So You Want To Be An Astrophysicist? Part 2.0 - GRE at

So you want to be an astrophysicist?
You've suffered through 3-4 years of undergrad, and you're ready for more.
You picked the places to apply to (or have you...?), and you're ready for the paperwork.

Another lightly retouched blast from the past...

So what do you do?

First you apply to the departments.
There are 35 astronomy departments in the US with a PhD program, and some sub-departments within physics departments.
As a rule, go directly to the department web site you are applying to and read carefully (ie do not go to the Graduate School at the University, until/if the department indicates you should), then do as they say.
Application deadlines should be around christmas, either just before or after. Most places move relatively quickly on the decision process, to get offers out early enough that people have time to visit (most US places will invite US candidates to visit, many will pay or part-pay for the trip).
Decisions deadline is April 15th, that is to say, you have until then to accept, after that the department may withdraw the offer (yes, they may hold it open, but they will either be overcommitted or wanting to move on any waiting list candidates; please do make your mind up by the 15th unless there are exceptional extenuating circumstances).
Foreign grad applicants - note this is early, much earlier than, eg, universities in Europe make decisions. You miss the deadline, chances are you're out of contention for a year.

Research the departments you're applying to.
That is trivial now on the web.
Take the time to do so.
Find out what research is done in the different departments, who is there, the age profile, the student population, check informal student web sites, boring faculty blogs, pompous administrative blurbs - everything.
This is your life.
Do the work.
Start early, take notes and take the time.

So, admission decision is done by the department (or possibly a sub-set of the department if you're applying for astrophysics in a joint physics and astro department).
It is a committee (always a committee), membership changes each year, typically (or not).
There is a chair. That is a person, and that person can be dealt with.
Decision is not anynomous or robotic. If in doubt, contact the person (email preferred), ask questions.
A good committee will have a blend of people: the one who just goes by the GRE score; the one who likes a good GPA; the one who only likes students from universities they've been to (or heard of); the one who thinks research is all important; the one who carefully reads the recommendation letters, and the one who thinks activities showing enthusiasm and interest are important.
There may even be that many people on the committee...

The committee will have too many applications (they wish) and (some) want to cut on some pre-set criteria.
A lot of places cut on GRE subject score, or GPA or both.
Most everyone says they have a "preferred minimum score", but most will look below the cut-off (a few don't, they just cut the pile).
Some departments game the system, since they know the process is flawed and the indicators inaccurate. eg. some just admit everyone (above some minimum cut) and figure they'll flunk some out (so look at sizes of incoming vs graduate classes - 20-30% attrition is normal, higher than that, check what is going on; oh, and do a box car average, small field, Poisson noise dominates...).
Some places ignore subject GRE and look for either GPA; good letters; or, interestingly verbal GRE general score (more on that later).
Then there is some horse-trading (plus look at research statements, etc) and then offers are made, typically k times as many offers as target number of intake, where k is an empirical ratio carefully based on historical experience.
Some places make heavy use of waiting lists, others do not.

Important things:
Your research statement should be short, clear, free of spelling errors, and please skip the "career counselor" poetic blurb crap you learned in high school.
What we want to know is whether you're interested in theory, observation or instrumentation - or don't know; and, whether you think you know what sub-field you want to work in.
[side-note: when I applied, a long time ago, I didn't know about this tradition of the "application essay" in the US. My statement fit on the box on the form, two sentences. Except the last application I sent in - a friend saw it and said "you can't say that!" and made me write a full page essay. That one place rejected me. Bastards. ]

If you don't know, be honest and say so, half of the incoming students will change fields anyway. And don't all say you want to do cosmology, it is soooo sophomoric. Er, unless, of course, that is what you really want to actually do...
Other important thing - don't say you desperately want to do, say, VLBI observations of CO at high redshift, if there is no one at all at the institution you're applying to who does that stuff! Either customize applications to each university (which takes like 10 minutes electronically) or check that you're applying to universities where they actually do the research you claim to be interested in doing! Most departments are not big enough to have a presence in all sub-fields (maybe 1 or 2, but not really, not any more).

So, where should you apply?
Well, unless you're the next Feynman, and everyone else knows it already, you should be applying to 6-12 places. Less is foolhardy, more is gratuitous.
Usual application rules: apply to at least one "safe" place, and one "better than you hope for" place (unless you're applying to top places only, in which case have a backup anyway).
Pick places that do research that you are interested in; if you're not sure, pick a place that is big enough.
Check if you're free to move around between research groups or advisors after accepting within the department, you can in most places, but it is harder some places than others. Think about whether you're better off in a pure astronomy department (smaller usually) or a joint physics department (where astro may be a minority activity).
Then apply secondary criteria: cost of living (the stipend will be enough for you to survive, just, in most places); style of living (big city, town, rural?); climate (like snow? heat? humidity?); two body and family issues.
If you didn't get away from your hometown for undergraduate, then get out now, while you can!
Oh, and staying at the same university you did your undergrad is a Bad Idea.
Most places forbid it or strongly discourage it except for the most exceptional of extenuating circumstances; and it is still a bad idea.

