A group of guildies and I took 3 level 20 pets into the desert, all of us spec'ed to 12 BM, one Elder Black Widow, one Hearty Black Widow, and one Dire Black Widow. We sic'ed all three of them on a Sand Drake, who proceeded to use Aftershock. A fixed-damage point blank AE makes a great health test for pets, and we found:
The Dire Pet lost roughly 20% of his hitpoint total.
The Elder Pet lost roughly 18% of his hitpoint total.
The Hearty Pet lost roughly 5-8% of his hitpoint total.
It was easy for us to run this test several times, just zoning in and out (the Sand Drake is the first creature just outside Elona Reach). Every time we got these exact same results....The Hearty Pet REALLY seemed to take the beating, even better than we had expected. We did not use any spells, calls, pet attacks, etc, and didn't bring any henchman...so we're certain that all three pets were subjected to the exact same amount of damage at the exact same time.
I'm speculating that perhaps the Hearty pet has a higher armor rating, because it makes no sense that an Aftershock would deal 20% damage to a Dire, and only 5-8% to a Hearty, even with the difference of 120 hitpoints between them. However, we didn't have any way to determine exactly how much damage, or exactly what % of their current health was lost, so the test is plenty flawed.
More tests to come in the near future, I'd love to find the exact hitpoint totals of each pet evolution. If anyone has any ideas or advice as to how to better test the pet evolution differences, post or PM me and let me know

-------------------
Edit: Tested the Aftershock damage against different Armor Ratings. The Aftershock *should* have dealt 74 damage to each pet, if it were true that they all have the same armor rating. And if their hitpoint totals were as speculated, we should have seen:
The Dire Pet lost 16.8% of his hitpoint total.
The Elder Pet lost roughly 14.8% of his hitpoint total.
The Hearty Pet lost roughly 13.2% of his hitpoint total.
But the hearty pet didn't lose anywhere near that amount of health, and the Dire lost significantly more than 17%. When the aftershock hit, the dire and elder pets were always VERY close in health, only 2-3% difference tops. The Hearty was always extremely far from each of them. The spread in the hitpoint totals of the pets was far more than 4%, which the math would suggest.
I think this further proves that the pets may have different armor ratings, and possibly not different hitpoint totals. I'll try to run a test using healing spells after the aftershock to see exactly how much healing it takes to repair the damage after the aftershock, and therefore see how much each pet is really taking >_>