My car has a very sensitive throttle, say for the first 10-15% of the travel. Think of the behavior as lively acceleration. Several things led to suspicion of the TPS and it's hard to gather the pieces that led me down this path. Some of this is seat of the pants, others are bits and pieces on the Web. Also, I have noticed other posters here commenting about this 'problem', with no apparent resolution. Some of these same people complain about poor fuel economy.
From day one after I got the car, I have noticed that off-idle acceleration is brisk, enough that I need to really feather the throttle once I am at speed, in town or on the highway. On the Web, there are cases of TPS wear leading to 'out of range' operation of the MAF, basically by exagerrated throttle position signals. The non-linear behavior of my throttle may be part of this. I already have a diagnosis of 'out of range' MAF from a technician, not necessarily a bad MAF but of something related.
The on-highway OBDII scan shows a distinct on-off-on of the cruise control, reinforced by a seat-of-the-pants thing again. At 60 mph, the throttle is operating down in that sensitive part of the throttle range.
Also, there is more to some TPS signals than just the absolute throttle position. The rate of change of the position signal can also be used to enrichen the mixture, similar to the accelerator pumps on carburetors. If the Mazda's have that feature in the ECU programming, the mixture will go rich whenever a quick movement of the throttle is made, especially if the movment is exagerrated by the TPS.
This is a SWAG on my part, supported by some evidence. The TPS is supposed to be here today, an aftermarket piece from Rock Auto. I will document the results. The nice thing is that the scanner helps take some of the guesswork out of the process and provides documentation.
Does anyone know if the TPS is a variable resistance device or a Hall-type device? I can see the variable resistance stuff wearing out.