2) You need to understand probability, and natures bell curve. In mating, most queens will be average. Not average in a rating, but average to what you are breeding. The further away from the bell curve top, the more extreme in traits you may desire or want to stay away from. But the more you vary away from the curve, the less likely you are to produce those queens.
With that said, if I could show this on a graph, sliding the bell curve in one direction or another can be accomplished in small increments. If the curve peaked on the graph at say a rating of 50 (of whatever you are measuring), then through selection, would you be happy with queens being produced on average at 60%, 70%, or more? I know I would. So selecting even from the 1 or 2 best queens in your operation, and not raising queens from your 1 or 2 worst queens, will allow a better overall standard queen to be produced.
It's kind of hard to write out. I have all this in a neat powerpoint, and when you see the graphs and bell curves, most get it.
I just looked at NSQBA's website since I know there is a page on small scale queen rearing. But it is inadequate to the points you seek. http://www.nsqba.org/smallscalequeenrearing.html
I hope some of that made sense.