for delayed development. The eHCI emphasizes improving early childhood development at a population level, rather than diagnosing individual children as abnormal. Future studies should take this into consideration according to their target population and goal. Third, although the reliability and validity of eHCI has been tested in this study, there is still no evidence to verify eHCI as a reliable predictor of long-term indicators of academic or working achievement, such as education level, income, and crimes. Longitudinal studies are warranted to test its predictive validity for later outcomes.