Part of the problem is perception. Keeping this tied solely to convection when watches are issued and of course with the SPC convective outlooks...these are probabilistic forecasts - meaning they are that...probabilities. It is very difficult to verify probabilistic forecasts.
If there is a large say severe thunderstorm watch box that extends from say Maine through New England and maybe into Pennsylvania and New Jersey the majority of people would probably anticipate and expected widespread severe weather or a significant amount of reports within the watch box - but is that really true? Is that a fair assessment to make? All a severe thunderstorm watch means is that atmospheric conditions are favorable (key word) for any thunderstorms which develop to become capable of producing damaging winds, hail > 1'', or a tornado. Now you would expect that there will be at least some of these events within the box.
But what accounts for verifying? I'm sure the SPC has a defined criteria which is probably in a publication somewhere but if let's say there were a total of 10-severe reports within that entire watch box, was it really a bust?
Anyways and this probably answers the question, but there is a reason why watch boxes probably only get thrown up when there is an expectation for a certain number of reports.
But in terms of the SPC probabilities if we have a 10% TOR contour and no TOR's is that really a bust? That defines that there is a 10% probability for a tornado within 25-miles of any point within that contour...or in other words, a 90% probability there will not be a tornado. Obviously in terms of climo that 10% is going to be extremely high...but at the end of the day the probability is still exceedingly low.
I wouldn't be surprised if some years from now you see major adjustments with convective forecasting, how it's communicated, and even with probabilities and so forth as the data set grows and we become more sophisticated.