Redefining the underlying structure of the ROSE instrument and its application to understand Chinese students' affective domain of science learning

The questionnaire instrument of the international project called ROSE (Relevance Of Science Education), which investigates the affective domain of science learning for 15-year-old students, was administered to 70 classes of Chinese students in Hong Kong, Shanghai and Guangzhou of China and 2,426 valid questionnaires were returned together with 251 student interview records. The present work focused on the quantitative analysis in which our preliminary analysis indicated that the reliability for the Section D on “Me and the environmental challenges” was rather low (just around 0.7) as compared with 0.95 or above for all other sections and so its 18 items were discarded in our subsequent data analysis. After using the principal component analysis for data reduction, our EFA (exploratory factor analysis as based on the maximum likelihood estimate) revealed 25 key factors for 173 items taken from this ROSE questionnaire and they could be grouped into 4 categories by our expert’s judgement, to wit, (A) 11 for what the students want to learn, (B) 7 for students’ experiences on science and technology within or outside school, (C) 4 for students’ science and technology related career orientation, and (D) 3 for students’ opinions about science and technology. A second level EFA on those 25 factors led to 6 categories (or known as factors of factors) in which the previous categories A and B were each subdivided into another 2 highly correlated categories. However, using the AMOS software for second level confirmatory factor analysis, it was found that the EFA findings were actually consistent with our initial expert’s judgment of having 4 main categories of factors. By using the factor loadings as weights for summation of items within each factor, our present work demonstrated that factor scores and their average values could easily be calculated for comparing the students’ level of interest in various science topics and attitudes towards science vs gender, socioeconomic factors and banding of their schools across different cities. For our subsequent full-scale ROSE survey as conducted in Hong Kong and Shanghai, the data could be used to verify the present structural model of the ROSE instrument for the Chinese learners and to yield the category scores (i.e. second level factor scores) for the individual students so that they would know about their own level of affective domain of science learning as compared with the norm in their own school and own city. Furthermore, it will be discussed on how those factor and category scores can be treated as some reliable and concise indices for teachers to provide counseling on students’ choice of further studies in the science stream or future career orientation in the field of science and technology.