You can look at DxOMark's Lens Ratings, and particularly the Optical Metric Scores, which include a T stop measurement. I don't put too much stock in DxOMark's overall numbers (which don't have much practical impact for real use), but if you're interested in this particular thing, here's a way to tell.

Manufacturers do not typically give this number, so the measurements DxOMark gives are probably your best bet. Fortunately, they have measured a lot of lenses.

As you can see from the numbers, it's pretty typical for the light loss to be about a third of the stop from the nominal f number, although sometimes it's almost a full stop, and occasionally it is almost nothing — and you are right that in the latter case, it's usually prime lens designs. But not always; for example, the Sigma 18-35mm F1.8 DC HSM is measured as having a T stop of 1.8.

I think coatings can make a difference here, but that's clearly not the only factor, or the main one.

Another big factor is certainly going to be rounding, not just for precision but because a lot of camera and lens numbers are actually given as "close-enough" standardized values — see How to test actual focal length? for example. This is for a number of reasons, starting with variance in focal length with distance, overall irrelevance of being exact in real use, and almost certainly also because when more impressive numbers can be written on the box without really lying, of course they will be.

As for overall impact: sure, it can make a difference, and sometimes every bit counts, but overall, if the difference is less than a stop, it's not worth sweating over.

And of course, depth of field isn't affected.