I'm going to take the contrarian point here:
If you're still designing elements for the game, a playtest is the wrong choice
See a much better written explanation here:
You need to stop playtesting. It's not just that you're doing it wrong, it's that you're doing it wrong in ways that hurt your games. Furthermore, by promoting a culture of design in which playtest is held up as the be all and end all of the design process, to which all other elements of design must bow down, you are causing at best cause other, newer designers to feel inferior and inadequate and at worse cause them to mutilate their own designs in ways similar to which you have mutilated your own.
Now, as he explains, there are excellent reasons to playtest, mainly to test the comprehensibility of your rules.
However, when working out the game mechanics, you should have a complete statistical model of your game. You must not rely on other people to figure out that the mechanics of your game do not match the intent.
Instead, using graph theory and statistics and game theory, you should be able to completely model the mechanical-theoretical (and mechanical-practical) domains of your game. You, as game designer, need to prove that every given combination of elements fits your intent and the expected behaviour of players.
This restriction of "complete mathematical model" also has the nice side effect of reducing reliance on silly systems that obscure and obfuscate the stats behind the scene in some vain hope that it will make them go away.
I currently have a Dice roll mechanic to cover combat/skill checks, character stats and skills. I have something similar to nature from mouseguard. Eventually I want to introduce feat like abilities but I don't think I'm able to make enough of them in time for the playtest.
This means that you aren't ready for a playtest. At all.
Instead, you should be using statistical models to look at all possible outcomes of the combat/skill check system and the modes of modification that feats introduce. Only when you believe that your statistical model is correct and complete (which by definition requires making all of the mechanical elements for that subsystem of your game) and rendered into some textual form, is it time to test.
You will be testing two things:
- Did you explain your system correctly?
You should be able to predict the outcomes of the mechanical interactions that your playtesters attempt. (See: making your beliefs pay rent.) If the outcomes do not coincide with your models, try to investigate where your model was incorrect or your explanation of system was incorrect.
- Did your models capture the complete range of action?
It is quite possible that you did not anticipate ranges of action being taken, and playtesting can show those.
The danger is Do not design your game for your playtesters. Instead, use them to identify flaws in your writing and your models of system.