Thesis Writing Tips

• The purpose of your abstract is to provide an overview of the entire project. Thus, it should briefly outline the background and justification, approach, results and discussion. A reader should be able to read your abstract and understand what the project as a whole is about.
• Remember that the abstract is the first thing your marker will read and first impressions count, so don’t rush it. Often best to write it once all other parts of your thesis completed.
• The abstract should have the following content:
• 2-3 sentences of general introduction: project background and why it matters.
• 2-3 sentences outlining the overall methodological approach (e.g. we used Planet Labs satellite imagery to classify land cover types) and aims of the project. Often this is started by a phrase like ‘Here we….’
• 3-4 sentences outlining your main findings. These can be a mix of results and interpretation, e.g. ‘We demonstrate that erosion rates were 0.2 mm / ka during the past 10,000 years, we suggest is similar to rates observed in other arid cold-regions”.
• 1 (possibly 2) sentences of broader implications.
  • You don’t reference in your abstract.
  • Abstracts are short, so you need to avoid waffle and be precise in your writing. It is also a great chance to show your examiner you can write well.
  • Look at abstracts of papers within your field and read them in terms of structure, rather than the specific content: you should notice they all follow a similar pattern.

• The aim of the introduction is to provide a brief overview of your project. It should provide the reader with the initial information the need to understand your project and outline the justification for undertaking it. It should lead the reader into your aims and objectives, so that they have enough context to understand the aims / objectives and they understand why they are important.
• The introduction should follow the main points of your literature review. Roughly speaking, each main section in the literature review should become a paragraph in your introduction.
• It should be structured from more general to more specific:
• The first paragraph is usually about broad implications / ideas /background relating to your study (e.g. importance of glacier response to climate change).
• The next 2-3 paragraphs are the key ideas the reader needs to understand your study, noting major research gaps. The content of these paragraphs will depend on your study e.g. it could be framed around key processes, key theories or specific areas.
• The final paragraph usually summarises the literature on your specific study area. Note, that this section is not always needed, depending on the amount of work published on your site.
• The introduction leads into your aims / objectives section, so you need to ensure that your reader knows: 1) the key ideas required to understand your study; 2) the justification for why your study is important and its place in the literature.
  • Due to the relatively short length, you generally only use case studies here that are imperative to understanding your study, e.g. if your work aims to directly compare with another site, or if it builds upon an existing piece of work at your study site/area (e.g. it extends the temporal dataset for the site).
  • You do need to reference in this section.

• Your aim should capture what you want to achieve in your project overall.
  • Generally, it should be one sentence. You can have a sub-aim if you REALLY need to.
  • It should include key information (e.g. time span and location).
  • It usually has an action word (e.g. evaluate).
  • It should specify the thing you are doing (e.g. quantifying glacier retreat).
• Your objectives should set out step by step how you will achieve your aim. Thus, they mark the main waypoints in your project.
• Be specific and concise.
• Each objective should be achievable, realistic, time constrained and measurable. Another helpful way to think about this is to mark your objectives “SMART”, i.e. Specific, Measurable, Achievable, Realistic and, Timely (or time-bound).”
• Ensure that each objective is distinct from the others: each one should be a major step in your project (e.g. collection of one of your primary datasets or a substantial component of your analysis).
• The objectives should be in the form ‘do x to achieve y’ (or vice versa). I.e. say what you want to find out and how, specifically, you will do it. E.g. Quantify glacier retreat rates between 1970 and 2012 at Glacier X, using satellite imagery.
• There are no hard rules about the number of objectives, but 4-5 is typically a sensible number for an MRes thesis.
• Avoid repetition when writing them, for example, continually using the same action word or the exact same turn of phrase.
• Try to avoid generalisms in the objectives, e.g. 'we investigate the characteristics of…' It's better to state what you will actually measure / determine e.g. the elevation, the depth, the area.
• Include timescales and temporal resolution, if relevant.
• Below are example objectives from a recent project on glacier melt ponds:
  • To assess the spatial and temporal patterns of supraglacial pond characteristics (including pond area, number and location) using high resolution (3m) remotely sensed Planet Labs satellite imagery.
  • To identify glacier ice cliff characteristics (including length, orientation, number and location) and investigate the interaction between ice cliffs and ponds.
  • To assess the role of ice surface velocities and surface gradients on the spatial distribution of supraglacial and ice cliffs.
• Some theses may include hypotheses: this depends on the project and is not compulsory. Hypotheses should be specific and testable

