That consumer reports test wasn't very scientific at all. It was rather disappointing in fact.
They did not test any variables besides force, and made no effort to replicate the real world.
For example, with a 3 point test, pushing on the middle gives it the least amount of leverage. In the real world, forces are often not placed precisely in the middle, causing a force multiplication from the leverage.
This also nicely avoided the known weak spot. You can even see it in their video/screenshots. Despite trying to bend it in the middle, you still see a bend and break at the lower volume button. How much less force would it require if each of the phones were bent on their weak points?
Also, what happens if the direction of the force is reversed? Perhaps they are weaker when the force is the other direction. I am certain we would see a more disastrous result here too, as it would compress the screen, certainly shattering it quite early.
Finally, what happens when the minimal force to bend the phone is repeated? That seems to be one of the key issues with the iphones: They are retaining their bent shape. A little force repeated over time could cause the permanent deformation to get worse over time.
I would really like to see a test where they have 2 static points on the left ( one on top, one on bottom over each phones weak point. ) then have the far right point being the press. This would better mimic a real world situation of being in the front pocket and someone sitting down. ( The top of the pocket, and a crease would hold onto the phone, then the leg bends up applying the force. ) As mentioned, it should be tried in both directions. ( This is definitely an important variable. I'm a screen in guy, but I've known screen out people. One could potentially have a very different result that the other. ) Record the minimum to get a permanent bend. Then repeat that same force to see if the bend worsens over time.
Perhaps some more people can help send in ideas for improving the testing, and actually see some real results from CR. You can email them from here:
http://web.consumerreports.org/customerservice/customer-service.html