Our system is sort of based on the 'skills' for the various sections:
- Problem-hypothesis-materials-procedure take a very systematic, structured approach, and are substantially similar from one competition to another. However, the basic structure needs to be adapted quickly to the specific situation of each event. The main skills involved are meticulousness and a good memory.
- Analysis-errors-conclusion-recommendation are basically writing. That's what ties them together in our system; they're the sections that basically have to be free response, straight-up mini-essays. That's definitely a skill unto itself: writing a lot, fast, coherently, and making the end product sufficiently eloquent and complex to impress the judges xD. [So even though the analysis is based in the statistics, the way it has to be presented groups it with the conclusion etc. The person does need some understanding of the statistics, but I think there are more people who understand the implications of mean/median/mode/range than who can write with the necessary clarity, complexity, and speed, so that's usually not an issue.]
- Collection of data (and the attendant stats and graphs thereof) needs a steady hand and at least a modicum of organization, and obviously the ability to calculate stats and graph data. In a perfect world, this would be the most straightforward section, but since data in this event doesn't always show what you want it to... this person gets the job of making up what the data "should" be if necessary (yes, bad scientific practice, I know; however, given the time-constrained nature of the event, sometimes things don't go quite the way they would in a real lab setting- e.g., sometimes the conclusion is written before the data collection is actually finished, so if the remaining data doesn't fit the conclusion, it can be easier to change a few numbers than change an entire conclusion). That takes a real ability to both understand the data and think on your feet.