There’s a genre to AAA presentations — like for all conferences — and there’s pragmatic reasons for the genre. Basically, it goes like this:
1) A paragraph long anecdote from your fieldwork
2) The presentation of your argument and brief mention of relevant literature
3) Case #1 to explicate your argument — something ethnographically rich, but no longer than one and a half or two pages
4) Case #2, as above
5) Conclusion, including a recapitulation of your argument
Anthropologists love their introductory anecdote — especially one that presents a conundrum or puzzle. This should lead pretty naturally into your argument, so you should pick something that has a logical relation to your thesis. If you’re coming from a different disciplinary background — or are an anthropologist going to a conference in a different field — the format may differ slightly, but, generally, leading with one’s evidence is a strong way to approach a paper. If you can lead in with a quandary of some sort, it tends to pique audience member’s attention; if you start with some dry theoretical debate, chances are you’re going to lose everyone in the audience who isn’t particularly invested in that debate…
Your argument should be rather straightforward — don’t give people too much to think about: just one idea will do. Remember, your paper is only 7-8 pages long, and thoroughly explaining one idea will take up that space quickly. Your mention of relevant literature can be quite short — just a couple names will do, if that. Assume that your audience knows the genealogy of your argument, unless it originates from some other, uncommon source. But even if that’s the case, keep your discussion of sources rather short — interested audience members can always ask you about it later or look it up online. You’re there to present your ideas, not someone else’s.
I may be alone in this, but most of what I remember from past AAA presentations are the empirical cases that people present in their papers. They don’t need to be especially long, but they should be good ones — you want your audience remembering what you do and what your research is based on. If you have a website — either a personal one or one hosted by your university or academia.edu — make sure that the keywords that people might take away from a presentation are reflected in that website. I often remember what people talk about, sometimes their institutions, and rarely their full names. Years later, when I try and look someone up, those are pretty much the keywords I plug into Google. You want to make sure that your conference keywords are the keywords that will bring people to your work. That might seem commonsensical, but it’s important to keep in mind that there is often drift between how one presents oneself online and in person (and his or her research) and what has been presented in the past. Keep your metadata fresh, but also make sure it reflects what you’ve done historically.
If your panel has a discussant (or two), please be kind to them. Usually discussants will ask for papers by a specific date; be respectful of that deadline. Yes, things come up, and sure, there can be delays. But your discussant is probably counting on those kinds of things happening in her or his life too. So a rough draft of a paper is better than no paper or a late paper. Also: sending things like the whole chapter that the presentation is going to be carved out of or a published article that you plan on basing your presentation on are not appropriate things to do. Yes, you’re busy, but so is your discussant (presumably, that’s what he or she is your discussant, after all). Trying to guess what your presentation is going to be — based on too much or the wrong information — isn’t going to lead to a very good discussion. And, it’s likely to just mean that your presentation isn’t going to be included in any formal remarks about the panel. That can be okay: some discussants are great at winging it and might be able to fit in a discussion of your paper based on what they hear. Rather than burden your discussant with a guessing game, you can always just make the suggestion that you’d be happy to have more freewheeling comments based on the presentation you eventually give.
If a panel goes especially well, panel organizers often get the idea that it could become something more, and conference presentations sometimes get spun into articles or book chapters in edited collections. That can be a fine afterlife for a conference presentation, with the caveat that guest-edited journal issues can sometimes be treacherous, and edited collections are sometimes not counted as peer-reviewed (which may or may not be important given your professional needs). For more on publishing strategies, look here.
Here’s an example of a paper (my AAA presentation from a panel in 2012) and the book chapter it eventually became. If you’re looking for other examples, you can always look up old presentations in conference proceedings and email the author to see if she or he will share the paper as presented. Some people (like me) keep this stuff for years…
Finally, be sure to keep in mind the recommendations for accessible presentations. Printing out copies of your talk can be helpful for many audience members, and ensuring that any visual component is high contrast (e.g. having a white or black background with the opposite colored text) helps people make sense of what they see.