- Here be dragons. Statistical dragons
- Writing and talking
- Getting it together
- The End
I’ve been lucky to have been included in three different research groups, with fantastic people in each. My friends have provided a venting mechanism, sounding board, debating forum, vast amounts of good advice, and – most importantly of all – a jolly good time. Get to know people when you start your PhD and keep on making friends throughout, get out there and join in – you never know when you might need them! My experience has been made richer by the people I’ve been surrounded by.
If you’re reading this, well, excellent. But if you’re in the middle of a PhD, perhaps you should stop here and save the rest for your lunch break! Procrastination is an art easily mastered, but apparently never perfected: you just have to keep going back for more practice. Access to high-speed internet is both your best friend (papers! access to distributed computing! downloadable software!) and your worst enemy (long reads on The Guardian! KittenCam! Wimbledon!). It’s good to have a break, and the consequences of getting entirely distracted are less severe at the beginning of a project, but like anything else, time becomes more valuable as it diminishes. To get my PhD finished on time, I had to be very strict with myself, actually signing off Facebook and Twitter for two months so that I wasn’t tempted to procrastinate. There are services you can sign up to that will restrict your access to distracting websites for you, but why bother? I found the threat of not actually being finished on time was enough!
5. Here be dragons. Statistical dragons.
If you’re doing a PhD in any sort of quantitative discipline, chances are that you will come across R at some point. Whispered along the corridors of educational institutions, and enough to prompt feelings of fear and uncertainty in the uninitiated, the 18th letter of the alphabet has become infamous among PhD students all over the world. Actually, it’s not that bad. In fact, it’s really great. R is a stats package and a programming language, which has several advantages over the competition:
- It’s free
- It has an ever-expanding library of functions and help manuals written by users, for users
- It’s infinitely flexible
- You get kudos for using it
- You get to pretend you’re a pirate
There are two disadvantages. Firstly, because R has a command-line interface, you can’t just point-and-click your way through functions, resulting in a steep learning curve. Secondly, when you do turn to the help, or online messaging boards, for assistance, you’ll probably find it rather hostile and obtuse. But don’t be put off – there are lots of great, accessible books on R out there for all disciplines, and so many people use it, there’s bound to be a helpful PhD student or post-doc you can turn to for help.
You might be lucky enough to know precisely what you want to do with your data, once you’ve collected it, but for most of us, choosing the best method for testing a hypothesis can be a tricky problem. This is where your institute’s statistician comes in. I found that the discussions I had with the statistician at my institite were helpful, not condescending, and versed in a way that was simple enough for me to go away and process the information into a decision. Not every institute will have such an approachable statistician, but the maths department might offer consultancy as a service to PhD students, or there’s your fellow students and post-docs. Many minds can often be better than one.
Once you’ve got your head around the statistical wizardry, it can be easy to plough ahead and start creating a host of complicated models. At this stage, I found it helped to return to my original questions. What questions was I trying to answer? Which techniques would be best for answering them? Once I had these ideas settled in my head, I drew a flowchart, starting with the data, stating the question, and finishing with the desired method I would use to answer it (usually an R function). The more specific you can be, the easier it will be to keep track of what you’re doing, and what you’ve already done – if you end up repeating tests without realising you’d already done them a couple of days ago, you certainly won’t be alone! In this respect, it’s a good idea to keep a journal for your stats, in the same way as you’d keep a lab book for your work in the lab.
Drawing plenty of graphs, at each stage of your analysis, will help you to interpret your results in the context of the real world, and discuss what your data are really showing, rather than confusing your audience with coefficients (while necessary, these aren’t the most reader-friendly way of communicating your results).
Remember when you were doing your fieldwork, and you decided to stop because you had enough data to answer the questions you’d set yourself? The same approach can be applied to your statistics. Once you’ve answered your questions using your chosen method, and your happy that your results are robust, don’t go back and try another couple of different methods, just for fun and because you want to flex your new-found statistical muscle. There be dragons! And much wasted time. Stick to the plan, otherwise you could be there forever.
