You should probably get more specific advice from your advisor/lecturer/tutor/whatever, but the basic guidelines I think you will absolutely require is that you frame the question as clearly as possible and do so in a compact way that is simple to understand and direct. I'm guessing you will have to do some research before you narrow things down to a specific question though.
In terms of sources, again speak to whoever you need to. I imagine though, that you will be able to get a lot of documents in digital form to support your research and your claims.
For the referencing, you will have to talk to who is supervising you or giving you the assignment as they will have different standards in comparison and contrast to others. The most conservative view would be to simply cite everything you refer to explicitly: in other words copy everything you cite and put a reference to it in your reference list.
Your arguments that correspond to what you are trying to answer will be your own work even if they draw conclusions or make inferences based on stuff you cite.
To be honest referencing is an absolute pain in the neck and I'm glad I don't have to do it. I understand it's use and all but personally I'm more in favor of listening to arguments that have a sound logical premise and argument rather than some poor excuse to cite 1000 people, but then again I'm not getting marked and you are. The reason I say this is because all inference of any sort by default is under uncertainty and because of this, it is better to take a default position to consider as many things as equally likely as possible rather than say "it must be x because y said so": george carlin referred to this phenomenon as 'thinking'.