My answer did not agree with that in the answer key, but I believe my solution to be correct. If interested, please take a crack at it before looking at my solution.
I second Todd Pedlar's post.
But as a personal pet peeve, I think questions like this are nonsense. Perhaps I am ignorant of something, but I cannot possibly imagine an engineer, scientist, or mathematician requiring the skill of translating a complex sentence structure into a formula.
The problem can be taken in a number of different ways. Since there are no limitations given on the range of these variables (no mention of fractions, max/min, etc.), I believe it would also be legitimate to interpret it as two separate formulas: g = b^3 and g = 1/(w^2).
That interpretation meets the problem's requirements perfectly in my mind. G varies directly with the cube of B, and inversely with the square of W.
Most people, when making problems like this, simply do not take the time and care to be precise.
The only time such a grammar-to-formula translation skill could be required is when a researcher/teacher/etc. is intentionally obtuse in recording the actual formula. In the real world, I would simply want to tell him to write it properly and be done with the issue.