6. Writing and talking
This is important, and worth getting right before you start. Knowing your way around a referencing software package will make your life many orders of magnitude easier when it comes to writing up: a good package will, as a minimum, maintain a database of papers you’ve read, insert references into your documents and compile a bibliography in your chosen style. This is neither the time, nor the place, for an exhaustive review of different packages – among the most well-known are EndNote, Papers, Mendeley and Zotero. I prefer Mendeley personally – it’s free, comes with plenty of cloud storage for pdfs (which you can upgrade for a fee), and is easy to use. Whichever package you decide to adopt, it’s worth learning how to use it properly before you get elbow-deep in writing, then sticking with it until the end, because you won’t want to be working out how to transfer databases between packages with deadlines to meet!
So you’ve spent many backbreaking hours in the field and lab, followed by numerous late nights in front of your favourite statistical package, analysing those hard-won results. Now the time has come to put all that work into words. When I was finally ready for this, time was short, so my supervisor suggested a radical new programme: I would write a chapter in a week. Easy!
Monday: Monday is methods day. The methods are the easiest place to start. Write down what you did, and while you’re at it, write an abstract: what was the worldly problem, what questions did you ask, what were the answers and the implications?
Tuesday: Results text, tables and graphs. Be prepared to cut your graph-tweaking time short – there’ll be time for polishing later.
Wednesday: Back to the introduction, which is composed of five bits: background, knowledge gap, questions, objectives and hypotheses.
Thursday: Four days into the week, you should still be feeling fresh, and a good thing too, for today is discussion day. Put those findings in context! Don’t forget to write a conclusion and discuss the implications of your work.
Friday: Go back to the start and review the whole lot. 5pm is pub o’clock.
Giving a talk before an audience is probably one of the most dreaded aspects of the research experience – I know plenty of postdocs who still get sweaty palms over it, despite years of experience. When I embarked upon my PhD, I knew that this was something I wasn’t terribly comfortable with, so I decided to make the best of it, rather than avoiding the issue. One of the best tips I can give is: be prepared. In reality, this is sometimes difficult, but it’s worth getting your talk ready well in advance and practicing it in front of a substitute audience. This will give you an opportunity to check that you run to time, and take some constructive criticism from colleagues. The less you leave to chance, the less nervous you’re likely to be when the time comes to face your audience for real.
Of course, standing up and talking about your work can be very nerve-wracking. There are three coping mechanisms that tend to work for me (the above isn’t one of them). Firstly, do something you enjoy beforehand, to put your mind at ease, and stop you worrying too much. Secondly, be confident. Or at least appear confident – you may well be quaking inside, but once you get through the first couple of slides with an air of authority, you’ll have your audience’s attention and will have generated some momentum to carry you through the next slides. Thirdly, smile and try to enjoy the experience – you deserve to be there, communicating your work.
7. Getting it together
It’s probably a given that most people will, at some point, fall foul of Microsoft Word’s irritating and often seemingly random formatting system. Since this is the system that most people (aside from those brave souls who have embraced LaTEX – all power to them, too) will use to write up their theses, I’ve taken some time to outline a few tips that might help to save you from text-based strife.
See this chap: ¶ – clicking the Pilcrow symbol on the Word toolbar will toggle formatting marks on and off. Some might find that leaving it switched on makes the screen a bit too crowded, but I find it useful for understanding the mechanics at work inside a document. It’s particularly useful when you’re working with page, section and column breaks, which can all be employed to good effect to reign your text in and keep it where you want it.
Styles in Word can be used to automate the process of setting up a table of contents, which will be much appreciated when you’re at the end of the writing process and just want to get shot of the thing. Headings one to five can be used to create indented headings in the table of contents, which can be automatically updated as you edit your document. This is one of those labour-saving devices that I found really useful.
