本研究以某高中升學班與普通班各一班為對象，追蹤調查班級內的友誼網絡與課業諮詢網絡，並透過分離時序指數隨機圖模型（STERGMs）分析網絡的生成機制。研究結果顯示：（1）高中生選擇成績相似者交往的傾向隨著班級的升學氣氛差異而有所不同。在高度追求升學績效的班級內，以成績擇友的情形更明顯。（2）高中生挑選課業諮詢對象時頗為精打細算。他們除了向朋友詢問課業問題，也很清楚某科的問題要去請教該科成績高的同學。（3）課業諮詢是高中生友誼的來源之一。曾經提供諮詢服務的同學，爾後更容易成為課餘時間的朋友。

This study attempts to carry out three pioneering achievements: comparing differences in friendship networks between an elite class and a normal class in a high school, conducting surveys of academic consulting networks among high school students, and taking a lead in analyzing how high school students’ friendships and consulting relation¬ships are generated by using “separable temporal exponential random graph models” (STERGMs). Previous studies have revealed that high school students tended to form friendships with classmates who had academic performance simi¬lar to theirs, and some studies have emphasized that teenagers’ friend¬ships developed from class contexts they were in. However, the present study first reveals that class contexts will influence high school students’ tendency to form friendships based on academic similarity. Many high schools in Taiwan put students with higher academic performance together in one class, and they strive all out to attend top universities or medical colleges. The academic differences between these selected elite students are usually smaller than the differences between students in normal classes. However, in this study, students of the selected elite class still can form friendships with classmates who have academic performance more similar to their own. By contrast, it is not apparent in a normal class that students form friendships depending on academic similarity. The results here show that a class context with intense focus on academic performance may indeed increase the tendency for stu¬dents to choose friends based on similar grades. Secondly, most studies of teenager social networks have focused on friendships. The current study attempts to conduct surveys of high school students’ academic consulting for each subject. Results demon¬strate that high school students are shrewd when choosing suitable peer consultants. In addition to choosing friends as consultants directly, high school students also clearly know how to choose peer consultants for each subject based on their classmates’ academic performance in each subject. Besides, it is not common to find reciprocity in such academic consulting. In many situations, the consultant does not intend to obtain helpful information from the receiver. It can be seen from this that the academic consulting network is more like a hierarchical structure. Students with lower academic performance often receive advice from students with higher academic performance unilaterally. This kind of network structure creates chances for knowledge to flow from students with higher academic performance to those with lower academic per¬formance. On the other hand, this study reveals that academic consulting is one of the ways for high school students to develop friendships. The receivers may develop friendships with the consultants who were not friends at first. This finding implies a more significant meaning. Assum¬ing that students with higher academic performance only establish relationships with students who are similar to them, the academic gap would be unfillable or even aggravated. If consulting relationships could be further expanded into friendships, the academic gap would be mitigated, since the consulting relationships often bridge students with higher and lower academic performance. The experience of the present study also provides beneficial inspi¬ration for future research designs.