New Perspectives in Science Education

The world has been globalized quickly. This globalization movement has also influenced on the education sector. Many countries join the international comparative studies to get suggestion for their future educational plan. TIMSS and PISA are the most well-known international comparative studies. The number of participant countries has been increased and now reached more than 60 countries in the world (IEA 2012, OECD 2013). The number of participants of developing countries is also increasing. However, some scholars question that these international comparisons with single standard are not so much appropriate particularly for developing countries. For instance, floor effect by low achievement and so on (Baba and Uchida 2008). But the number of participants of developing countries is still limited, so that substantial reports are also limited.  This study aims to add substantial verification on the theoretical study, by seeing the adaptability and problem of international comparative study for developing country. The study conducted simplified TIMSS test in one of the developing country, Bangladesh, which have never participated in any international comparative study. This study picked some test problems and background information questionnaire from framework in TIMSS. The city of Mymensingh was selected, where have both urban and rural area. Target grade is 4^th grade. The achievement test includes 21 questions both for contents domains and cognitive domains more than two, The questionnaire includes 22 questions consisted of 8 questions of family background, 7 questions of individual background and 8 questions of schools and teachers background. The scores and answers were then analyzed by factor analysis, multiple regression analysis and Covariance Structure Analysis with using SPSS and AMOS. Total number of  data was 30 schools and 1194 students. Result of the multiple regression analysis and Covariance Structure Analysis from student’s questionnaires shows that most influential factors for students’ achievement are “Gender” and “Parental education” and “Reading environment”. From teacher’s questionnaires, there are significant difference in “town and country” and “Gender of teacher”. Coefficient of determination of family and individual factor is 3.4%, and school factor is 31.2 % on the multiple regression analysis. It means that school factor is more influential for students’ achievement in Bangladesh. These results look like a nothing problem. By looking in details, there are some problematic issues raised though analysis and data themselves, For example, even though the most of tests style was multiple choice questions of four options, the average achievement was 5.95/ 21 (SD=2.80) and therefore the Cronbach's alpha was not so high (0.61). Some questions were unsuitable as data such as most of the answer for “Number of books in your house” was 0, and “Mothers occupation” was mostly Housewife). These can be suggested that there are some rooms for improvement about test questions in order to appropriately adapt in developing countries.