Inserting images in Word documents is always a bit of an unpredictable affair. I like to try and hang on until I’ve finished writing the main text before I go ahead and insert the images where I’d like them to appear in the final document – if you’re still moving large chunks of text around, you can put money on the likelihood of Word doing something absurd with the text wrapping or anchor placement, leaving your image several pages from where it should be. Once you’re happy with the text, switch on those formatting marks (¶) and create spaces where you’d like your images to sit – usually just one line is enough. With the cursor where you want the image to be, press the insert button, try not to panic about what might have just happened, and set the text wrapping to ‘in-line’ with text. I do things this way for two reasons: firstly, you know where you are, and your image won’t wander the pages unduly; secondly, in order to speed up scrolling through your document once you’ve accumulated several images, you can tell Word to just display the placeholders (Options | Advanced > Show document content) – this option only works with in-line images.
Reference software is jolly useful, but can have one drawback connected with the creation of the table of contents, or tables of figures. Because reference software often uses active fields to link to the reference database – these show up with a grey background when you click on them in your document – they can conflict with the table of contents, which also uses active fields. To avoid any unseemly conflicts, once I’d finished inserting all my references and the bibliography, and was happy with them, I used the option to export the document without active fields to create a ‘clean’ copy – all the references and bibliography are exported as plain text. Most reference packages will have this option, and it’s worth using if you want to avoid potentially confusing conflicts between field codes. Something to bear in mind, however, is that once you’ve exported the document without the field codes, there’s no longer any link to the reference database, so you’ll have to make further changes manually.
At this stage, you’ll probably be quite good at keeping it together emotionally. After all that time spent writing those precious words, it’s vital to be organised and keep your stuff together physically, too. At the very least, keeping your files backed up will save you having to do everything all over again. This is where cloud storage services like Dropbox, Google Drive, SpiderOak, Skydrive, etc., come into their own. As long as you have an internet connection, you can synchronise a folder on your computer at university, continue to work on your documents, then go home, synchronise the folder on your laptop, and work some more. These services aren’t infallible, but they’re much more robust and less easy to loose than portable hard drives and USB sticks. For those with tech-savvy supervisors, you can use shared folders to send your work to supervisors as soon as you’re ready, which avoids those awkward missed emails and filled inboxes.
8. The End
Congratulations – you’ve made it! Three / four years’ work, all summed up in an epic thesis. Now all there is to it is to print the thing, get it bound, and hand it in. At this point, if you haven’t already (in which case, you do like cutting it fine, don’t you!) a thorough read of your institute’s guidelines on thesis submission comes highly recommended. Even if you’re really up against it, keeping a couple of days free to get your thesis printed and bound will avert any last-minute crises. You may well be stressed and sleep-deprived, so getting your friends involved – particularly when out-sourcing your proof-reading – is a good idea!
After all the parties and holidays have dwindled and your life has returned to (an albeit empty-feeling) state of relative normality, it’s not a bad idea to start thinking about your viva. Picking up the thesis and reading it through can be a dispiriting exercise, especially if you had as many spelling mistakes as I did, but it’s absolutely necessary. Pre-viva nerves can be effectively mitigated by knowing the answers to some simple (!) questions:
- What did I set out to find out / achieve?
- How did I structure my project, and why?
- Why did I use those methods? This is a big one: you have to show that you’ve understood the limitations of the methods you used, for data collection and analysis. You’ll need a scientific basis for all the decisions you made – ‘it was what everyone else did’ won’t cut it with your examiners.
- What did I find?
- How do my findings fit in the context of other work?
- What would I do differently?
Like the PhD itself, viva examinations are very personal, and it’s quite likely that someone else will come up with a different set of questions to those above. Asking around can be another good coping strategy for pre-viva nerves.
What I’ve written about here is based on my experiences of a very personal thing – the PhD – and I’m sure that not everyone will agree with the points I’ve made above. But one of the aspects of being a postgrad that I’ve enjoyed the most is being able to share experiences with other students in a similar position. If you have points and comments of your own to add, please do add them below!