Image stabilization will prevent 3 stops of camera shake, but it does nothing for subject movement. You can take photos by hand at much, much slower shutter speeds with image stabilization, but the movement of subjects will be a problem.

Keep in mind there is also far more to lens speed than shutter speed. You also have the depth of field impacts of being able to open the aperture further. This can be a strength or a weakness depending on the situation. If you want to take a photo with a big depth of field and an unmoving subject, the stabilized lens will serve you better as it can get the shot at a lower aperture. On the other hand, if you need to shoot a moving target or want the highest degree of background blur possible, then the fast lens is the only way to go.

The optics in a prime are also much simpler and thus the image quality is generally better than a similarly priced zoom lens. Comparing apples to apples (same focal length and aperture), you'll probably still notice a slight improvement, but it will be countered somewhat by the gain that the image stabilization provides. If you use a tripod, the prime will have a clear advantage however.

A better comparison is really the slower lens with VR on at minimum f/number and the faster lens at it's max f/number. This is the real world apples to apples comparison for low light since you would shoot wide open for both and in decent light, the shutter speed is much less of a factor.

Your observations also seem to be fairly true across lens manufacturers for similar situations. I did a similar test with a stabilized Canon lens that was f/4 and compared it to an f/2.8 that was $700 more (both were zoom lenses). In that case, when shooting the same f/4 on both, there was fairly minimal quality difference, but when shooting freehand at f/2.8 and f/4, it became more noticeable and when comparing the background blur in a well lit shot, the added sharpness and the shortened depth of field became major factors in the quality of the lens. (To the point that the IQ of the f/2.8 was probably roughly double the f/4 with less than a 50% difference in price. $1500 vs $2200)