2 people mentioned it here but incase you missed it
1. Larger transistors have a lower gain, eg hfe of 100 vs a small transistor npn might have say a 300 gain
In otherwords, if 1ma is allowed to flow via the base, the collector / emitter a smaller transistor would allow 300ma but a bigger transistor only 100ma
Say you only have 1ma current from a micro controller, you need 1amp of current, it's not going to happen so using a darlington pair you can get a 1000 gain
The other thing is switching speeds and other stuff like capacitance