• A literature review is a critical review of existing work, that aims to set the context and background for your project, identifies gaps in understanding/areas of disagreement in the research area, and acts as a springboard for your research and thesis. A literature review is not just a list of what has gone before, it must critically evaluate the calibre of existing research and provide a justification for why your research is worth pursuing.
• Critical review, or critical evaluation, involves thinking about, reflecting on, and critiquing (where appropriate) the ideas in the literature. This could include, but is not limited to:
  • Noting limitations, uncertainties and/or gaps in current knowledge and/or key theories.
  • Discussing how theories / ideas have changed over time.
  • Considering exceptions to theories or ideas and how applicable they are.
  • Discussing how a given control or process might vary spatially and/or temporally, and how it might evolve over time.
  • Noting temporal / spatial patterns in a particular factor.
  • Using case studies to demonstrate where theory does / does not fit with reality.
• A good way to start your literature review is to write out your main headings and sub-headings, then add 1-3 bullet points, stating the main contents of each paragraph. You can then insert your reading into this main skeleton and it will really help you to structure your ideas and the literature.
• Structure is really important for your literature review. As a general rule, you should go from more general (e.g. broad research area) to more specific (e.g. specific geographical study are or specific process being investigated). You may also want to go from longer to shorter timescales, new to old theory, or topic based.
• Within each paragraph, you should have:
• Topic sentence: what is the paragraph about?
• Description / explanation: explain any key ideas or theories.
• Examples: back up your explanation with case studies, facts or figures, as appropriate.
• Critical evaluation: How applicable are these ideas? Are there any limitations or exceptions? What are the research gaps? Noting the research gaps and areas of disagreement is important, as it makes it clear to the reader why you are doing this study (i.e. the justification for it) and how your work fits within the literature (i.e. the context).
• Link back to the point: The final sentence should link back into the argument and state the key point, i.e. what does this paragraph tell us that is relevant to your topic?
• Following this paragraph structure will really help the flow of the argument and make it easy for your reader to access information.
• Within paragraphs, work on linking between sentences, rather than simply having a list of things
  • An example of a poorly linked sentence is: Glaciers in Bhutan retreated between 1990 and 2010. Glaciers in Bhutan have lakes.
  • With a small change, the link between the two statements is clear: Glaciers in Bhutan retreated between 1990 and 2010. Many of the retreating glaciers had lakes, which likely accelerated their retreat.
• The reader could put two and two together here, but much better to spell it out and make it easy for them.
• Paragraphs should be separate entities, so they do not link directly back to the previous paragraph. E.g. starting a paragraph by saying ‘this shows that’, when ‘this’ is in the previous paragraph, as it is then not a discrete entity.
• However, you should think about the order of your paragraphs and the flow between them. E.g. you could start a paragraph with the sentence: ‘In contrast to the UK, floods in northern Germany have…’ if you were linking between a paragraph on UK floods and floods in northern Germany. This way, the two paragraphs are linked, helping the flow of the text, but also independent.

Writing style & content

• The literature review needs to be written as ideas NOT in terms of what each author said. I.e. NOT as ‘Author X said’ or ‘Author Y outlines this process’. Writing in terms of ideas will help you to synthesise the information and connect ideas between papers. This is really crucial to having a good literature review.
• Ensure that you fully explain key processes and ideas. Sometimes, people note a process (e.g. that lakes impact ice loss) but don’t actually explain HOW this works. Your thesis needs to be self-contained, so has to explain everything it introduces. Do not assume that your reader (or examiner) has the same background knowledge that you do. They may not be specialists in the area of research.
• Make sure you define key terms and all acronyms. Imagine you are writing this for a university friend who studies a different subject and think about which terms you would need to explain to them.
• Ensure you define more general terms, that are specific to your project, e.g. long-term versus short-term. What you think of as long / short in your study could mean something very different to other people. Same for spatial scales.
• Be precise in your use of terms. For example, glacier retreat, glacier area change and glacier mass loss are NOT the same things (although clearly linked).
• Avoid including lots of material that is very, very general. For example, if you are doing a thesis on mountain glacier change, you do not need several pages to explain how mountain glaciers form, how they flow downhill etc. The key idea is that the material should be directly relevant to the project. If it’s just generally related, you don’t need it.
• Clearly note research gaps and broader importance as you go through. For example, ‘The processes driving Himalayan ice loss are poorly understood, but it is crucial to improve our understanding, as the supply water to almost 1 billion people in downstream catchments’. It usually works well to note significance / gaps in the relevant sections / paragraphs. You may also want to highlight these research gaps in the introduction and /or if you have summaries within sections or at the end.
• You don’t need to have summaries in the literature review, but sometimes it does help to remind the reader what the key points were: it depends on the topic and structure. If you do choose to include them, you need to make sure you are not just repeating material and the summaries are actually helpful. They should come at the end of the section, not the start.
• Make use of examples to illustrate your points. These help to break up the text and show how key ideas relate to reality.
• When adding case studies / examples, make it clear how these relate to your argument and try to put them in context. E.g. is your example particularly representative or is it an exception? Is it a similar example of what you are planning to do, just from another area, or a very different site? Your reader could probably put two and two together and figure out why you’ve included that case study, but it’s much better to spell it out for them.
• Be consistent in the tense used. The vast majority of the time, you should use past tense, as the studies and the data collected are in the past. You may occasionally need to use the future tense (e.g. for prediction) or present tense (e.g. for describing what the study site is), but you should only do this where appropriate. Otherwise, you need to be consistent in your use of tenses:

