How to write a PhD thesis

Professor_Fink

It is the time of year when grad students throughout the world are struggling to get their thesis finished in time to start that postdoc in October, or even to enter the real world, so I thought I would start a thread to gather tips for these poor souls. So here are my thesis writing tips:

1. Use LaTeX
   Latex can easily be learnt in an afternoon and will make your thesis look more professional. It will save you hundreds of hours of correcting formating and will be far more consistent than a word processor. It handles equations and vector graphics with ease and is extremely versatile. Latex also allows comments which do not appear in the document, which are extremely handy for keeping notes to yourself. It is also the prefered format for most preprint servers. If you are planning a career in the physical sciences or maths, then latex is a must.
2. Work out the structure first
   Spending a day or two at the start working out exactly how you want to structure your thesis may seem like procrastination, but will prove incredibly valuable later on. As you near the end of the thesis you will undoubtadly get the urge to restructure things. Putting in a little time at the start can save a lot of time spent on rewrites.
3. Leave introduction and conclusions until last
   Believe it or not, the introduction is the very last thing you should write. It might seem logical to work through in the order the reader will encounter chapters, but this is a big mistake. In the first chapter you will want to describe the structure of the thesis, and outline how you are going to proceed. Even though you may should a plan for this at the start, it will change as space and time constraints begin to take their tole. Believe it or not, you will almost definitely over shoot the word limit and have to axe material.
4. Break each chapter up into managable chunks
   Everyone gets writers block. You come to a chapter and the thought of filling twenty pages with the intricacies of some mathematical technique or experimental setup seem overwhelming. In order to avoid this, it is a really good idea to break down the thesis into smaller and smaller named sections (\section, \subsection and \subsubsection in latex) even before you start writing. Break it down into chunks on the order of 200 words so that they are easily attacked. It makes it easy to work on lots of different sections simultaneously, and means you can write a section in less than half an hour. Even if you don't want your thesis to have such fine graining, add the headers at the start and remove them later.
5. Use at least one chapter to introduce techniques used
   It is often a mistake to introduce a new mathematical or experimental technique in the same chapter you use it. Having a seperate chapter in which you introduce necessary techniques can help you from making chapters overly long, and gives the reader time to absorb this new information. This also provides a very clear cut line between your work and previous work from others upon which you intend to build. This is essential to avoid annoying examiners. Poor students often try to blur this line, and so it is a red flag for anyone reading the thesis.
6. Use a bibliography manager such as JabRef
   It is scarily easy to create duplicate references. I spent almost a week weeding out duplicates and triplicates from my bibliography file. Using a reference manager from the start avoids all this hassle by automatically detecting duplicate entries.
7. Use BibTeX
   Don't write in references by hand. It will be hard to give them a consistent look, hard to keep track of where references what, and hard to update. Use latex, use JabRef (or similar) and use bibtex. You'll avoid all sorts of formating issues.
8. One chapter per paper if you have them
   Every paper you wish to include should have its own chapter. If you have several papers on one topic, then you probably need several chapters on it too. You may need to mix content between these chapters, but one paper one chapter is a good rule of thumb.
9. Neglect earlier work if it isn't to as high a standard as the rest
   Let's face it, that paper we got in our first 6 months isn't really our best work is it? If you have some particularly weak material, don't feel obliged to include it. You absolutely do not have to include everything you did over the course of your PhD, so don't feel obliged to.
10. Keep to less than 80% of the word limit/page count
    It's just a fact of life, examiners don't like long theses. Don't use the full word limit, even as a guide. Keep it short and punchy, rather than long and drawnout.
11. Add a page listing collaborative work
    No man is an island, and that goes double in science. Almost everybody collaborates on projects, and this is certainly not a bad thing. It is however very important for you not to appear to be passing off others work as your own. To avoid this, add a page to the start of the thesis listing any collaborative work, describing honestly your role. This makes it much clearer to the examiners who did what. Again, poor students often try to blur the line, but such bluring raises an immediate red flag and will cause pain in the viva. Avoid it.
12. It is ok two have two themes
    Sometimes our work is in more than one area. Mine quite distinctly focused on two different topics, which could be unified under the broader topic of the thesis title. It can be hard to see how best to mesh two seperate topics together into a readable document. In my experience having a common introduction followed by to distinct parts to the thesis was the answer. It's quite common to do this, and can save you quite a headache.
13. Don't fear THE FEAR
    Are you freaking out because you have been procrastinating and now you have little time left and little progress made? Good. That's called THE FEAR, and pretty much everybody gets it. It will motivate you to do superhuman amounts of work in those last weeks. No matter how tight the deadline, THE FEAR will see you through.

Good luck.

If anyone has any other tips, perhaps they could add them below.

