Ph.D. tips
Here are some thoughts I have about how to make the most of grad school, loosely divided into themes. Of course there’s lots of different ways to enjoy your PhD and produce meaningful work, and there’s certainly lots of things I wish I had done differently and better. But perhaps there are nuggets of useful stuff in here.
Productivity
- Keep a running list of potential research ideas. Every time you have an idea for a question that’s interesting or a study you might want to run, write it down. You never know when you might want to mine your past thoughts. 99% of the things I’ve written down are stupid, but it’s helpful to get bad ideas out too.
- Organize your work flow as much as possible so you don’t find yourself re-doing work. Everyone has different systems, but in general, you should have a system for:
- Organizing data
- Organizing analysis code and results
- I think the ‘results’ aspect of this is super crucial. When working on projects, you’re likely going to try out a million different analytic approaches (especially if the project involves modeling) and ideally, you want to keep everything organized from the get-go so you know what you’ve tried and what you haven’t tried, and can easily pull up output/ figures without having to re-run everything every time you want to check something. R markdown files where you ‘cache’ models can work well for this, but so can just writing code that outputs figures / results with a clear naming structure. The key thing is to document everything so when you pull up a random figure or results file, you know what you actually did. Again, R markdown files / jupyter notebooks where you can add text are helpful, but there’s a lot of systems that can work for this.
- Organizing papers
- Organizing notes on papers (this is the hardest and I still don’t have a good system).
- Organizing notes on meetings (see ‘meetings’ section below).
- Plan out the work you want to do each week, each day, and in each chunk of day. Once you have figured out what you want to work on, break every task down into as many little, concrete tasks as possible. This serves many purposes:
- By spending time planning, you can be more efficient when you actually go to execute something (e.g., you won’t have to backtrack because you forgot to do something).
- Little tasks are more manageable than big tasks (e.g., ‘convert existing function from project A to work for project B’ is easier than ‘implement model’). By figuring out all the little steps involved in doing something, you can more realistically assess how long it will take and where you are likely to hit snags.
- There are lots of tools you can use for these things but many of them are unnecessarily complicated. I’ve exclusively used the free apple notes app and google docs for 10+ years. The one valuable ‘feature’ they have is they are cloud based so I can access on multiple computers/ phone/ etc.
- Never set goals based on time. Work expands to fill the time you have. Don’t tell yourself you’re going to ‘work for one more hour.’ Instead, pick a few of your small tasks, and decide which of them you’d like to finish.
- When you feel like you can’t motivate yourself to do anything, either: force yourself to complete one of the small, concrete tasks on your list OR read a paper.
- Usually, my list of small tasks involve some easy but tedious things that I can just force myself to get done. Then I’ll feel better about having done something.
- Reading a paper is ’easy’ in that I know I can do it successfully, and often reading about cool things other people have done inspires me to do more and better work.
- People have different systems for rewarding themselves for completing tasks. I find that causally irrelevant rewards don’t work that well for me. For example, telling myself, “I just need to finish X and then I can watch TV” doesn’t work that well for me because there’s nothing stopping me from just watching TV now. Instead, I like to think about the fun and interesting stuff I will be able to do as a result of finishing something. Things I generally think are fun are: seeing what data from a new task looks like, discussing results from an analysis in a meeting with mentors / collaborators, doing an editing pass over a written draft, sharing my work in any capacity (tweeting a preprint, presenting at a conference), planning a new experiment, etc. and pretty much anything that elicits external validation. Things I generally don’t think are fun are: programming new experiments, any of the steps involved in data collection, debugging analysis code, writing responses to reviewers, etc. Of course, these lists are going to vary a lot across people, and for me, they vary on a daily basis. Writing can easily fall into both categories. But I find the best way to motivate myself to get something done is to think about what it will lead to. Debugging analysis code is going to lead to me being able to share and discuss results, write a paper, and design a new study, and those are things I generally enjoy.
- If you find yourself not being able to look forward to any stages of the research process, that might be a good sign to do some self-reflection and decide if you want to be doing research at all. Of course, that’s not to say you’re not going to find activities other than work more appealing, but if you don’t find any aspect of this rewarding in and of itself, you might want to reconsider. There’s not really anything on the other end of a Ph.D. except the opportunity to do more research. If you don’t like it, you shouldn’t do it. Your work probably isn’t helping the world very much — at the end of the day, it’s really for you.
