I think what you may have overlooked is that mods are not 100% full on or full off. ...
... i disagree that you won't feel the effects of a sway bar change at safe speeds. the mazda5 has horrible suspension tuning from mazda, and will benefit in driving confidence from a simple sway bar and damper upgrade. even a simple change to mazdaspeed3 swaybars and konis/bilsteins will be noticed during daily driving activities.
I totally agree that mods aren't on/off. But I am making a distinction between ride motions and vehicle "balance". Things like stiffer springs and shocks are easily discernible at low speeds regarding their effect to ride comfort and roll gradient deg/G (what "most" people associate with a "good handling" car - it is stiff and corners "flat"). However, while it technically exists, the affect on the balance of the vehicle isn't easily discernible at low speeds/lateral accelerations/slip angles. Here is why, from a vehicle dynamics perspective:
This balance is also commonly known as understeer or oversteer, or in vehicle dynamics terms is the "understeer gradient", K. K defines whether the car is inherently understeer or oversteer prone.
Skipping all the derivation, steer angle, delta, required to negotiate a turn is:
delta=(L/R)+front slip angle-rear slip angle
If you do some math and convert slip angles into lateral forces using cornering stiffnesses, the equation becomes:
delta=(L/R)+K(V^2/R)
Or in terms of lateral acceleration, Ay:
delta=(L/R)+K(Ay)
Note the V^2 term - lateral acceleration is dependent on velocity squared, as is the steer angle. This is why autocrossers go to such extreme lengths to try to get the car to "rotate", because physics says that at low speeds the effect of K will be small. This is why a properly set up auto-x car (30mph corners) will be virtually undriveable on a track with 100mph corners. And consequently why it's difficult (but I will agree not impossible) to evaluate your latest swaybar change at legal (low) speeds (and consequently low lateral accelerations and slip angles)
Here's a graph that shows the velocity squared effect:
Btw, how do you feel about removing both bars all together or going as small as possible on the track?
(Assuming we are talking auto-x/track setups, not street)... I would not remove the rear bar in this FWD car, you typically need a lot of rear stiffness in a FWD vehicle and this car is no exception (in my admittedly limited experience). Removing the front bar is a definite option, and common in many FWD auto-x setups. Removing the FSB often allows better (earlier and stronger) power application, which in a FWD
My auto-x'ing with this car reveals that some form of RSB was needed for my setup, but I was delaying my purchase of the RSB to save that money for the turbo. I was planning to test the disconnected FSB and the event this weekend, but I stumbled into a free ms3 RSB from a local. So instead I will be running this weekend with just the RSB (only change one thing at a time logic). If it becomes evident after the first run or two that the ms3 RSB is not "enough" for my tastes, I *may* try to disconnect the FSB between runs to test out the setup without it. That would be a nice back-to-back test that could give some pretty good data about which setup is faster, if I do choose to try it.
Keep in mind that my spring rates are stiffer than stock (though not super stiff by racing standards). I have 450lb/in springs, IIRC the stock springs are under 200? So on stock springs the ARB contributes a higher percentage to the effective wheel rate during cornering than it would with stiffer springs (so far I've seen some serious ms3 guys running 1000lb/in in the front). The effect of running a disconnected ARB with stock springs may be significantly different than with stiff springs - turn-in response and front camber control during roll may be unacceptable with a stock sprung vehicle, whereas they may be acceptable and able to be tuned around with a modified vehicle.