AtomicHorror

*takes notes furiously*

Planning to hand in around Christmas time! :eek:

GuanYin

Good Job

fits

rightio,

Im working on the final formatting of my thesis for hard-bound submission atm. Two of the papers (out of four) have been reviewed and modified since it was submitted and theres shedloads of formatting to do on the rest of it.

Am I stupid to be starting from scratch with a new document as table of contents and headings were not working in the other one? (btw I also did not do the formatting last time so am on a very very steep learning curve).

epictetus2009

i will be doing a standard thesis in 2 years and this is invaluable advice and tool tips.

thanks for posting.

stephenlane80

whats the best Latex program to use on windows, im using TeXnIx Centre with Mitex ?

2Scoops

stephenlane80 wrote: »

whats the best Latex program to use on windows, im using TeXnIx Centre with Mitex ?

I imagine they're all much the some. I use ProTeXt.

Time Magazine

stephenlane80 wrote: »

whats the best Latex program to use on windows, im using TeXnIx Centre with Mitex ?

MiKTeX for the background stuff (that you don't actually see, it does the compilation stuff) and I use WinEDT for actually typing the document.

WinEDT is cool:



It's got a spell-checker; to compile the document all you have to do is click the button at the top; when there's an error it automatically brings you to the line with the error; you can select TeX commands (e.g. bold, or mathematical symboles) from the buttons at the top; you can select columns of text instead of just rows (very handy for pasting in tables of results) and so on. Well worth the €20.

Sparks

I really wish IDEs like that would let you use your own editor. I'm too used to Vim to try to relearn a less optimised editor, even for the nice features.

Mind you, I just use latekmk -pvc -ps on linux to get real-time updates of what I type shown longside the gvim window, and vim has some handy plugins for latex too. And there are generic tex-aware spellcheckers too, IIRC.

Sparks

BTW, most linux distros are using texlive now rather than miktex.
Just FYI - it doesn't change much at the user level.

bangophinx

I use Refworks for my references. Do LaTex programs work with Refworks?

mikhail

bangophinx wrote: »

I use Refworks for my references. Do LaTex programs work with Refworks?

Yep.

bangophinx

mikhail wrote: »

Yep.

Thanks mikhail ... nice link
~1

Gaspode

This may be of interest to anyone starting out, written by an Irish researcher.

Black Swan

There are several pages dedicated to thesis proposal writing, as well as the evaluation of a thesis in this handbook:
ISAAC, S., & Michael, W. B. Handbook in Research and Evaluation. San Diego: EdITS Publishers

Ostrom

I have found the forums at findaphd.com a huge comfort

Well worth a read during the darker hours

Metalpanic

A few people I know found this a very useful reference management program and it's free.

http://www.mendeley.com/

Gaspode

If you are going about a PhD (or any piece of research work for that matter), deciding what statistical tools & techniques you are going to use on each piece of work should happen before you do the work, and not at the end when you're probably going to be in a hurry to write up.

Otherwise you could end up with tons of results that you cant make sense of.

Our statistician has this quote above his desk:
'To consult the statistician after an experiment is finished is often merely to ask him to conduct a post mortem examination. He can perhaps say what the experiment died of'.

So think stats before you start. and if your institution uses a particular stats package get familiar with it from the start, not as you start writing up. That might save you from last minute panic stations trying to navigate software you have no idea about (like one or two of these guys perhaps)

djpbarry

Gaspode wrote: »

So think stats before you start. and if your institution uses a particular stats package get familiar with it from the start, not as you start writing up. That might save you from last minute panic stations trying to navigate software you have no idea about...

Or worse still, it might save you from suddenly realising that your results are statistically insignificant.

itzme

Over the summer in NCG we put together an internal seminar on finishing a PhD, two professors and three recent graduates gave their experiences. We got great feedback from the postgrad students who attended, the recordings of all of these are available at http://ncg.nuim.ie/redir.php?action=events/20110325

kevmy85

Reading this link may help you get over some thesis/viva fears.

http://www.studentservices.uwa.edu.au/__data/page/65297/Its_a_PhD_not_a_Nobel_Prize.pdf

Black Swan

Gaspode wrote: »

So think stats before you start.

Our programme requires us to develop a research proposal that clearly specifies the statistical procedures that we will use. This proposal must be presented and approved by committee before we can begin research (i.e., main study).

djpbarry wrote: »

Or worse still, it might save you from suddenly realising that your results are statistically insignificant.

For deductive, quantitative research, we use one or more pilot studies specified in our research proposal to preliminarily test the research hypotheses (stated in the null) against one or more small samples (rule-of-thumb n=30 for parametric; n=10 or more for nonparametric statistics), which allows us to dry run the research design, statistical methods, and hopefully reduce the risk of insignificance, and such problems as multicollinearity if using regressions, etc., BEFORE we being the main study.