So, where will you get in?
Even if you have a perfect GPA from a top ranked university and 99% on the GRE, there are still no guarantees on admission into any one of the top 4-6 schools; but anyone else will take you, unless you have something really bizarre in your application.
After that, the strongest determining variable in applications is the subject GRE score.
Some places cut on it; a high score guarantees you a close look from most places, a low score means you have to hope something in your application catches someones eye.
What is "high" - well, 50% is the crude cut-off line in a lot of places. Higher is better, lower is a concern.

So WTF is the GRE. See GRE

It is a set of multiple choice tests.
See comments for update: it is a set of multiple choice and written tests.
The General: quantitative, analytic and verbal sections.
The Subject (physics).
Scoring is reported by percentile rank, not raw score.
Supposedly normalised.
With declining numbers taking the physics subject test, this is a concern, since we don't know if the people now taking the test are more biased towards being good in the subject!

Analytic test is sort of like an IQ test, like one of those stupid MENSA tests.
Quantitative is straightforward and the only catch is that there's a lot of it; so practise doing dumb number and logic problems quickly.
Verbal has changed since I took it.
Here's the catch: astro and physics grad students have the highest GRE general exam averages. No one will be impressed if you do well on the general exam, it can only hurt you if you do badly. In fact you damn well better do very well in Q&A; so Verbal is really the only discriminant there.

Subject test: it is a friggin' multiple choice test. Such tests are inherently flawed; taking them can be gamed, and they are usually poor tests of knowledge or skills relevant to research. I despise it (and I did very well on it when I took it).
So why the emphasis on it?
Well, it works.
GRE subject score correlates reasonably with later performance - it shouldn't, the skills that make people good at fast multiple choice tests are not what make good researchers, so it must be due to a cross-correlation.
I still worry a lot that we miss good people who do badly on tests.

So, what is the catch?
The GRE physics test covers all of physics (caveat - I took it some years ago).
If you're a US student, then you will probably not have the breadth to do very well in it, unless you were taking intro calculus college classes while a junior in high school and went straight into advanced electives at a Good University.
The US system starts students too late in physics, does modular immersion courses with poor student retention (in both sense, students don't take additional classes, and individual students retain the material poorly). So most students do not have a comprehensive education in physics at the end of their junior year.
An equally good European student will outperform you on the test, on average.
So will the Asian students, if good enough.
(There was a cheating scandal, or two, some years ago, test forms sold on web sites in south-east Asia - this has tainted both past tests, and current candidates; universities have responded in different ways. Asia is still the source of some of the very best students, but they find it harder to come to the US to study.)
See comment below
Other catch: preferred test taking dates are Nov/Dec - the score from the earlier test comes too late to decide on the second test. If you take both (expensive) then you better do better on the second test. There is a test in spring, but then you take the exam with one less semester of study, which is a gamble.

There is a huge amount of analysis of GRE exams and their difficulty and how they compare and the sample exams.
Looking at the physics (gre) forums is not necessarily a bad idea.

So, what do you do?
I recommend taking the Nov test.
Take a test where you get the score back early enough.
You can get the score before christmas break and you can adjust your application strategy.
If you do much worse or better than predicted, e-mail or phone your advisor, that's what they're there for - to advise you. Oh, and talk to your advisor at the beginning of this process. Don't show up out of nowhere in the middle of january. and say you need letters of recommendation for PrincetonandHarvard, and, by the way, they're due tomorrow.
Bring a list of places with names of admission committee chairs, and due dates. Then send short, polite e-mails to remind the advisor the week before they're due.
Don't worry. Too much.

Aside: most universities now do rec letters through website, some are run by the individual graduate schools, others contract with commercial providers, which offer customizable rec forms. They are all different.
We Hates Them, We Do!
What ought to streamline, speedup and reduce the workload in writing rec letters is a nightmare of heterogenous crap design websites, some of which are so evil I will not use them (yes, I know some graduate schools require the departments to use them - know what, scan the f#@king letter in yourselves if you must - there is one commercial provider I will not use at all, the one that requires the faculty to sign off on a terms of use agreement to submit a letter - any dept doesn't like this, I'll be happy to send a letter handwritten in #2 pencil on yellow paper).

Ok, enough for now. There you go, don't have to worry about it for months yet.
Good luck, and apply to Penn State!