Its interesting your article, one thing its not clear enought to me is why do you open your article with a GROUP and COUNT issue but later you develop a SELECT JOIN for 1-m topic ?

The first, taking case of GROUP BY, it would be iteresting to know how it works if you want to calculate for each ManagerId, two clumns, COUNT of workers with SALARY>1500 and COUNT of workers with SALARY<=1500, not using a CASE, but using a SUBQUERY to count vs. any other approach.

Related to the analysis of the execution plant its very helpful your comments and the bibliography you have referenced.

One last advice, not to use classic SELF linked case of EMPLOYEE-MANAGER as data sample since its not normalized as two tables but one table with references to it self.

Thanks,

L.