ArikBY,
The sinking/sourcing relationship is not necessarily as simple and straight forward as you suggest.
http://www.plcs.net/dcforum/DCForumID1/2645.html
MACI,
As far as Field Sensors (Inputs), there was a time (not long ago) when the selection was pretty limited. These days it seems that, for any given sensor, there is an NPN version, a PNP version, and quite often there is an NPN/PNP (selectable) version.
Also, not too long ago, some PLC's restricted the Input devices to one type or the other - at least, for a particular card. Some manufacturers provided both Sinking Cards and Sourcing Cards. That issue has just about disappeared. That is, these days, generally, PLC Inputs accept either type of input device.
Now, regarding Outputs, if you are using solid-state Outputs, they are either NPN-Sinking, or PNP-Sinking, or NPN-Sourcing, or PNP-Sourcing. Refer to the manual to determine the case for the particular PLC.
Of course, Relay Outputs don't care... as long as you are careful to maintain the proper relationship within any "grouped" Outputs.
As far as troubleshooting goes...
Each method has their own set of pluses and minuses. And in any case, you still have to determine the particular configuration before your readings make sense.
So, as far as which is more common in terms of a NEW construction, the real answer lays not in the historical evidence, but rather, in the current availability of the various versions.
Since there appears to be less and less difference between the availabilities of the various types, it boils down to cost. That would include the cost of the sensor and any extraneous expenses associated with the design and installation of the particular sensors - whatever those might be.
As far as which method is better...
It's getting to be more like "six-of-one or a half-dozen of the other".