- I think it’s also informative to think about whether I’m prioritizing other things in my life versus procrastinating. There have been times where I’m busy with lots of things that I generally think I should prioritize over work (spending time with friends visiting from out of town, supporting friends / family members in crises, taking advantage of opportunities to travel, etc.), and at these times, I like to cut myself some slack and not feel bad about not working. But there have been other times when I intend to do work on a Sunday afternoon and instead watch 5 hours of Gilmore Girls episodes I’ve already seen, and I think in these cases, it’s helpful to reflect on what’s up with my motivation (Do I not like what I’m working on? Am I secretly stuck on something?)
- Set deadlines for yourself. It’s hard to work without deadlines, and setting them for yourself is hard because you will know that they are fake. I sometimes like to get around this by scheduling meetings to do a specific thing (e.g., read a draft) so that I know I have to finish that thing before the meeting.
- This can backfire if you still don’t take the deadline seriously, because then you’re wasting the time of the person you asked to meet with.
Working hours
- Everyone has different feelings about how much one ‘should’ work and it’s a loaded topic I don’t want to weigh in on. In general, the more time you spend working, the more you can do, but the relation is obviously non-linear. Everyone also needs time for life, and the amount of time needed for ’life’ is going to vary based on a million person-specific factors.
- I try to minimize the amount of time I spend in ‘half-work’ states, because they are the absolute worst. A ‘half-work’ state is when I pretend I’m writing a paper but actually I’m just checking twitter or browsing the internet for random things. It’s neither productive nor fun. If I find myself drifting into this, I will sometimes just fully commit to not working because I know I’m unlikely to be productive, and I might as well actually just relax.
- I try to be honest with myself about how much I’m actually working, and how long things actually take. The flexibility of grad school is great, but I think too many people interpret ‘flexibility’ as meaning that they don’t have to work that much, i.e., flexibly devoting classic ‘working’ hours to fun, but then not making up that time anywhere else.
- My hot take is that I think most grad students should spend their first few years working 9 - 5 (or some regular schedule like that), in the lab every weekday. You can get a lot done in this amount of time, and I found it helpful to establish a routine like this early on. Establishing a set working routine means you don’t have to decide how much you are going to work everyday, because you have a default policy for what to do. As soon as you give yourself freedom to choose when to work, you then have to converse with the devil on your shoulder that’s telling you that oh wow there’s 7 seasons of Great British Bakeoff on Netflix. Make things easier for yourself and just pretend it’s not a choice. Stick to your work routine, get shit done, and feel good.
- One note on this is that your PhD literally IS a job. You are getting paid to learn and do research. The ‘flexibility’ of the job really means that you have flexibility to figure out how to divide your time between different things (e.g., reading, learning new skills, working on classwork, directly working on research, etc.) but all these things should be PhD-related in some way.
- Sometimes Ph.D. students justify working less than they are expected to because they feel as though their salary is too low. I think that it is completely reasonable for students to feel as though they are not paid enough and to advocate for higher pay. However, as with any other job, silently working less than the amount you are expected to work is not going to rectify the underpayment problem, and in fact, likely only bolsters the case against paying students more.
- Some people feel the need to completely shut off over the weekend. I find I do my best work on some weekends, from a combination of having uninterrupted time and feeling like anything I accomplish on a weekend is a win. I think that if you are working roughly 40 hours per week, you can finish a Ph.D. in a reasonable timeframe without working on the weekends. However, I think extremely few people successfully log 40 hours of true, focused and productive work during the week.
Meetings
- I love meetings because talking to other people about my work (or their work) is one of my favorite parts of research. But to make meetings good, you need to do a lot of pre-meeting and post-meeting work. As a trainee, it’s important to remember that for the most part, you are in charge of your meetings, and you can control how useful and engaging they are.
- Whenever I have a meeting scheduled, I typically spend 15 - 30 minutes strategizing for it, not including the time I spent doing the work the meeting has been set up to discuss.
- To strategize, I list all the things I want to discuss and the specific questions I want to emerge from the meeting with answers to. Then I order them by priority so I can start with the most pressing things first. After that, I go through each thing, and think about all the information the people I’m meeting with need to be able to answer my questions and help me. In general, I rely on my advisors to help me make decisions, but I assume that I am capable of and responsible for gathering all the information necessary for making those decisions. Their role is then to consider the information I bring them and advise.
