People very often compare images which are shot at far too low shutter speeds. If for example 35 mm lens is used than shutter speed should be at least 1/60 sec, but for most people 1/80 or even 1/100 sec should be better. And if you compare those images shot at 1/100 sec without VR and those shot at 1/30-1/100 sec with VR turned on than pictures shot without VR will be sharper.
Not true. If lighting conditions require VR for short FL lenses, the faster shutter speed will
require higher ISO. I have never seen a case where the IQ loss due to VR was worse than 2-3 stops higher ISO.
I don't understand what's not true? Question was prime or VR, and primes are most usually faster lenses than most VR lenses.
Simple. For a given shutter speed, the prime will yield a lower ISO, due to it collecting more light (faster lens). Thus, if you need a particular shutter speed (such as to photograph moving children), then a prime is ideal.
However, VR lets you get a sharp photo with much slower shutter speeds than you can get with a prime (handheld, otherwise "VR" wouldn't have been in the original question). Take the 35mm F/1.8G. For a 50-70% hit rate, you need about 1/40th of a second shutter speed. Now take a nikon 18-200 VR at 35mm. That's about F/4.2. In terms of F ratio, F/4.2 is 2.7 stops slower than F/1.8, meaning the Nikon should require that many stops higher ISO (ISO 2600 vs ISO 400 for the prime).
Now, the VR on the 18-200 is worth about 3 stops in the field. So, if unstabilized requires 1/40th of a second, 3 stops better will be about 1/5th of a second. In terms of ISO, that would be a 3 stop improvement vs the ISO needed for 1/40th of a second (or ISO 350 vs ISO 2600)
I've managed to get about 1/8th of a second on the 18-200 to get a similar hit rate vs my 35mm prime. That isn't quite 2.7 stops better, but close, which is why I found that the ISO of my shots with my zoom was similar to my supposedly "better" 35mm prime. And that was what clued me in that the "primes are better for lowlight" idea wasn't entirely true in all cases.
Now, take a Tamron F/2.8 VC lens. Not only it it faster by 1.6 stops, but the VC is slightly better (IMO) than the Nikon version. What does that translate into? Well, at 1/8th of a second, it can yield a further stop ISO improvement over the nikon (either), yielding about 1/2 the ISO required vs the 35mm prime. (400 for the prime, x4 for the slower speed so 1600, divide by 8 for the 3 stops of stabilization yields ISO 200, or 2x better than the prime)
For a 50mm F/1.4G prime, the math works out similarly. The Tamron at 50mm is 2 stops slower in F ratio, but 3 stops better(due to VC) in how slow you can make the shutter speed yet get similar sharpness. Not only that, but the Tamron 50mm at F/2.8 is generally sharper than virtually any F/1.4 lens at F/1.4.
Practical experience has validated the math above. From my experience, for shooting stuff that doesn't move, a fast zoom with image stabilization (like the Tamron) yields lower ISO shots than a 35mm F/1.8G or a 50mm F/1.4.
The math doesn't lie, nor does the real world experience. That is the basis for my statement that VR is generally better if your target isn't moving, otherwise use a prime. And it's because you don't have to use the same shutter speed, VC/VR lets you use a
much slower shutter speed than you would otherwise need.