关键词: 入学准备 Rasch模型 项目反应理论 分部评分模型

Abstract: As an important task of early childhood education, getting every child prepared for school is increasingly emphasized by governments, stack-holders and parents in recent years. Against this background, researchers paid more attention to children’s school-readiness. In this line of studies, it is essential to accurately assess children’s school-readiness status. Using Classic Testing Theory (CTT), children’s scores of each item are added together to get a total raw score for each individual. The reliability, validity and discrimination of the test are then analyzed based on these raw scores. This method meets challenges in school-readiness testing. First, the difficulty of the test and the ability of children are dependent on each other so that one cannot gain objective and independent estimates of children’s abilities and difficulty of test as well. Second, school-readiness tests tend to have complex errors due to young children’s low developmental level and lack of examination skills. However, it is hard to decompose and analyze errors using Classic Testing Theory. Third, Classic Testing Theory takes raw score as equal interval variable while it’s merely ordinal variable in essence. The Rasch model on the other hand, can solve these problems well. The recent study aims to use Rasch model in developing and analyzing a math school-readiness to investigate its effectiveness and superiorities in school-readiness testing. The study developed a test containing 13 items and used it to test 150 children (mean age=6.6 years, 89 males). Using Rasch Model, the study analyzes the results and examines the reliability of the test and model fit of each item. The study also investigates the items and their categories and subsequently modifies the original test. The final version has an individual separation reliability of .70. All except one item have acceptable model fit indices, i.e., mean square of both infit and outfit between .7 to 1.3, and t statistics between -2 to 2. Most items’ difficulty is less than 1, indicating that the test is relatively easy for participants. However, this is appropriate for a school readiness test because it aims at testing how well is each child prepared for going to learn in primary school. Normally, most children should be well prepared. The Rasch score for each participant is then calculated and is found to be significantly correlated with participants’ age (r=.16, p<.05) and socio-economic status(r=.47, p<.01). The superiorities of using Rasch Model in testing school-readiness status are also discussed. In general, Rasch Model provides more effective methods that facilitate item diagnosing and also helps generate more reliable scores for subsequent analyses compared to Classical Test Theory.

Key words: School Readiness, Rasch Model, Item Response Theory, Partial Credit Model