The Development of EHCI in Chinese Version
In 2014, under the guidance of Brinkman, an early childhood development specialist in Australia, the China Development Research Foundation (CDRF), in collaboration with the Shanghai Children's Medical Center affiliated to Shanghai Jiaotong University School of Medicine, started working on the Chinese version of the early Human Capacity Index. Through discussions among experts in different fields such as pediatric medicine and education, the team translated each item into Chinese and made necessary amendments to reflect China’s cultural characteristics and avoid the ceiling and floor effects in the Chinese population. In 2015, based on the survey data of 3,698 Chinese children, the developers carried out a Rasch model analysis of eHCI. The overall chi square fit statistic of fit to the Rasch model was 2078.773 (df 540), p < 0.001 and the item fit residual was − 1.1258 (7.6952). The distribution of the item and person locations relative to one another on the same continuum is shown in Fig. 1 (Online). The Person Separation Index (the Rasch equivalent of Cronbach’s alpha indicating level of reliability) was 0.88509, which indicates high reliability. The power of the tests of fit was rated Excellent. The final scale has 62 entries, which can be completed by any person familiar with the child, such as parents, teachers, social workers, etc. The scale includes 9 dimensions: 1) verbal communication, 2) approaches to learning, 3) numeracy and concepts, 4) Reading, 5) writing, 6) cultural identity and spirituality, 7) social-emotional wellbeing, 8) perseverance, and 9) physical health. From these 9 domains an overall literacy and numeracy score is derived, as well as an overall development score, both also ranging from 0 to 1 with 1 being the best score.
Study Sample And Data Collection
This study used data to assess the reliability and validity of the eHCI mainly from the 2016 Shanghai Children’s Health, Education and Lifestyle Evaluation, Preschool (SCHEDULE-P) study. The 2016 SCHEDULE-P study was a cross-sectional survey that investigated the lifestyle, home environment and development of preschool children in 2016. The design, sampling and procedures of the survey have been described previously (20). A representative sample of newly enrolled preschoolers in Shanghai kindergartens was obtained by stratified random sampling design. There were 20,899 children (age 36–58 months) from 191 kindergartens who enrolled in the study. From these, there were 20,324 families who consented to participant, and the parents then completed the online questionnaire of the eHCI. The response rate was 97.2%.
To evaluate the discriminant validity of the eHCI, the demographic information of children and their family was also obtained in the survey. Age and gender of all participants were obtained from the Shanghai Kindergarten Registry Database of the Shanghai Education Committee and further confirmed by parents at the beginning of the study. Maternal education, paternal education, and annual household income were self-reported. The questionnaire included a family assets scale which sought information relating to the number of household cellphones, television, computers, cars and bathrooms (21). The principle-factor analysis was conducted to obtain a factor predicting the family assets. This same family asset scale was used in the Program for International Student Assessment (PISA) conducted by Organization for Economic Co-operation and Development (OECD) to reflect family wealth.
To assess how constant eHCI scores remain from one occasion to another, a test-retest reliability survey was conducted in two Shanghai kindergartens not involved in the SCHEDULE-P study. Parents of 183 kindergarten children aged 3–6 years old completed the eHCI for a second time 9.1 [SD:0.6] days after their first completion. In order to investigate the rater agreement, the eHCI ratings of 168 children from the two kindergartens were also compared between teachers and parent.
To test the correlations between the eHCI and other metrics of ECD by site, we used data from the SCHEDULE-P study, the One Sky program in Ye County and the kindergarten survey in Daming County. The Strengths and Difficulties Questionnaire (SDQ) was used in the SCHEDULE-P study. The Strengths and Difficulties Questionnaire (SDQ) and the Age & Stages Questionnaire: Social Emotional (ASQ: SE), two internationally recognized tool, were reported by parent to assess the psychosocial wellbeing status and the social-emotional development of the child in the SCHEDULE-P study (22, 23). The One Sky program conducted the study to describe the situation of the left-behind children in Ye County in August 2015. A total of 60 villages were selected from a list of all villages in Ye county provided by the Education Bureau of Ye county and the Bureau of Civil Affairs. All children aged 3 to 4-years old and their families were interviewed in these villages. The Age & Stages Questionnaire (ASQ) and the eHCI was filled in by the caregiver of 1918 children. The ASQ was designed to measure child development in the domains of communication, gross and fine motor, problem-solving skills and personal-social skills (23). The kindergarten survey in Daming County aimed to evaluate the early childhood development of children and the quality of preschool education in rural areas. 62 kindergartens were randomly selected from all 217 kindergartens in Daming County in October 2017. The eHCI was reported by caregivers of all children (age 3–4 years) in the first year of kindergarten. The total sample of 6,974 eHCI included 2203 paper and 4744 online questionnaires. The East Asia-Pacific Early Child Development Scales (EAP-ECDS) containing 99 items includes seven domains: cognitive development; cultural knowledge and participation; language and emergent literacy, motor development; health, hygiene and safety; socio-emotional development; and approaches to learning (24). The EAP-ECDS was tested in 1199 children on site by well-trained assessors. When interpreting some of the results presented in this paper it is important to note that for some aspects of the SDQ and the ASQ:SE higher scores represent children with greater development problems, which is opposite to how the other measures of child development are coded.
The study was approved by the Institutional Review Board of the Shanghai Children’s Medical Center (SCMC), Shanghai Jiao Tong University (SCMCIRB-K2016022-01). Statistical Analysis
We conducted descriptive analysis on demographic characteristics using the SCHEDULE-P data. Reliability was assessed from analyses of internal consistency, test-retest reliability, and inter-rater agreement. Internal consistency of eHCI was assessed using Cronbach’s coefficient alpha. The intraclass correlation coefficients (ICC) between eHCI scores by parents in two time points were calculated to assess the test-retest reliability. The ICC and paired t-test were calculated separately to assess the agreement and difference of eHCI scores rated by parents and teacher.
To validate the eHCI, we conducted the tests of structure validity, criterion-related validity and discriminant validity. Criterion-related validity of the eHCI was conducted by calculating its correlations with other metrics of ECD. Discriminant validity was tested through multi-level linear regression models assessing score differentials with respect to child gender, parental highest education, family income, and quantiles of family assets. The above analysis was conducted using Stata 14.2 (from StataCorp LP, College Station, Texas, U.S.A.). The Confirmatory Factor Analysis (CFA) model was established to confirm the dimensionality of eHCI. The robust weighted least squares estimator (WLSMV) was used, as the item variables of the eHCI are categorical (25). The CFA was operated in Mplus 8 (from Muthen & Muthen, Los Angeles, CA, U.S.A.).