No stats, but originally I did disconnect the old battery in my mods; later I became lazy and did no longer care. I haven't found any difference yet. Leaking will not happen in either case, unless you completely remove the compound covering the old battery (but even then it's unlikely). However, since the old battery is a resistor when not being disconnected, the new battery should discharge faster. Not a single of my modded Dallas chips ran empty so far, so time will tell.

I was also thinking about internal leakage, that is, destroying the module's internal time clock chip. I'm not sure if the internal battery is entirely sealed from the clock IC inside the module, but there is a metallic path connecting the battery to the IC, so possibly it could corrode alone this path. Anecdotally, it seems like dead batteries more readily leak than charged ones. I'm not sure how much validity there is in this, but that was my rationale at the time.

Corrosion over the metal connection is unlikely, as it's a Lithium cell, not a NiCd battery. And I'm pretty sure these are dry cells anyway.

Leakage is the least I would care here at all. Should it ever happen, you can just replace the Dallas chip.

Thought I recall someone mentioning here that their board's coin cell battery was leaking... I don't think I've seen one of mine leak, but I usually recycle them once their voltage doesn't hold.

Yes, they can leak. But the result is some white residue that won't cause damage. I think there are examples in the SC-55 thread about removing the battery.

Anybody else perform this RTC battery mod and notice any difference in CR2032 battery discharge rate for disconnected vs. non-disconnected internal battery?

Disconnecting the internal battery is so easy after grinding through the epoxy that I can't imagine why someone would let it connected and take any chance of leaks and/or faster depletion of the new battery.

One thing I've noticed about these batteries. Most of the times I get a really old motherboard with the coin cell still attached, it measures 0 volts when in circuit, but immediately after removing it from the holder it starts building up a voltage... 0.8, 1.x volts... I guess that its internal resistance is high enough if that happens, and there's even an experiment that can be set up. Two CR2032 holders in parallel, insert one new and one depleted battery, measure the current once every few days/weeks, draw the conclusions.