Your math is right, but you are missing a tool. Your conclusion is right (given the assumptions) but it's coincidence.
One of the most important keys to success in a SnG is the difference between cEV (chip expected value) and $EV (your expected portion of the prize pool). $EV should be the most important consideration in almost every decision you make. Often you'll make a move that is +cEV and -$EV. This is not one of those cases.
http://sharnett.bol.ucla.edu/ICM/ICM.html
If you plug in all of the players' stack sizes for each scenario you outlined, you can see the effect of your action on your % of the prize pool.
You'll come out with something like this:
=.7*.1824+.067*.0599+.033*.2644+.067*0+.033*.2613+ .067*.0768+.033*.2367=.1620
Assuming instead that it's folded around to the BB, your $EV is .1567
Since it's higher if you push, you're correct here.
So that's the methodology. I think Durron had a problem with your assumptions. I do, too.