This research proposal and its preliminary results will be summarized in the thesis/dissertation research methods chapter; i.e., it becomes a part of the foundation argument used to justify the design and methods of the main study.

ulyssescohen

These are very useful guidelines Professor Fink- but it seems discipline specific, and is not necessarily appropriate for all disciplines. I think you should provide a clarification of such issues to avoid potential problems

Ulysses

Professor_Fink wrote: »

It is the time of year when grad students throughout the world are struggling to get their thesis finished in time to start that postdoc in October, or even to enter the real world, so I thought I would start a thread to gather tips for these poor souls. So here are my thesis writing tips:

1. Use LaTeX
   Latex can easily be learnt in an afternoon and will make your thesis look more professional. It will save you hundreds of hours of correcting formating and will be far more consistent than a word processor. It handles equations and vector graphics with ease and is extremely versatile. Latex also allows comments which do not appear in the document, which are extremely handy for keeping notes to yourself. It is also the prefered format for most preprint servers. If you are planning a career in the physical sciences or maths, then latex is a must.
2. Work out the structure first
   Spending a day or two at the start working out exactly how you want to structure your thesis may seem like procrastination, but will prove incredibly valuable later on. As you near the end of the thesis you will undoubtadly get the urge to restructure things. Putting in a little time at the start can save a lot of time spent on rewrites.
3. Leave introduction and conclusions until last
   Believe it or not, the introduction is the very last thing you should write. It might seem logical to work through in the order the reader will encounter chapters, but this is a big mistake. In the first chapter you will want to describe the structure of the thesis, and outline how you are going to proceed. Even though you may should a plan for this at the start, it will change as space and time constraints begin to take their tole. Believe it or not, you will almost definitely over shoot the word limit and have to axe material.
4. Break each chapter up into managable chunks
   Everyone gets writers block. You come to a chapter and the thought of filling twenty pages with the intricacies of some mathematical technique or experimental setup seem overwhelming. In order to avoid this, it is a really good idea to break down the thesis into smaller and smaller named sections (\section, \subsection and \subsubsection in latex) even before you start writing. Break it down into chunks on the order of 200 words so that they are easily attacked. It makes it easy to work on lots of different sections simultaneously, and means you can write a section in less than half an hour. Even if you don't want your thesis to have such fine graining, add the headers at the start and remove them later.
5. Use at least one chapter to introduce techniques used
   It is often a mistake to introduce a new mathematical or experimental technique in the same chapter you use it. Having a seperate chapter in which you introduce necessary techniques can help you from making chapters overly long, and gives the reader time to absorb this new information. This also provides a very clear cut line between your work and previous work from others upon which you intend to build. This is essential to avoid annoying examiners. Poor students often try to blur this line, and so it is a red flag for anyone reading the thesis.
6. Use a bibliography manager such as JabRef
   It is scarily easy to create duplicate references. I spent almost a week weeding out duplicates and triplicates from my bibliography file. Using a reference manager from the start avoids all this hassle by automatically detecting duplicate entries.
7. Use BibTeX
   Don't write in references by hand. It will be hard to give them a consistent look, hard to keep track of where references what, and hard to update. Use latex, use JabRef (or similar) and use bibtex. You'll avoid all sorts of formating issues.
8. One chapter per paper if you have them
   Every paper you wish to include should have its own chapter. If you have several papers on one topic, then you probably need several chapters on it too. You may need to mix content between these chapters, but one paper one chapter is a good rule of thumb.
9. Neglect earlier work if it isn't to as high a standard as the rest
   Let's face it, that paper we got in our first 6 months isn't really our best work is it? If you have some particularly weak material, don't feel obliged to include it. You absolutely do not have to include everything you did over the course of your PhD, so don't feel obliged to.
10. Keep to less than 80% of the word limit/page count
    It's just a fact of life, examiners don't like long theses. Don't use the full word limit, even as a guide. Keep it short and punchy, rather than long and drawnout.
11. Add a page listing collaborative work
    No man is an island, and that goes double in science. Almost everybody collaborates on projects, and this is certainly not a bad thing. It is however very important for you not to appear to be passing off others work as your own. To avoid this, add a page to the start of the thesis listing any collaborative work, describing honestly your role. This makes it much clearer to the examiners who did what. Again, poor students often try to blur the line, but such bluring raises an immediate red flag and will cause pain in the viva. Avoid it.
12. It is ok two have two themes
    Sometimes our work is in more than one area. Mine quite distinctly focused on two different topics, which could be unified under the broader topic of the thesis title. It can be hard to see how best to mesh two seperate topics together into a readable document. In my experience having a common introduction followed by to distinct parts to the thesis was the answer. It's quite common to do this, and can save you quite a headache.
13. Don't fear THE FEAR
    Are you freaking out because you have been procrastinating and now you have little time left and little progress made? Good. That's called THE FEAR, and pretty much everybody gets it. It will motivate you to do superhuman amounts of work in those last weeks. No matter how tight the deadline, THE FEAR will see you through.

Good luck.

If anyone has any other tips, perhaps they could add them below.