This study uses hierarchical linear modeling to examine the influence of a range of factors on the science performances of Hong Kong students in PISA 2006. Hong Kong has been consistently ranked highly in international science assessments, such as Programme for International Student Assessment and Trends in International Mathematics and Science Study; therefore, an exploration of the factors that affect science performances of Hong Kong students can give a lens to examine how science education can be improved in Hong Kong and other countries. The analyses reveal that student backgrounds as male, at higher grade levels, and born in mainland (when in the same grade) are associated with better science performance. Among the attitudinal factors, enjoyment of science and self-efficacy in science play important roles in scientific achievements. Most of the parental factors, on the other hand, are not having significant impacts on achievement after student attitudes are taken into account, with only parents' value of science having a small effect. School student intake is found to be a strong predictor of school average achievement, as well as a major mediator of the effects of school enrollment size and school socio-economic status. The findings differ from recently reported results, which suggested that school enrollment size was associated with achievement. This study also points out the problems of the use of science instruction time as a school-level variable to explain science achievement in Hong Kong.
[Copyright of International Journal of Science Education is the property of Routledge. Full article may be available at the publisher's website:
http://dx.doi.org/10.1080/09500693.2013.879223]

Data for this study was collected from the Program for International Student Assessment in Hong Kong (HK-PISA), which was conducted in 2002. A total of 4405 fifteen-year-old students from 140 Hong Kong secondary schools completed the student assessment. A total of 4336 parents also completed the parent questionnaires. These questionnaires was used to provide information about family social capital and family background.
In conclusion, the results of this research support Coleman's thesis of social capital in the creation of human capital. It also underscores the importance of bridging social capital in student's learning. Finally, the given recommendations based on the results of this research may be useful for policy makers, educators and parents in assisting students to improve their learning and develop their potential fully.
The CFA results suggest that the applicability and validity of the indices related to the three types of capital created by OECD is fair, and therefore the local social context should be taken into consideration when modifying these indices for adoption in the research. The results of the HLM analysis demonstrate that most measures of family social capital are consistently associated with students' literacy. They also show that effects of the process attribute of social capital on student's achievement are stronger than that of the structural attribute. Among the social capital variables used at the individual-student level, students score higher if their parents communicate (both social and culturally) with them more and also if their parents expect their child to attend university. Among the social capital variables used at the school level, the school mean parental network has a stronger positive contextual effect on students' achievement. As for the variables of physical and cultural capital, only part of them has a significant effect on student's achievement, and the effect is weaker than that of family social capital.
The