在第二个查询中，当您将temp表连接回 enrollment ，它为每个类生成一行；这就是 totalpaid 纵队是从哪里来的。
第二个查询没有使用任何不在temp表中的列，因此您根本不需要这些联接。这应该产生你想要的。

SELECT 
  t.country, 
  ROUND(AVG(t.totalpaid) OVER (PARTITION BY t.country),0) AS avg_country
FROM #totalpaid t
GROUP BY 
  t.country;

因为你两次加入你的table。
把 INSERT 以及 SELECT 您的查询语句等价于：

SELECT
  DISTINCT s.country, 
  ROUND(AVG(t.totalpaid) OVER (PARTITION BY s.country),0) AS avg_country
FROM (
  SELECT s.stud_id, s.country, SUM(e.paid) AS totalpaid
  FROM oc.students AS s
  JOIN oc.enrollment AS e ON s.stud_id = e.stud_id
  GROUP BY s.country ,s.stud_id    
) t
JOIN oc.students s ON t.stud_id = s.stud_id
JOIN oc.enrollment e ON e.stud_id = s.stud_id

在那里你可以清楚地看到table students 以及 enrollment 连接两次。这将产生一个倾斜的平均函数。

SELECT s.country, sum(e.paid) / count(DISTINCT s.stud_id) as average
FROM oc.students s 
JOIN oc.enrollment e ON e.stud_id = s.stud_id
GROUP BY s.country;

与此同时，我找到了解决办法：

SELECT
      country,
      ROUND(AVG(totalpaid) OVER (PARTITION BY country),0) AS avg_country
    FROM #totalpaid;

超级简单：）