Re-analysis of Scientific Creativity Test for Pre-schoolers Using Rasch Model

Abstract

The Figural Scientific Creativity Test (FSCT) was developed using Exploratory Factor Analysis (EFA). As there are limitations of the Classical Test Theory (CTT), this reanalysis of the original study aimed to ascertain the construct validity of the five subscales of the FSCT using Rasch analysis. These five subscales of scientific creativity were fluency, originality, elaboration, abstractness of title, and resistance to premature closure. Rasch analyses were conducted on data from a sample of 50 pre-schoolers (29 females and 21 males, aged 6 years old) from two pre-schools in Kota Kinabalu, Sabah. FSCT consists of six items, which are re-scored according to criteria for four or five subscales. The subscale of fluency was analysed using the Dichotomous Rasch Model, while the other four subscales were analysed separately using the Rating Scale Model. The five subscales were then reanalysed as a whole to check the unidimensionality of scientific creativity. It was found that the subscales are individually unidimensional and the scientific creativity, with five subscales, is also unidimensional. When analysing the five subscales as a whole, the originality subscale in most of the items show misfit. The originality subscale needs to be reviewed in future studies if researchers wish to measure scientific creativity as a whole. When five subscales were analysed separately, two items in the originality subscale needed to be revised and an item in the elaboration subscale needed to be modified. The reliability and separation indices of the items in four subscales (fluency, elaboration, abstractness of title and resistance of premature closure) are still not within the acceptable range, while the item and person reliability were in acceptable range for scientific creativity. The elaboration scale was reduced to a 3-point rating scale, while abstractness of title and resistance of premature closure scales were changed to dichotomous.