Mainly Agree here.
In the GMAT RC Inference you do not need to infer very much. In fact, when looking at answer choices for such questions, those that are furthest from the text are wrong.
It is analogous to my walking into a room sweating. Based on that information, can you infer I have been running? No. Can you infer that it is hot outside? Not really. That the heating is up in the room? Nope. All you can infer is that I walked in and am sweating for some reason (in paraphrased words of course). Okay this is simplistic but you get the idea. You should not go too far away from the text and everything should be supported in some way. The easiest way to check this is to divide the answer choice into components and check each one. You will find that one or more of these informational components of a wrong answer choice is missing or the info presented in the text negates it. Try it out with some practice questions. Compare wrong and right answers and ask yourself which is closer to the text? Most of the time you will find the right answer choice parallels the text most closely.
Economist GMAT Tutor