http://www.ssc.wisc.edu/sscc/pubs/sfr-combine.htm
The above explains things like for many-to one merge , which file is using data set?
"Next add information about teachers. In teachers.dta each observation represents a teacher, and each teacher has many students. That makes this a many-to-one merge (since the many students are currently in memory and the one teacher is in the using data set)."
