How to write a great dissertation

Every summer students arrive in OpWall field sites to collect data for their undergraduate or MSc research project. As a field scientist at the Madagascar site, I have seen many of these students in the forest. As a lecturer, I have supervised many student dissertations at Oxford Brookes University. I think this allows me to make some suggestions about how you can go beyond writing something that is merely ok, and work towards producing an excellent dissertation instead.

It helps to select a good topic that can be addressed using data you can feasibly collect in a short field season. If you selected a project idea form the OpWall brochure you should be fine on this score. Good topics will usually sit within a theoretical framework e.g. resource selection by animals, which will allow you to formulate hypotheses informed by ecological theory about what you expect to observe in the field.

Photo by Callum Evans

Good dissertations have a clear structure which is a bit like an hourglass: wide at the top and bottom, with a narrow neck in the middle. The introduction should begin very broad. You might mention climate change, the global biodiversity crisis, land cover change, human population growth and growth in consumption footprints. You would then narrow the focus to a particular ecosystem such as tropical dry forests, the country where your study site is located e.g. Madagascar, and the taxonomic group you are working on e.g. lemurs. Next, you need to introduce the theoretical lens which frames your study. For example you might introduce landscape ecology and discuss the key papers in this area. It is a very good idea to summarise what is already well understood, and identify some gaps in knowledge. Ideally your research project will then address one of these gaps. At this point your introduction can narrow to a focus. You need to explain roughly the approach you will take, for example an observational study walking sample routes observing animals. Finally you can state the objectives of your study, which should be testing a series of hypotheses. For example, an aim might be ‘To test the hypothesis that lemurs are more abundant in forest interior than at forest edges’.

The methods section is part of the narrow neck of the hourglass. You need to explain your study design and be clear about what the units of analysis are. These might be forest plots, sample routes, villages etc. How many are you going to sample? Are the sample units stratified with respect to important environmental variables e.g. forest type, elevation, human disturbance etc. You need to explain your field data collection protocol and how data was handled to prepare for analysis. If you are using secondary datasets these should be cited appropriately. For example published databases of plant or animal traits; or spatial data such as landcover, elevation, climate etc. Next you need to explain your analytical methods such as statistical tests. It is best to keep this as simple as possible. Typically you might do a test on a contingency table to see if the frequencies of observations are in an expected ratio, or regression, or an ANOVA. Make sure to check for assumptions of models. If your response variable is a count or a probability make sure you are fitting an appropriate model e.g. a GLM.

Photo by Alex O’brien (@whatalexseas)

The results section is also part of the narrow part of the hourglass. This should be fairly straightforward to write. It is usually best to begin with a summary of the data collected. How many years of data do you have? How many sample units? How many species? How many individuals? This sort of information helps the reader judge whether the results that follow are based on a high-quality dataset. Next identify the key tables and figures that you need to present to tell the story of what you have found. Undergraduate projects usually have far too many figures. If a Nature paper only needs 2 or 3 figures, does your dissertation really need 20?  You can always put additional material in the appendix. Take particular effort to state your results clearly in the text. Don’t just say something was significant. Explain the direction of the effect e.g. ‘greater abundance of lemurs was associated with greater distances from forest edges’.  Then in brackets provide the evidence from your statistical test.

The discussion is where you broaden out again. You might begin by summarising your key findings. These should be compared to other published studies. If you found that elephants like to be near water bodies or that monkeys like fruit, then what did other published studies find? Are your results in line with what you would expect from ecological theory? Next your discussion should discuss: were there any limitations of your study?  How could this study be extended? What do your results mean for conservation management of the ecosystem you studied?

Photo by Roger Poland

The best projects are those that are well structured and clearly written. Use sub-headings. Break up long paragraphs. Write in short sentences and use simple vocabulary. Too many students write ‘avian’ instead of ‘bird’ because they think it is more scientific. If the maximum word count is 10000 words remember that this is a maximum, not a target. It is much better to write a clear and concise 6000 word project than a tedious 10000 words of waffle. The more similar your project reads to a manuscript that is ready to be submitted to a journal, the better. Take care to make high quality figures with informative figure captions. Pay attention to the detail of in text citations and references.

When you are writing it helps to refer to your university’s marking criteria so you can assess whether what you have written is on course for the grade you want to achieve. For example for >70% you might need to ‘offer analytical comment, critical evaluation and independent discussion’, whereas for >60% you may just need ‘information to be reasonably full and accurate’.

Finally, take care to manage your time as you write your dissertation. It is a very bad idea to open a blank word document and just try to write the whole thing from start to finish in a few days. Professional journalists can write high quality text under huge time pressure, but you probably can’t. It is better to have a draft document, put the headings in and then take lots of notes as you go about reading literature, collecting data and doing analysis. Then it is relatively easy at the end to turn your notes into nice paragraphs of text. You may need copious quantities of coffee, chocolate etc.

Good luck.