• Use figures and / or tables to illustrate your literature review where appropriate: this helps to break up the text and can save you a lot of words trying to explain complex ideas / processes.
• Tables can be useful if you need to summarise a lot of previous work. They can really help to identify similarities / differences between approaches or studies. This can be a lot more efficient than simply listing each study in the text.
• You do not need to create your own figures / tables (although you can if you want) and you can use ones from the literature. However, you do need to state the source and say if you have modified them. This should be in the figure caption, usually at the end (source: Jones et al., 2010). The reference then goes in the reference list, as normal.
• Refer to figures indirectly (Figure 1) and write in terms of ideas, not what the figures show, i.e. don’t say ‘Figure 1 shows that……’.
• When referring to figures in text, you just need (Figure 1), not (Figure 1, demonstrating / showing, illustrating). The extra words are unnecessary.
• You need to cite references for all ideas that are not your own.
• Where possible, try to use up-to-date references, to show you understand the literature and are reading the newest material. A good example is the IPCC report: best to cite the most recent one, rather than an older one, which could be outdated.
• However, you should also ensure you cite older, but important (i.e. foundational/seminar) literature, and that you go back to the original paper, rather than citing someone who has cited the paper. This shows you have read the literature in depth and understand where these ideas come from. Also, it’s important to use the original source, to avoid the academic version of Chinese whispers and potentially changing the original message. For example, a quote appeared in the 2007 IPCC report stating the Himalayan glaciers would entirely disappear by 2035. This had been taken from a report by the WWF, who took is from an interview with a glaciologist for two popular science magazines. The glaciologist in question said they had never published the work as it was speculative! This illustrates how easy it is to get the message wrong by not going back to the original source, and how dangerous it can be.
• If many papers have said a certain thing, pick the 3-4 most important and use e.g. at the start, to show you know there are more. This is better than having 10s of references in the same bracket.

Study site & representativeness

• A methods section is essentially a recipe for your research. When writing your methods section you want to ensure that a researcher 10 years (or more) from now could pick up your thesis and reproduce the investigations/analysis simply by following the instructions in your methods section. Reproducibility is important for high quality science.
• Considering having a study site section at the start of the methods section, if appropriate for your project. Some theses have distinct study site chapters.
• Study site should DESCRIBE you study site only, e.g. how big is it, what is its elevation, what is its climate. If you’re reviewing the literature at your site (i.e. synthesising the results of previous studies at your site) this should come at the end of the literature review.
• When describing your site, be precise and give salient detail. E.g. the study area extends from 5,600 m a.s.l. to 38,00 m a.s.l.
• Note why you chose your study site(s). This applies to both your field site as a whole and any sampling sites within it. It is fine if it was chosen because of access / data availability, but this should be made clear.
• Give an indication of the representativeness of your site(s) so it’s clear whether your results can be transferred elsewhere. It’s also fine if you’ve chosen a site that’s the exception to the rule, but its position in the field needs to be clear.
• If appropriate, justify the spatial scope of your study and/or why you chose certain sampling locations and strategies. How representative are these sample sites? How might your choice of sampling sites impact your results? As above, practicality often plays a part in site selection and spatial scope, but discussing why you chose it shows that you have thought about it and reflected on it, thus showing critical evaluation.
• A map of your study site is a very good idea. Remember to label all key features and to have an inset map showing where your site is within its country. Consider including a range of GIS information (e.g. DEMs and contour maps) not just satellite or aerial photos. Also look for national or regional datasets of key features (e.g. roads, rivers, towns) that will help to provide content. Remember that not every knows your study site as well as you!
• Particularly with projects involving fieldwork, it may be useful to include photographs of the field site. To add value, annotate them with key features and the thing you want the reader to look at: it may seem very obvious to you, having been there, but it may not be to someone who is unfamiliar with the site.

