Perhaps Zivilin could elaborate, but I found this post (from another thread) very useful with regards to experiment design:
[...]
1. Run a few controls to determine variance.
2. Each experimental group should perform identical tasks, i.e. face identical enemies.
3. Consider the type and quality of data gathered. Perhaps time to kill, or number of dead enemies after a given amount of time, etc.
4. Presentation is important.
[...]
Every project has its own nuances. In general:
1. Familiarize Yourself thoroughly with the researched subject. Decide on exactly what You are setting out to prove/disprove.
2. Think long and hard on what types of parameters/data will conclusively prove/disprove Your hypothesis.
3. Design an objective experimental setup.
4. Identify all conceivable variables that may influence Your setup, and modify the setup to eliminate them. Repeat until satisfied that all variables are eliminated, or at least accounted for.
5. Ascertain that Your results proved/disproved Your hypothesis. See if any other interesting research byproducts can be extracted from collected data.
6. Present the results in a clear manner, to the best of Your abilities.
And while I'm posting, I believe
Modding material properties vs. how the properties are used by the game has a place in this compendium, possibly under a new "Game Mechanics" subsection. It has a treasure trove of material and combat mechanics-related information, forum-exclusive.