- For example, sometimes I want help deciding on which version of a figure I should include in a paper draft. I try to make sure I have both versions of the figures already pulled up on my laptop prior to the meeting. Of course, finding and pulling up a relevant figure only takes like 20 seconds, so this doesn’t seem like a big deal, but I may have 10 of these types of decisions to get through in a meeting. I don’t want to waste 5 minutes out of a 30 minute meeting looking for files on my computer.
- For more complicated analysis discussions, I make sure that I have clearly written out what each analysis I ran actually is. Again, I don’t want to have to spend time in a meeting going back and being like, hmm what did I do again? I can easily do that myself beforehand.
- In discussing results of experiments, I try to do all the analyses and create all the visualizations I think might be useful to understand the data before the meeting. Obviously there’s some limit to this (you can never do everything) but I try to anticipate which things might be most useful to have.
- I try to send any relevant information to people I’m meeting with before the meeting. How much before varies depending on the person I’m meeting with and the topic of the meeting (e.g., if I know people won’t look at stuff until the meeting, I only send ~5 mins before. If the meeting is to go over complex results, I’ll try to give people time to digest findings prior to meeting).
- At the end of meetings, I’ll also look at my list of specific questions that I brought into the meeting and make sure the most important ones were answered.
- I always end meetings with a list of concrete steps. I comb through meeting notes and look at action items. Often, this list is just for myself, so I know what to work on next. Sometimes, I like to double-check this list with my advisor to get help prioritizing items. In collaborative projects, this list of concrete steps becomes a plan for everyone that is then communicated to the group.
- After meetings to discuss project ideas or progress, I usually spend ~15 - 30 minutes afterwards organizing my meeting notes. Typically, I take notes within meetings, but these are often scattered, incomplete thoughts that come up as we discuss ideas. After a meeting, I will try to organize these thoughts into more coherent notes that I can refer back to, and flag anything that I need to discuss more for future meetings. For this reason, I actually don’t love scheduling back-to-back meetings even though it can seem more efficient.
- This list makes it sound like I enforce rigid meetings where every second must be spent doing something productive to check a box off a list. But actually, it’s the opposite. I find that this level of preparation and respect for other peoples’ time actually helps create opportunities for the types of free-ranging scientific discussions (and gossip) that are most fun and most fruitful. By preparing this much, there is more time and energy left to think about broader ideas or more distant, big-picture plans.
Learning
- One of the joys of being a student is that it is your job to learn. You should take advantage of opportunities for learning by going to talks, engaging with visiting speakers, talking with your labmates, etc.
- It’s often hard to balance going to things vs. putting your head down and getting work done. When on the fence about going to something, I think it’s generally a good idea to go, but only if you can commit to focusing.
- Don’t bring your computer to talks with the intention of half-listening / half-working.
- You won’t get anything out of the talk, and it’s rude to the speaker.
- You won’t get anything out of the talk, and it’s rude to the speaker.
- At talks, try to find something interesting and relevant to you. There’s always something there if you look / listen for it — an idea, a method, even a presentation technique.
- After talks, talk about the ideas with other people. Talk about the presentation with other people. Most of my learning from talks hasn’t been from the talk itself, but from listening to other people’s opinions about it.
- Asking questions in talks can be scary. I like to think of questions, and then ask them afterwards (either privately to the speaker (e.g., at lunch) or to another student, professor, etc. who may know the answer). Over time, based on these more private responses, you can figure out whether the questions you are thinking of are smart and interesting, or naive and bad (Yes, there are bad questions to ask publicly). Then hopefully you’ll gain confidence in your ability to discriminate good questions to ask during a talk from those questions that are better to save for less public settings.
- That said, ask ‘dumb’ questions in the appropriate setting! (to other students, in individual meetings, even in smaller lab meetings).
- That said, ask ‘dumb’ questions in the appropriate setting! (to other students, in individual meetings, even in smaller lab meetings).
- Being able to converse with other researchers is an important skill. Go to lunch with visiting speakers and practice, and attempt to copy the behaviors of people who you perceive as socially savvy. It will become increasingly comfortable over time, and one day, you’ll find that you have now become the socially competent person whose behavior people are copying.
- Strategically choose projects based on skills you want to learn. You’ll never learn computational modeling / neuroimaging / etc. without doing a specific project that involves those techniques.