• Avoid having sub headings that are just ‘primary data’ and ‘secondary data’. This tells your reader very little. Better to have headings focused on the data type (e.g. meteorological data) or data collection method (e.g. Ground penetrating radar data).
• Be really clear what is your data and what was provided by other people. This includes stating what analysis / processing you’ve done on a given data set versus what was done by others. A specific standalone paragraph detailing this can be useful, and is often requested by examiners for PhD theses.
• You may want to include some general background material (e.g. theory/application) on how a given dataset was produced or a given technique works, e.g. the theory and application of how a ground penetrating radar operates. This is fine BUT you need to make it clear how it relates to your work specifically, and not just have a general description of the technique. It’s best to do this as you go, rather than describing the technique / data collection method extensively and only linking to your project later. Only include more general material like this if it is directly relevant to your project.
• Where appropriate, reference previous studies. This can be referencing specific methods publications (e.g. British Standards for soil and sediment analysis) or providing justification for your sampling procedure. For example, previous studies might suggest that hourly sampling is enough to capture variations in water chemistry.
• Use figures to illustrate, if appropriate. Photographs of field equipment set up can be important, and can aid the goal of enabling reproducibility, particularly if you have an unusual field set up.
• Be precise in your descriptions of what you did. E.g. rather than just saying air temperatures were measured, state the sensor used, the timescale, sampling interval and sensor resolution.
• Give salient detail for all of your datasets. For satellite imagery, this includes things like image resolution, bands used, data source and image frequency. For instruments, this would include things like the instrument make and model, its resolution and your sampling frequency. If you have a substantial number of datasets, a table is a good way to present this information. Some of this information can be included in the appendices if unwieldly in the methods chapter.
• Consider what the (temporal / spatial) resolution, frequency, sampling strategy and/ or data gaps. For example, if you are looking at glacier fluctuations, you need to get images from roughly the same date each year, to minimise seasonality. This is not always possible, so you have to consider the potential impacts and how much this ‘matters’ for your results.
• If you have to make choices about using certain parameters over others, show that you have tested them and provide justification for your choice. For example, if you were automatically classifying glacial lakes, there are several different techniques and thresholds you could use. You want to show you’ve tested them (e.g. by manually comparing to visual imagery) and that you’ve looked at the options in the literature, so you know you’ve used the most appropriate technique. Often, there’s not one ‘right’ answer, but instead it’s about showing that your approach does a good job.
• If you have had to compromise data acquisition in the field (e.g. due to bad weather or equipment failure) be open and honest about in your methods, even if it means that your data are not perfect as a result. It shows you have put in the effort to collect the data and also reflected on its quality.
• Describe any statistical analysis you conducted on your data. Note that you should avoid using words like ‘correlation’ and ‘significant’ in your thesis unless you have actually tested this statistically.

• Uncertainties and errors provide us with information how reliable our results. Importantly, they tell us what is / is not real change, and therefore strongly inform our interpretations. For example if we are trying to assess how water pollution levels have changed over time, we need to know the errors in our measurements, to ensure any change we detect are real and are not just the results of things like measurement error.
• State clearly how you quantified your errors.
• For data you collected, you will need to calculate this and you should look at similar papers to see how they have done it. If you cannot calculate it from your data, the next best option is to reference a paper that collects very similar data and state the errors from that. However, you need to make sure these errors are applicable to your study and datasets.
• For data someone else produced, you should be able to get error values. For example, a number of satellite products come with error maps. Think about how you present these and what is most relevant for your study: you could present a map of errors for your entire study area, but you may also want to calculate them at any sampling points where you extract data.
• You can then feed your error calculations into your results figures, e.g. by adding error bars.
• Whatever you do, you have to be sure any patterns you see in your data are real and not just because of errors in your data.

