A few days ago I combed through the most recent 270 questions tagged diablo-iii, and identified all questions which had been tagged with such class tags, or had not been tagged with such tags, though they probably should have.
A google doc of all my data collection is here, though the data is already quite old, and Oak has already retagged some of them.
First, let's get some stastistics from the sample:
% of sample questions with at least one class-specific-tag = ~13%
% of questions within the overall tag with at least one class-specific tag = ~14%
So it looks like we got a pretty reasonably representative sample.
Before we go further, I want to talk about the individual tags themselves. When I first started collecting data, only diablo-3-monk had a tag wiki (though diablo-3-witch-doctor has since gotten one as well).
This is a bad thing.
Without tag wikis, it is less obvious to new users what they should be tagging questions with, or how the tags are supposed to be used (more on this later). Regardless of your views on whether they're working or not, these tags need wikis, a fact made all the more embarrassing since they've already got more questions than other, less useful tags that do.
During the approximate 2-day time period in which the 270 sample questions were asked, I found a total of 33 class questions and 9 questions that should have had the tag. My criteria for determining the latter category was simple: questions that referenced a class, class ability, or class achievement by name in the title but did not have the related tag.
This means that about ~21% of all questions that should be tagged are being tagged inappropriately, at least when first asked. (There were also a few questions which did not seem to warrant a tag ((but had it anyway)), but I did not include these in the figure)
And it's not just new users who are not tagging appropriately, as even 10k+ users have left it out of their initial question.
As an aside, I'd be curious to know (and unfortunately, do not know an easy way to compile this info) how many of the class-specific questions have been retagged to add them after the fact. Those ~21% were only the questions that other users hadn't edited already; how much larger would that number be without the diligence of folks like Oak?
Generic Questions and New Users
The saying goes that when you have a hammer, every problem starts to look like a nail. The same goes for the class tags. If you are playing a barbarian, you're more likely to tag your question with diablo-3-barbarian even if class doesn't figure into your question at all, or someone hasn't asked a dual-wield vs. 2-hander question about your class.
Finally, some new users do not understand that simply using the class-specific tag isn't appropriate and we get questions like this that need to be re-edited by users.
I am not really sure how to accurately judge the effectiveness of a tag, but I thought I'd get this data out to more people, and encourage their own reflections.