1 What is the research question?
"Research" is hard to impossible if it's not looking for something. We've had the process laid out for us early in the semester by Booth et al. and Lipson - there needs to be a progression from topic to theme to question. That's not a logic to be treated as relevant to the fourth week of class and then forgotten. Research questions don't fall out of the sky or jump off the page. You have to work - blood, sweat, and tears - to arrive at them.
2 Avoid the anthropology or sociology of me.
One can ground one's interest in a topic in one's own experience, but one must abstract from that starting point toward a question that is of wider interest. Sure we want to know what turned you on to the topic, but you you caring deeply about it is not a reason that others should be interested. There must be something that others are a priori interested in that you can show a connection to.1
3 Bias is not restricted to people whose conclusions you don't like
There is likely a high correlation between one's level of ideological engagement, one's acuity in detecting bias in others' methods and blindness in recognizing it in one's own. The more you care politically about something, the more likely you are to overlook your biases. A nod to "privilege" does not solve this problem.
4 You probably cannot write the sentence "There is nothing written on this topic" yet.
Some humility is in order here: either you have not found it yet or you have not read it yet. Suggesting it does not exist releases us from anxiety around something that we should be anxious about: there is more to read and we need to be reading it. The most valuable work you can do this summer is reading the things you should have read this semester. Any sentence that says something like "most of the research out there says X" or "looks only at Y" had better include a lot of footnotes so that your reader is persuaded you have actually looked at "most of the research out there."
5 Sampling on the dependent variable ruins research
Suppose I think one-eyed, Republican bird watchers face unique problems in getting their voices heard in political debates. A study of one-eyed, Republican bird watchers will not help to ascertain this.
6 Snowball and convenience samples should be where you might end up, not your initial aspiration
Most of the time when we propose snow ball or convenience samples what we are really saying EXPLICITLY is "I'm not willing to think about the work it would take to get a better sample." And IMPLICITLY we are saying "I don't really care about whether my findings tell us about any wider a group than the people I talk to." The expedience of the first can have pragmatic appeal. The apathy of the latter is unfortunate.
7 If you have the urge to prove something, pick another topic
Always ask yourself (and others) whether you are trying to find something out or trying to show something. When it's the latter, you should admit that you think you already know the answer and that the project is about marketing or political persuasion rather than research. No shame in those pursuits, but don't undermine the social value of science by dressing up advocacy in its robes.
8 The true test of a good topic and/or presentation is not when the audience says "Amen! I know…" but rather when they say "Oh, I didn't know"
Everyone likes affirmation from people around us. And a sure path to nods, pats on the back, and compliments is to tell people what they already believe. But that's not your job as a scientist.
9 Replication research is good. Redundant research less so.
Science depends on researchers who try to replicate one another's findings. This is especially true in the social sciences where it is far less common than in the physical sciences. One doesn't even need to justify the replication by finding some new population or context; straightforward "do what they did" replication can be a contribution. On the other hand, simply confirming what's already known (women earn less than men! finishing school while raising a kid and working is hard! urban school systems are underfunded! kids from well funded suburban schools do better in college admissions! male students talk more than women students in class!) is usually not a contribution. It's fine to say "Jones (2012) looked at X and found Y. I wanted to see if these findings were robust and so I replicated that study."