Often ignored in weighing is the need for consistent environmental conditions. Powder can absorb a percent or so of moisture weight and then needs the equivalent of another half a percent or so in energy to evaporate it. Someone mentioned recently that the Norma manual also mentions that increased moisture content lowers powder burn rate, so you can add that error source into the mix. So if you don't keep your powder in fairly consistent temperature and humidity conditions, you can get something on the order of the equivalent of a 2% error in charge error even though you weigh every charge to the nearest tenth of a grain.
If you dispense by volume perfectly you eliminated the moisture weight portion of the error above, which is about half of it. This is one reason benchrest shooters loading at the range use the high dollar powder measures rather than bring a scale and a draft shield. When the climate changes, they want to minimize the effect of moisture content.
The other factor concerning weight vs. volume is that stick powder burn rates, in particular, vary with packing density. Some will self-compensate in some cartridges with some bullets. Other combinations won't work out that way. That is, if you get a charge that's heavy because it is packed more densely, it also is slower to burn by an amount that compensates for the heavier charge quantity. Other's don't work out that neatly.
An example of compensation would be Hatcher's report of a powder that metered so poorly that it gave a spread of 1.7 grains from the equipment used to load National Match ammunition, but that shot consistently more accurately than a shorter grain similar burn rate powder than metered to just a 0.6 grain spread on the same equipment. All other components were the same.
An example of not compensating would be the late Dan Hackett's report of a load that shot well when loaded at home, but that caused sticky bolt lift when he loaded it at the range. Same charge. What he finally traced the difference to was the powder settling due to vibration in his vehicle when transporting it to the range.
Powder measures tend to vibrate some during use, particularly when cutting grains, so powders that can pack much with vibration tend to pack to different densities over the course of the hopper being emptied. So to get the best use of a measure, you want a powder that tends to compensate well in your load. If you don't have that, you may have a situation in which weighing is going to be your better option.
Toward weighing, because of the packing issue in measures, the newer electronic weight dispensers are probably an advance over the old method of throwing a charge of varying density, then trickling up to final weight. This is because the electronic dispenser spreads the grains out and vibrates them uniformly into the pan. I would expect, therefore, the packing density of the resulting charges to be more uniform. So if you have consistent temperature and humidity conditions and are weighing, as long as you handle funneling the powder into the cases the same way each time, this should be a very tight charge metering method.
At some point you will have to wonder why we set powder measures up with a scale, and if we go back and tweak the measure because the scale says it has moved off target weight a little, then, in fact, we are really metering by weight rather than volume, and the whole thing feeds back on itself. Well, you have to start somewhere, and manuals have loads by weight, and even Lee's VLD's are used to figure a volume for a weight or vice versa. So why don't manuals publish loads in cc's or some other unit of volume? The answer is that the bulk density of powder varies even more than moisture affects do. If you look at Accurate Powders, for which the distributor, Western Powders, publishes bulk density and its tolerance, you see some vary by over 5%, lot-to-lot. That means VLD numbers are no more accurate than that.
So we start with weight. Then we work up a load. Then we are where we are. It's often not exactly on any published number for charge weight or volume (based on a powder VLD). It's only certain to work with the lot of powder we used to develop the load. If we get there by using a micrometer adjusted powder measure for which we have determined how far to change the adjustment to get about one test charge of weight increment, we can actually work a load up just using the measure and record the setting. Again, that's just good for the lot of powder we have and we'll have to do it again if we change lots.
Interaction between scales and the powder measure that are useful are to find a starting load. To see if your measure trends toward increasing or decreasing charge density as the hopper gets lower. Seeing if a different dispensing technique seems to produce less variation in weight (and therefore in metered density) than another. Lots that is useful may be done there. But I don't think it's necessarily useful or even a good idea to readjust a measure in the middle of loading an ammo batch. That's as likely to introduce an inconsistency in performance as not. As always, though, YMMV, and it should be tested.
. 
that matter.