And yet... I wonder how much it matters. Many universities don't even look at the AW score when evaluating you as a candidate for graduate school. I think that's too bad. But for the purposes of this post, I'm assuming my audience is people who are hoping to score a 5.5 or 6 on the AW section because they know they're applying to a school that views AW as important.
If you're taking the GRE, I can safely assume you're an adult, so the advice I'm going to give you here is adult advice: work your ass off. What that means, in practical terms, is that you need to be writing, writing, and writing some more, and you need to be doing what you can to get better at it. Specifically, you need to sharpen your essay-writing skills, because Analytical Writing isn't creative writing: it's about engaging in logical argument and identifying flaws in other people's reasoning.
In a previous post aimed at high schoolers, I wrote on writing logically and clearly. You might want to give that post a visit, because it applies to at least one of the two AW tasks that await you on the GRE: the Analyze an Issue subsection. (The Analytical Writing section is composed of two subsections: Analyze an Issue and Analyze an Argument.)
Let's focus, then, on the second subsection: Analyze an Argument. In this section, you'll be given an argument that contains flaws, and you'll have 30 minutes to note as many flaws as possible while also suggesting ways in which to repair them. Here's an example from my Kaplan manual (Kaplan New GRE Verbal Workbook, 7th Edition; New York: Kaplan, Inc., 2011), and I can vouch that this example conforms to what you'd see on the actual GRE:
The following is a letter to the editor of a psychology journal:
The data collected from a variety of studies now suggest a relationship between the medicine Hypathia and the heightened risk of anxiety in patients afflicted with bipolar disorder. In 1950, before Hypathia was widely used to treat bipolar disorder, relatively few patients were diagnosed as anxious or had symptoms that suggested anxiety. However, in five studies published between 2005 and 2010, more than 60 percent of the subjects with bipolar disorder who took Hypathia demonstrated symptoms of anxiety or reported having episodes of heightened anxiety.
Write a response in which you discuss one or more viable alternatives to the proposed explanation. Justify, with support, why your explanation could rival the proposed explanation and explain how your explanation(s) can plausibly account for the facts presented in the argument.
The nice thing about Analytical Writing is that you're not required to name or classify the fallacies you detect. You won't have to talk about post hoc, ergo propter hoc or tu quoque or argumentum ad ignorantiam (and so on ad nauseam). It'll be enough to spot the flaws and identify them in plain language. Let's try spotting a few of them now, shall we?
1. As a general comment: I notice the entire argument falls prey to a massive "correlation is not causation" flaw (this is, by the way, the post hoc ergo propter hoc fallacy: event B happened after event A, so A must have caused B!).
2. There's a faulty diachronic comparison between "relatively few patients" before 1950 and "more than 60 percent of subjects" between 2005 and 2010.
3. Sticking with the faulty comparison for a moment, I'll also note that the time periods aren't being rigorously compared: we seem to be matching up (a) a vague, pre-1950 era with (b) a more clearly-defined period. If "pre-1950" means something like "1900-1949," then wouldn't it make sense to compare that era with, say, 1960-2009? And what if the data from 1950 to 2009 undermine the writer's argument?
4. The relationship between "more than 60 percent of the subjects" and "five studies published between 2005 and 2010" is unclear. One gets the impression that the writer is saying that 60 percent of all the subjects in the five studies, taken together as a single group, experienced some form of anxiety after/while taking Hypathia. But what if some of the subjects in one study were also involved in one or more of the other studies? Or: what if the percent figure is hiding a real disparity in raw numbers (e.g., if the five studies all had very different numbers of patients, with Study A having 20 patients and Study E having 10,000)? Or again: what if four of the five studies showed only 20% of patients suffering anxiety, while the fifth study had radically skewed results, thus leading to the 60% figure? If one study is an outlier, then we need to take it with a grain of salt, don't we? At the very least, that study makes it dangerous to lump the five studies together in the service of some general claim.
5. Doesn't a distinction need to be made between patients who begin suffering anxiety and those that suffer increased anxiety? Is the writer conflating causation with exacerbation?
Can you find any other flaws in the writer's argument? Think about it for a bit, but now let's turn to the question of offering solutions, because you need to do more than pick out argumentative flaws in this subsection of the GRE.
1. How can we fix the correlation/causation problem? If you're a David Hume partisan, then you know that the best you can come up with is a "regularity theory" of causation (event B regularly occurs after event A). Because this flaw permeates the writer's entire argument, we really need to focus on how to fix problems (2) through (5).
2. For (2) above, it would have been good to see less ambiguity about the number or proportion of patients being compared. A raw-number comparison might have made for a stronger case.
3. For (3) above, it would have been nice for the writer to have been clearer about the time periods being compared. Why is "pre-1950" being compared with "between 2005 and 2010"? A more rigorous comparison would have noted (and related) trends between corresponding time periods.
4. For (4) above, well... a lot needs to be done. It would have been nice had the writer both spelled out the number of patients involved in each 2005-2010 study and offered similar figures for a corresponding pre-1950 period (say, 1945-1950). The writer would also have strengthened his/her case by noting whether there had been any patient overlap between/among studies.
5. For (5) above, the writer should have been at pains to discuss what s/he feels Hypathia does. Does it cause anxiety? Does it increase anxiety? Is there some sort of equivalence between causation and exacerbation?
It occurs to me that I've missed one rather obvious flaw in the writer's argument. Can you guess what it is? I'll write it below in white font between [brackets]; just highlight the space to see my thoughts on the matter. This flaw may, in fact, be one of the very first ones that you thought of.
[The writer never considers the possibility that something other than Hypathia might be causing the anxiety. Even if only to dismiss the possibility, the writer should have at least raised the possibility that the anxiety's etiology might stem from a different source.]
There are doubtless other flaws to be found, but if you were able to pick out the ones that I picked out, you'd be on your way to getting a 5 or a 6 for this essay.
Remember that this is an essay you're writing. You might be asked to write it as if it were a letter to the editor, but it's an essay all the same. The quality of your organization counts, as does the quality of your writing ability in general. Avoid typos, poor grammar and mechanics, misspellings, and all the other pitfalls that come with writing under pressure on an unfamiliar computer keyboard.
Based on my own experience with the AW section, I'd recommend going for at least five paragraphs-- more if you're a fast and competent typist. Organize your paragraphs this way:
1. quick intro
2. flaw #1 & #2 detected; solutions to patch them
3. flaw #3 & #4; solutions
4. flaw #5 & #6; solutions
5. quick conclusion
[NB: the Kaplan instructions call for a bit more than just finding problems and proposing solutions, but the question as written will still allow you, more or less, to craft an essay along the lines of the above-suggested template.]
For both of the AW subsections, leave yourself time to proofread your essays. You'll be embarrassed at how many mistakes you can make when typing in furious desperation.
Apologies for concentrating so heavily on the Analyze an Argument subsection, but I feel that the skills needed for the Analyze an Issue subsection have already been adequately covered elsewhere (and here as well).
In the meantime, get writing! Start up a blog; write letters to the editor; append detailed, well-argued comments to blog posts and news articles, and watch how your arguments get picked apart by the ensuing commenters. Learn from these experiences; strive constantly to improve your writing until you feel confident that you can "argue on command." Grad school involves plenty of discussion and debate: you'll often be required to think on your feet, so the GRE isn't a completely unrealistic simulation of that sort of pressure.