Getting started with results

• Your results chapter is a really important part of your thesis, as it forms the basis for your discussion and presents the content you have produced. Thus, it’s a very good idea to get the results right before you go on to the discussion. You may also need to revisit your methods / literature review once you’ve written the results and discussion.
• Your results section should provide a description of your observations, highlighting and describing the important patterns, trends and anomalies in your data. It is distinct from interpretation and discussion of your observations, which should be left to the discussion section. This is important because knowledge and understanding may change over time, so that someone might interpret your results in a different way if they revisited them in the future, with this new knowledge (e.g. if there had been a paradigm shift like deformable beds under glaciers or plate tectonics).
• A good way to get started with your results is to produce figures that display your data. These don’t have to be perfect and you don’t necessarily have to think about synthesising them at this point, just produce the figures. This way, you can get an idea of what is going on and look for patterns / commonalities / trends / anomalies. This is a very good time to meet with your mentor and you can use these figures as the basis for discussion.
• Once you’ve made your initial figures, try to put them in some kind of order. The best way to do this is to add in sub-headings and then put the figures into that structure.
• Look at the figures (with your mentor) and think about what your data show.
• Write 1-3 bullet points stating the key points in each section (note this should be per section NOT per figure: as outlined below, you need to write in terms of ideas not individual figures). You can then think about the key point for each paragraph and begin to flesh out the text around these key points.
• Once you know the key messages, you can also think about which figures you really need, which you can remove and which could be synthesised / turned into summary figures.
• Taking this ‘skeleton’ approach is much easier than simply trying to write from start to finish. It also means you don’t inadvertently miss anything by synthesising the figures too soon and also means you can ensure that your final figures really show the key points.
• You need to pull out the key ideas / arguments and avoid just describing one figure after another, as this gets very repetitive and it becomes hard to see the key messages for all of the detail. It also helps you to group similar ideas together, as you group figures showing similar things. In this way, you are ordering and synthesising your data, rather than simply reporting it in the order you collected it.
• You can then use specific values from your results to illustrate your points. Try to be specific wherever possible. For example, don’t just say that a profile on a graph goes up, state clearly how much it goes up by.
• Additional data and/or extra figures (e.g. if you have graphs of sediment characteristics for +50 sites, and only want to include the summary data in the thesis) can go into appendices.

Structure

• Think about the overall structure for your results. This could be:
• Thematic e.g. you present your flood reconstructions first, then your models of where would be inundated in different size of modelled floods.
• Order of importance / complexity, e.g. you may need to discuss where certain features are found before you can say how they change
• Spatial / temporal patterns e.g. you may want to discuss spatial patterns first, then say how these vary over time.
• Spatial / temporal scale e.g. discuss changes across the whole study area, before focusing on sub-sections.
• Consider all of these principles when structuring your results and think about what makes most sense for your data.
• To aid the structure, ensure you have sub-headings. These should match the content of the section and be descriptive (e.g. Seasonal patters in water chemistry). Avoid using ‘primary’ and ‘secondary data’ as headings, as these don’t tell the reader much about the content and also mean that you may split data you want to directly compare, which makes the structure awkward.
• There’s no hard and fast rule about numbers of sub headings but generally, you shouldn’t have one sub-heading per paragraph and many pages without a sub heading are hard to read. Think of the sub-headings as rest points for your reader during a hike: if you take too many breaks, you don’t get anywhere, but too few and you get tired.

