Yeah, in general with signal strength about the only things that is going to give you an idea on is interference from other wifi networks (and only if they are active. An idle wifi network causes minimal interference, only its beacon) and it is somewhat comparable on the SAME device.
If you look at most routers that have information from the manufacturer on connection speed, you'll see that a lot of them can connect at max speed down to around something like -68dBm...which generally means a stronger connection will have NO bearing on transfer speeds (other than interference, because in that case, the louder your stuff is, the easier it is to hear. SNR is not something that router manufacturers generally tell you, but they will sometimes tell you signal minimums for difference connection speeds). Often times it drops off quickly from max down to around 54Mbps which might be a difference of only 3dBm, from -68 to -71dBm. Then it might only be a drop of another few dBm to drop it to 11Mbps and another couple to drop it to 1Mbps.
Personally testing in my house, since I have zero interference, or near zero (closest other Wifi network is well below -85dBm) I see effectively NO difference in transfer speeds between the max signal strength of -25dBm on my router and my tablet and around -65dBm. The only difference is if I move around and suddenly get a big drop, the transfer speeds will drop a lot for a couple of seconds before jumping back up again. Once I get much below -65dBm I'll actually start to see my performance drop off and it is generally pretty quick. Between around -65dBm and -70dBm I'll go from ~10MB/sec to ~5MB/sec. From -70 to -75dBm it'll drop from 5 to 1.5-2MB/sec and from -75dBm to -80dBm it'll drop to around 1MB/sec. I'll still connect down to around -90dBm but the connection gets steadily crappier to one or two hundred KB/sec with complete drops outs for seconds at a time.
On my laptop I notice a slight performance hit once it goes below around -50dBm, but it is small. From around 22MB/sec to around 18-20MB/sec and then nothing further until it drops to around -68dBm. Then the same story as my tablet, except that my laptop is typically 40-100% faster at any given signal strength (its 2:2, versus my tablet which is 1:1).
On 5GHz, my signal strength is roughly 15dBm lower than 2.4GHz on my router according the InSSIDer, however I get roughly the same kind of throughput (actually about 10% faster on my laptop, but the same on my laptop 2.4 versus 5GHz) and in similar locations. Its only when I am really stretching signal strength that 2.4GHz is faster. For example in my hallway with a 4ft masonry chimney between my tablet and router and 25ft It'll run along at around 7-8MB/sec on both 2.4GHz and 5GHz. If I walk down the hall in to my bedroom and put another 10ft as well as 3 interior walls between me and the router suddenly I can't connect on 5GHz or BARELY (-90dBm to not seen, alternating) where I might get a spurt or two of 50-100KB/sec and then nothing for seconds at a time, but on 2.4GHz its around -75 to 80dBm, showing me 1-2 bars and I can generally get around 1.5MB/sec and pretty solid connection still.
So 5GHz seems to work at lower signal strengths at full/very high speeds than 2.4GHz does, but 2.4GHz deffinitely still has better penetrating power and range when pushing to the limits.
Anyway, the whole example is simply the point that signal strength doesn't necessarily mean everything. If you are comparing directly to the same thing, like -50dBm versus -70dBm on the same piece of gear, it MIGHT make some difference, but in generally once you are in to the good/excellent signal strength range, it has no bearing on performance. It might/probably make a difference once you are pushing the limits, because what was -25dBm at 5ft with no obstructions might be -70dBm with 50ft and 3 walls and what was -40dBm at 5ft with no obstructions might be -85dBm with 50ft and 3 walls and I'd expect the stronger router to perform much better in that scenario...but they still aren't necessarily equal (different CPUs, different firmware, different radios, amplifiers and signal processors)...the weaker router might have a much better time processing everything. Especially when you are talking MIMO, there is a HUGE amount of signal processing required to parse the different radio streams.
Its a small part of the reason why you might see one manufacturer list their N300 router as handling -68dBm at 300Mbps connectivity, -68dBm at 270Mbps, -68dBm at 230Mbps, -70dBm at 135Mbps, -72dBm at 54Mbps and so on and another one might list theirs at -71dBm at 300Mbps, -72dBm at 270Mbps, -73dBm at 230Mbps, -75dBm at 135Mbps, -78dBm at 54Mbps and so on.
Its not just about the strength of the signal, you have to weigh all of the processing bits as well as noise (if you don't have a lot of noise or if the noise is still well below your signal, then the strength of the signal doesn't matter much until it starts getting really low).
Keep in mind, this is a decible scale. That means the difference between 3dB, is 2x, 6dB is 4x, 10dB is 10x, 20dB is 100x, 30dB is 1,000x, 40dB is 10,000x and so on. From a strong signal of -30dBm to a weak signal of -80dBm is 50dB. That is 100,000x difference in signal strength. From a really strong -25dBm to -90dBm at the limit of connectivity of most devices, that is around 3 MILLION times difference in signal strength.
Same goes with interference, even a heavily used wifi network won't cause much interference if it is 30dB weaker than your wifi network. Its 1,000x fainter and your router/client should easily be able to be heard over it. The way Wifi works with OFDM is the edges of the frequency your stuff is using is a lot weaker than the center part of the channel, which is why a competing network can really start interfering once it gets vaguely close. There is roughly a 20dBm difference between center and the very edge of your channel, most is in the strong center part, but some of the data is transmited at weaker strength. So anything more than about 30dBm of your wifi network will really start to interfere. Weaker than 30dBm and in general the interference ranges from none to very minimal.
.
