由國立交通大學學生背景特質與學業表現看多元入學政策

Abstract

本研究運用國立交通大學94學年度至100學年度之學生資料，以實證研究的精神探討多元入學管道之政策在本校的實施成果，希望了解(1)多元入學政策下錄取學生之背景；(2)各科系的招生策略；(3)入學後的大一專業必修科目表現。 我們發現受全國政策影響，透過考試分發進入本校的學生比例逐年遞減，個人申請逐年遞增，繁星計畫近年則維持在10%左右。其中畢業於公立明星高中（共31所）之交大學生，有較高比例透過「考試分發」或「個人申請」入學。「繁星計畫」的推動增加了學生背景的多樣性，但對其他公立高中（指非上述之明星高中）和私立高中則較有利。分析學生大一成績表現，透過繁星管道入學的學生若只單看平均成績，表現並不差，但異質性大。明星高中學生若能在繁星管道被挑選，學業表現往往比其他高中背景更出色。我們亦整理了本校學生所畢業之高中在全台的地理分布情形。 此外我們建立數學模型以探討何謂「最佳的招生策略」，藉由電腦模擬檢驗所提出檢定統計量的分布在虛擬假設下的正確性，並透過機率的運算說明最佳的招生策略會是當“各入學管道表現最差者已達均質”時，也就是各科系表現不佳的學生均勻地出現在各管道，這樣的招生策略會使整體能力最優。在結論中我們基於此資料分析之心得，陳述了對相關教育議題的建議。

In the thesis, we study the influence of the multi-route admission policy on NCTU students during the academic years from 2005 to 2011. We especially focus on the impact on three aspects, namely (1) students' background, (2) the admission strategy for different departments and colleges and (3) students’ academic performances in their freshman year. Similar to the nationwide trend, we find that the proportion of students entering NCTU through examination and allocation (EA, 考試分發) gradually decreases; on the contrary, the proportion of those admitted via recommendation (AR, 個人申請) is increases. The proportion of the “twinkling star” program (TSP, 繁星計畫) remains around 10% in recent years. We also find that students graduated from the 31 well-known public high schools (公立明星高中) are more likely to enter NCTU via the EA or AR channels. TSP is more advantageous for students graduated from less competitive high schools. On the other hand, it increases the diversity of students' background on NCTU campus. For students admitted via TSP, their average performances in the first year are not much different from those via other routes but there is significant heterogeneity among TSP students. If TSP students coming from well-known public high schools, they tend to do well in their freshman year. By providing some geographical information for the high schools of our students, we believe that the analysis can be used to examine the educational resources on the island. We also construct probability models and perform simulation analysis to give suggestions about "the best admission strategy" for a department. Finally, we discuss the issues related to recent change on the nationwide education policy.