Content and writing style

• The results section should only describe your data, not analyse or interpret it: someone else should be able to pick those results up and come to their own conclusions.
• Write in terms of ideas, not what each figure in turn shows, i.e. ‘The moraines become progressively older with distance from the ice front (Figure X)’, rather than ‘Figure X shows that the moraines are progressively older with distance’.
• Following on from this, avoid describing the figure itself, e.g. ‘The red lines show the oldest moraines’. This kind of information on what a given symbol means in a figure goes in the caption. The results text should describe the patterns in the data, not the figure itself.
• Reference figures indirectly, i.e. (Figure 1) and figures should be placed after where they are first referenced in the text.
• Think about the narrative: as per the literature review, each paragraph should have a topic sentence and a summary sentence that links back to the project and states the key point. Clearly stating the key message at the end of each paragraph will really help the narrative and argument.
• Signpost your reader and remind them why you are looking at a particular thing. E.g. “In order to assess the impact of ice cliffs on pond melt rates, we compared changes in pond volume for ponds with and without cliffs. Overall, ponds with cliffs expanded 20% more than those without during our study period.”
• This kind of signposting reminds your reader of the linkages and why you’re discussing certain things.
• Most of the time, you don’t need chapter summaries at the start of each chapter: if it’s well-structured and has appropriate sub-headings, these are redundant.
• When writing your results, try to be precise and clear about what you are discussing. This includes:
• Stating dates when things happened or date ranges if discussing change over time. E.g. ‘We estimate the Little Ice Age began in the area between 1890 and 1900, and persisted until ~ 1930’.
• Being clear what exactly you are discussing: your entire study area? One part of the study area? One sample point?
• When comparing things, state clearly what you are comparing. E.g. ‘Flow speeds were 15% (25 km/hr) faster during Lahar event 2, compared with Lahar 1.
• If you say a thing has ‘changed’ specify what you mean by changes: increase in area? Differences in a specific parameter? An increase in the mean?
• Be clear whether you are discussing trends (i.e. how a variable has changed over a period of time) versus change between two time intervals (i.e. how a given variable is different in Time Y, compared to Time X).
• Make any categorises / groupings clear and explain why you’ve grouped them. E.g. ‘The deposits over 100 m2 (Deposits 1, 3 and 5; Figure X) had larger and more angular grains than the smaller deposits (Deposits 2 & 4) which were under 100 m2.’
• Try to keep the language simple and straightforward (see Section 9.1). Avoid phrases like ‘it was found to be’, ‘it was observed to be’. Just state what the results show and remove the extra words. Write scientifically. Look very carefully at every sentence, and look for ways in which it can be broken up into shorter sentences. As a rule of thumb, a sentence that exceeds 2-3 lines is too long.
• There are a number of useful guides on writing scientifically, for example: www2.le.ac.uk/offices/ld/resources/writing/writing-resources/science

Level of detail & using your results

• Your results need to have a balance between detail and maintaining a clear narrative.
• Remember that your job is to convey a clear message to the reader NOT to write down every single detail in your data.
• Often, when people first write their results, they describe every little feature in the data and the results can become repetitive. This means that the key message is lost: you can’t see the wood (the key points) for the trees (lots and lots of data).
• Instead, you want to look at your figures (with input from your mentor) and decide what your key points are, both within each section and within each paragraph. You then use the data to illustrate and demonstrate these points. This way, you synthesise the data and tell a clear story. For example:
• The following has lots of detail, but it’s hard to read and it’s unclear what point the writer is trying to make:
• The temperature of Pond 1 increased by 3.5 °C between 5.00 am and 10.00, then increased by 1.4 °C from 10.00 to 14.00. After this, the temperatures were around 6 °C and fluctuated by around 2 °C, then decreased to 1°C by 19.00.
• It’s better to synthesise this information, say what the key point is, and illustrate:
• The temperature in Pond 1 warmed during the morning, until 14.00, fluctuated during the day (by ~ 2 °C) and then cooled by evening (19.00).
• When using data to illustrate, think about which values best illustrate your point. Sometimes less is more!
• Also think about the clearest way to present your data values. Sometimes brackets can be really helpful for clarity and clearly linking values with the things they are referring to. E.g. “The level of pollutants increased in all sites over time, with the greatest changes being at Site 1 (+60%), followed by Site 3 (+49%) and Site 2 +(43%).”
• When stating values, think about what is the most appropriate way to state them, eg you might want to use a single value or give the range of values. Also consider whether to give percentages as well as absolute values: particularly when discussing changes, percentages can put things in context.
• Remember to give units for all values given and be consistent in the units used (e.g. kilometres or meters). Use metric as standard.
• Stick to what your data show and avoid extrapolating.

Errors and statistics

• Only use words like ‘correlate’ and ‘significant’ if you’ve done statistical tests.
• Think about where you can add stats in and what they tell you. For example, if you talk about two sets of data being ‘different’, can you test for statistical difference? This will help to bolster your argument and strengthen your work.
• Check your statistical tests are appropriate (e.g. does a test require normally distributed data?).
• Clearly state what the outputs mean, e.g. ‘t-test results show that sediment size was significantly larger at Site A than Site B (p value = >0.05)’.
• Give p-values when quoting stats. These tell you whether your results are significant or not. For example, R2 values tell you how well your line of best fit describes your data, not the significance, so you should quote R2 and p-value if fitting trend lines.
• Think about what is ‘real change’ (i.e. change above your error values) when presenting your results. Do this in your figures (e.g. by adding error bars or greying out areas above the error threshold) and in the text.
• You can only include patterns / data values that are ‘real’ data. This is very important, as not separating values within / outside of your error margins is misleading.
• When discussing / presenting data think about an appropriate number of decimal places: how accurately can you really measure a given variable? Giving more decimal places is mis-leading, as it implies you can measure it to that accuracy.

