How to get in to grad school

Many people have asked me for help on applying to grad school. I think that they assume that, since I got into Caltech, I must know something. That's doubtful. I chalk it up mostly to dumb luck. But here's a core dump anyway:

• Are you sure you want to go to grad school? There's more to life than academia. Most of the best products and services that affect people's lives are developed and produced outside of the ivory tower.
• Are you thinking, "I want to help people," or "I want to make a lasting contribution to society?" Everybody thinks this at some point. It's a phase that most people grow out of. Your near term education should hinge on what society can do for you, not what you can do for society. When you're famous and have some skills, then it can work the other way.
• Write a good Statement of Purpose. Nobody will read it, but it'll help you figure out what you want to do. No, you may not read mine. This is a rite of passage, and you have to do it yourself. For science and engineering, the general format is straightforward:
  • Know why you want to go to graduate school. Describe this in the first paragraph.
  • Do not try to be clever, unless the mantle of Mark Twain has descended upon your shoulders. It has been known to happen, but chances are that most of the people who apply to grad school are more clever than you.
  • Past research success is the primary indicator of a good graduate student.
  • Name dropping is only appropriate for names of people who wrote letters of recommendation for you.
  • Don't say "I was inspired by Bob Kirschner's lecture...", because you need to inspire yourself to get through.
  • Write in present tense when possible. Emphasize the positive. Have a plan for what you want to learn.
  • Don't apply for a masters program if the school is at all reluctant about it ("non-terminal masters degrees are rarely awarded..."). If all you want is a masters degree, you'll get it faster by dropping out of a PhD program.
  • Use a spelling checker!
• Now that you know why you want to go to school, it's time to figure out where.
  • What specific two or three problems do you get really excited about?
  • Where do the people who study them work?
  • There are a lot of schools out there. You don't need to apply to every one.
• Get three good letters of recommendation.
  • If you're smart, you did research with three professors while you were and undergrad.
  • If you're really smart, your profs know people on the admission committee.
  • Give your prof the opportunity to decline to write a letter. Maybe she doesn't like you.
  • Recommending students is part of their job, but don't remind them of this.
  • Provide them with appropriate stamped, addressed envelopes for each university.
  • Remind them of due dates.
  • If you took a class from this prof, your grade in that class will show up in the letter. If a prof should ever give you a 4.0, make friends with her in case you need a recommendation later.
• Have good GRE scores.
  • Do not take the general and physics GREs on the same day.
  • Don't change your answers. If you do, you will usually change from a right answer to a wrong answer. Besides, if you have time to check your answers, then you are smart enough that you didn't make any mistakes in the first place.
  • Have someone drive you home after the test, 'cause you'll feel like a zombie when you're done.
  • The tests are written by the same people who wrote your college textbooks. The questions should be straightforward.
  • You don't have time to solve every problem. So don't. Use dimensional analysis and estimation.
  • Bring a random number generator with you. If you can eliminate several answers, making a truly random guess will improve your score. Making a "best guess" will lower your score, because the ETS is smarter than you are.
  • It doesn't test anything that you've studied recently. But that's OK, because you've looked at old tests, and you know what you have to learn:
    • Memorize every standard freshman mechanics and E&M question.
    • Caltech does not have a standard freshman class, so get a copy of Resnick, Halliday, & Krane.
    • You should be able to do all the homework problems in both volumes in one week, and most of them without paper.
    • There will be a couple of questions at the end about recent discoveries. Browse through Science News and pay attention to orders of magnitude.
  • Don't pay extra to find out your score early. It isn't going to change anything, and you won't want to take the test again anyway.
• Apply for a NSF fellowship. This requires one more letter of recommendation and one more essay. You probably won't get it, but it's worth a try. Better hurry, though, because the deadline is early. As a side benefit, your essays will have gone through an extra round of revision before you send them to schools.
• Include your GRE scores on the application. ETS can be slow.
• Complete your applications one at a time. Wait a day before mailing them. You don't want to send the wrong documents to the wrong universities.
• Send in your application early. A month early would be good. This means that you start working now.
• Are you sure that you really want to go to grad school? There are plenty of worthwhile things that you can do without a PhD. You might not get into grad school. What will you do then? Prepare yourself to find a job. Pick up a marketable skill or two.
• Recommended reading:
  • Beamtimes and lifetime, Sharon Traweek (Mil 7fl, QC774.A2 T73).
  • A PhD is Not Enough, Peter J. Feibelman. Don't follow his advice -- you want to become a scientist, not a politician. But read the book to get an appreciation for the culture you're getting into.
  • Alternative Careers in Science, Cynthia Robbins-Roth
  • Advice for a Young Investigator, Santiago Ramon Y Cajal
  • etc...