PISA 2009上海地區數據之多層次分析方法與應用

The Application of Hierarchical Linear Model on PISA 2009 in Shanghai

大數據（big data）的時代來臨，讓我們以前所未有的觀點，並透過極大量的數據統計對教育進行分析。鑑於國際學生能力評量計畫（Programme for International Student Assessment, PISA）在教育評價的理念、技術和結果報告，對我國的大規模教育評價改革都有值得借鏡之處，本研究以PISA 2009上海地區數據為例，樣本為參與上海市2009年「國際學生能力評量計畫」正式施測的152所學校所屬的5,115名學生；研究方法採用多層次分析模型進行分析，以探討家庭社會經濟文化地位、學校教師品質對學生閱讀、數學及科學等三項素養之影響。研究結果顯示，上海市PISA 2009閱讀、數學及科學在不同學校的組間差異實不可忽略。對各學校而言，家庭社會經濟文化地位每增加一個單位，PISA成績可以有效提高五分左右；家庭社會經濟文化地位具有能夠正向解釋成績之差異的效果，當學生家庭社會經濟文化地位愈高，其PISA成績愈好；此外，當學校具有高品質的教師時，更能提高PISA閱讀、數學及科學方面的成績。

Big data makes us collect and work with a large volume and variety of data and apply massive data analysis on education. Based on the missions, techniques and reports of PISA, it can be worthy as the mirror of our large-scale of educational evaluation reform. This study applies PISA 2009 in Shanghai. The subjects contain 5,115 students in 152 schools who attended in PISA 2009. The methodology is hierarchical linear modeling (HLM) and we analyze that the effect on index of economics and social and cultural status, proportion of qualified teacher to reading literacy, mathematical literacy and scientific literacy. The results show there are differences among reading, mathematical and scientific literacy in schools. To each school, once index of economics and social and cultural status raise one single unit, the outcome of PISA will promote approximately five points. Index of economics and social and cultural status positively explains the relationship with the scores. The higher index of economics and social and cultural status students are, the better PISA they perform. In addition, when school employ high proportion of qualified teacher, the students’ performances in reading, mathematical and scientific literacy are better.