一般人與過去研究都強調，國家推動之各種教育與教育行政，是形塑同質的國家認同的重要機制。臺灣過去有威權式、由上而下的民族教育，但是仍然產生了不同的認同對立，並反映在統獨立場的差異上。本文探討教育程度的高低，和統獨意識、統獨立場的關係為何？本文提出兩項假說，一為「直線假說」：教育程度越高的人群，內部越趨於一致，也越可能傾向「以後統一」；另一為「折線假說」：假定當教育程度達到大學程度之後，內部歧異越大，傾向「以後獨立」的人會更多。本文利用經驗調查資料分析，發現教育程度有三種作用：1) 門檻效果：以國╱初中程度為分界，在此之前， 人們對於統獨問題比較疏離，常無法認識或回答。 2) 同質效果：當控制族籍、分類意識、政黨意識、年齡之後，教育程度的淨作用，如同「直線假說」，有將不同人口同質化之積極作用。 3) 異質效果：透過「互動項作用」，臺灣閩南人和外省人、中國人有三種作用︰１門檻效果︰以國╱初中程度為分界，在此之前，人們對於統獨問題比較疏離，常無法認識或回答。２同質效果︰當控制族籍、分類意識、政黨意識、年齡之後，教育程度的淨作用，如同｛直線假說｝，有將不同人口同質化之積極作用。３異質效果︰透過｛互動項作用｝，臺灣閩南人和外省人、中國

　　　　　Common understanding and previous studies tend to show that nationalized curriculum and education system have a strong unifying effect on the forming of national identity. Taiwan had been under authoritarian rule with a strong emphasis on nationalistic ideology enforced from above; but nationalistic unity in Taiwan in recent years has been divisive, reflecting people's different standings on the tong-du (unification vs. independence ) issue. This article explores what the relations are between education level and tong-du standings. It suggests two hypotheses: 1) the linear hypothesis: the higher the education level, the more homogenized people become, and the more they are inclined to ward " unification", and 2) the non-linear hypothesis: the higher the education, the more diversified people become, while the chances for people to incline to ward "independence" also increase. Using survey data, the paper finds three kinds of effect of education level: 1) the threshold effect: the junior high school level constitutes the threshold. If being under the threshold, a great proportion of the people cannot articulate the "tong-du" issue. If being above the threshold, the proportion who cannot articulate the "tong-du" issue drops significantly; 2) the homogenizing effect: when ethnic background, self-proclaimed identity, party inclinations, and cohorts are accounted for, education level has a "net" and positive effect in increasing the direction inclining to ward unification as the linear hypothesis suggests; and 3) the dividing effect: through statistical interaction, we find that the tong-du differences between Taiwan-minan and mainlanders, self-proclaimed Chinese and Taiwanese and that between New Party and the DPP supporters is enlarged when the education level increases. The article also notes that the amount of accountable variances by education level, though statistically significant, is small. In the end, we suggest that national education not be considered as the only source of political socialization. The effect of national education on people of different backgrounds is expected to vary from group to group.