For talking about complexity, we ignore constant factors. Its all about how the function scales - N^2 means, if you double the elements, the running time quadrupples. It doesn't matter if its 1/2*N^2, N^2, or 100*N^2, scaling always works the same. (For those three: N=2 => 2, 4, 400. N=4 => 8, 16, 1600).
In the same way, only the highest exponent counts. If you've got quadratic scaling, adding (+) a linear scaling (N) still results in a quadratic scaling in the end.
So yeah, a function might have a complexity different than exactly N, N^2... but generall you'll just see O(N^2) since thats just how the notation works. You can easily have a O(N^2) function that runs faster than a O(N) function for small Ns with certain constant factors. But eventually the O(N) will always be faster than the O(N^2) function => if you have (n^2)/2 and 10000000n for example, you can calculate the point at which the quadratic scaling will start to be slower (and it will NEVER return to being the other way around at that exact points, thats important) :)