Tx, Your spec limit is the point at which you expect to see a failure. In other words, what is the probability that you'll see a short when you have .008" of solder paste? What is the probability that you'll see an open if you have .004" of solder paste? Your spec limits are very dependent upon your level of expectation with respect to quality as well as your component population. I'd love to be able to say .006" -.001" +.002" but that would be a total guess. The best thing you can do, is to not spiral into "analysis paralysis." Print 30 boards, pick 5 spots on each board but make sure your selections are representative of the finest and largest features on your stencil. Measure those 5 spots and get a standard deviation and average. Take your average and add 3x the standard deviation for the upper control limit and subtract 3x the standard deviation for your lower control limit. If your process is in control, you'll see that your control limits are withing .001" to .0015" mils of your average. Now, add / subtract another mil from that for your spec limit. Watch your quality data and make sure that boards that fall within the spec limits don't fail for shorts or insufficient solder. That will tell you if your spec limit selection was effective. You have to constantly watch your spec limits to make sure that the setting does indeed have a high probability of failure. Use your control limits to improve your process yet still remain fiscally responsible. Fiscal responsibility is important. It's very easy to set super tight control limits but if operators are wiping / cleaning every board for a misprint, how effective is our process?
When it comes to solder paste, you should worry about volume. Granted height is a part of the equation, but you could have perfect height and still get defects. I wouldn't get too wrapped up in volume either, unless you're evaluating stencil height, squeegee blades or such.