Exploring School Improvement in Hong Kong Secondary Schools

Abstract

Scholars have increasingly sought to understand how the process of school improvement differs among schools operating in different school levels, conditions, and contexts. Using Rosenholtz's (1985) conception of “moving” and “stuck” schools as a framework for thinking about school improvement, this study examines the learning outcomes of 39 Hong Kong secondary schools over a 3-year period. We examine whether features of leadership and school capacity differed with respect to these learning outcomes within the sample of moving and stuck schools. This research in Hong Kong has identified several factors that appear to synergistically contribute to differences in patterns of improvement in learning across different subjects in both moving and stuck schools. These factors include resource management of principals and school capacity in terms of professional learning community; workload of teachers; alignment, coherence, and structure; and resource capacity. This study extends the research on leadership and capacity building as a means of school improvement, in the process elaborating on their impact within a non-Western society.

Acknowledgments

We acknowledge the funding support of the Research Grant Council of Hong Kong for its support through the General Research Fund (GRF 451407).

Notes

According to the most recent data available, there are only 457 schools in the major districts. Our 39 schools represented about 7%, 8%, 7%, and 10% of these districts. There are 77 fee-paying schools receiving the government direct subsidies, but only three were in our sample. All other 36 schools are aided by the government and offering free education to students. Regarding school sponsoring bodies, the 39 schools represented 33 sponsoring bodies with five schools from Hong Kong Catholic Diocesan, the largest School Sponsoring Body in Hong Kong. Our sample may be slightly underrepresenting schools from the Anglican Church and the Chinese Christ Church, but we considered this would not affect our findings much.

We examined the scales against the data using LISREL 8.8 (Jöreskog & Sörbom, 1993). We ran a confirmatory factor analysis (CFA) for each scale allowing the factors to correlate freely. The fitness of the CFA model was evaluated using multiple indices such as chi-square, root mean square error of approximation (RMSEA), together with the non-normed fit index (NNFI), the comparative fit index (CFI), Incremental Fit Index (IFI), Relative Fit Index (RFI), and the squared root mean residual (SRMR). In general, the chi-square will be less reliable and become significant when the degree of freedom increases and other fit indices are considered better indicators of the model fit (Bollen & Long, 1993).

³χ²(443) = 899.12, p = .00; RMSEA = .077 with 90% CI = .070, .085 and p value for CFit = .00; NNFI = .98; CFI = .098; IFI = .98; RFI = .96; SRMR = .052.