I'm a 2-decade subscriber to Consumer Reports, but sometimes they just get their science (engineering) completely wrong.
A force doesn't bend an object. A moment does. That is, the propensity to bend is not proportional to the force applied. It's proportional to the force times the lever arm. i.e. A 90 pound force applied to one point on an object may not bend it, while applied to a different point it can easily bend it. So the bigger (longer) phones were actually resisting greater moments, even though the force was the same.
Another problem is the test they came up with supported the phone at both ends, while pressing down in the middle. Basically a
simply supported beam. The important thing to note here is that in such a config,
both sides of the phone are resisting the bending moment. If it took 90 pounds of force applied to the middle, then the left side was resisting 45 pounds, the right side 45 pounds.
When a phone in your pocket is bent, it is in a
cantilever configuration. One end of the phone is held rigidly, while the other end is free-floating. If the phone reached sufficient deflection to permanently bend in a simply supported config at 90 pounds, it will reach the same deflection at just 45 pounds in an equivalent cantilever (more precisely, 45 pounds pushing one way at one end, while your body weight holds the other end of the phone in place). You can try it in the calculators I've linked. Give both the same load, make the cantilever half the length, and you'll see the cantilever has twice the deflection. Make the load on the cantilever half that of the simply supported beam, and they have the same deflection.
(The actual force and moment diagram when you're sitting on your phone is a lot more complicated, since the force is distributed along the phone instead of all at one point. Integrating this is trivial for anyone who's taken a structural engineering course, but explaining it is beyond the scope of a forum post.)