This is a gross oversimplification, when it comes right down to what's happening on the molecular scale. We are leaving out the fact that any particular molecule that bounces off the ground will certainly crash into another atmosphere molecule before it gets very far up.
In fact, any molecule on its way up will have many such collisions on its way up and down, and its progress will be not simple ("ballistic") force-free motion under the influence of gravity, but rather the net result of all of these collisions, some of which will propel it up, and some of which will of course propel it down. Interestingly, however, the fact that there are so many collisions means that they tend to largely cancel each other's effect. So much so, in fact, that the net result tends to be just about the same (at the level of this investigation, into the rough overall shape of the atmosphere) as if we ignore the collisions altogether.
In the pantheon of scientific models of gases, what we are doing here is treating the atmosphere as an ideal gas, as is mentioned further down in the explanatory notes. The principle error we make is actually not so much ignoring the influence of collisions ("nonideality") per se, but ignoring the influence of the inhomogeneity in solar heating of the Earth and of the rotation of the Earth, i.e. winds.