本文利用「華人家庭動態資料庫」的合併樣本，以Ejm*s and Partner (2004) 所提出之 「固定效果下有序羅吉特模型」 進行 「性別偏 好」與「出生排序」的實證分析。我們發現「重男輕女」普遍存在於 當時的台灣社會，且外省籍家庭「重男輕女」的程度並非所有族群 中最輕微的。出生排序越後面子女的敎育程度會越高，但是出生排 序越後面的兒子與女兒間之敎育成就差異則較低。前後手足的生育 間隔對於該子女的敎育成就具有「非對稱效果」：與相鄰兄姐的生育 間隔愈大，愈有利於該子女的敎育成就;與相鄰弟妹的生育間隔愈 大，反而對該子女的敎育成就有不利的影響。最後，雙胞胎的身份 會降低該子女的敎育成就。

This research utilizes PSFD (Panel Survey of Family Dynamics) and conducts an empirical study on the “son preference” and “birth order” effects by using the ordered logit model with fixed effects proposed by Ejrn^s and Portner (2004). We find that, first of all, there is a significant son preference effect found when parents make their education investment on children. However, the sex inequality of the education investment of mainlan-der families is not the smallest among all ethnic families as claimed by the existing literature. Secondly, children with later birth orders are found to have higher educational achievements. Furthermore, the difference in educational achievement among sons is less than that among daughters; and thus the difference between sons and daughters is decreased when they are in later birth orders. We also find that birth spacing will also have an asymmetric effect on a child’s educational achievement. A child who was born at a distant time to the adjacent older sibling tends to have a higher educational achievement; while his/her achievement shows a negative relation with the spacing to the next younger sibling. Finally, being one of twins may have an adverse effect on his/her educational achievement.