Figure should include error bars where possible.

• A discussion section provides an interpretation and critical assessment of your results, explaining why your results show what they do (e.g. the processes behind them), a discussion of the limitations of your particular study, ideas for future work beyond the scope of your current project, and an explanation of the broader implications of your findings. This discussion should be undertaken whilst directly and critically engaging with, and referencing, relevant literature. It should contain few, if any, figures.
• Your aim in the discussion is to explain your results. Thus, you need to link what you found to the literature and attempt to explain it. You are NOT aiming to simply restate / summarise your results.
• Every point you make in the discussion should be supported by a reference to either the literature OR your results, either by referencing a figure and/or stating a specific result.
• You should refer to your figures wherever you are discussing the results they show, not just once in the paragraph.
• Depending on your thesis, it can be helpful to have your discussion in a similar order to your results section, so that the reader can link the two. However, in other theses, it may be more helpful to structure the discussion e.g. by different theories or different parts of the study site. In all cases, make the links to the results clear by giving specific values, using phrases like ‘Our results show that….’ and referencing figures wherever the data are discussed.
• As for the other sections, paragraphs should have a topic sentence, descriptive material (i.e. explanation of your results), examples (where appropriate), critical evaluation and then link back to the key point.
• Each paragraph should have one key point, which needs to be clearly articulated. A good way to start your discussion is by writing out your headings and then the key point for each paragraph. You can then flesh each one out, without losing the overall narrative.
• Throughout, try to critically evaluate your results. This can include (but is not limited to):
• Spatial patterns / variations: How do your results vary across your study area or between your study sites?
• Temporal patterns / variations: How do your results vary over different timescales? Do the patterns / relationships change at different scales?
• Relationships between variables: How do different factors link together, e.g. how do air temperatures correspond to ice loss?
• Theory: How do your results compare to key theories? If they don’t fit, then why not?
• Comparison to other sites: How do your data compare to similar studies at different sites? How about different time periods?
• Exceptions / unusual patterns: Do all your sites / data points follow the same patterns? Are there exceptions? Why are these different? Importantly, are these actually different, or is this to do with your errors / data collection techniques?
• Avoid having lots and lots of results, as this clouds the argument. Instead, try to make a key point and then show clearly how your results support it (e.g. ‘Climate has warmed dramatically since the 1970s in our study area, which has coincided with our observed 27% increase in non-native plant species) You could also frame this the other way round and state your results, then give the explanation (E.g. The number of non-native plants specifies in our study area increased by 27% since the 1970s, which we link to regional climate warming).
• As in the results chapter, always try to be specific and precise when giving examples or values, e.g. state the time period, location, value.
• Try to give specific values in the same units, so they are directly comparable. E.g. if you are comparing flood magnitudes with a previous event in the literature, with your own, ensure they are in directly comparable units.
• Where appropriate, include case studies or examples from different areas / times, to illustrate your argument. Make it clear what the comparison shows: are the two sites different? Or the same?
• If you are unsure of the explanation for a given thing, e.g. a specific result or a pattern in your results, then it is fine to speculate, within the confines of the literature. This shows you have thought it through. One way to do this is to note what is unusual and why it’s unusual, e.g. by comparing it to other sites or theory. Then run through the potential explanations and tick off those you can show aren’t possible or are unlikely based on evidence. Also note those that are supported by your evidence. This often allows you to narrow down the explanation and you can then note what data would be need to confirm it.
• With the above in mind, it is important that you make it very clear what is speculation and what we can say, based on the data. It is not a good idea to take things too far and speculate wildly, without any constraint in the literature / your results.
• Likewise, it is important to make clear in your writing what is shown by your results and what comes from theory / previous studies. E.g. Our results demonstrate that Glacier X has retreated (Figure 1), which fits with previous measurements of retreat at the site between 1990 and 2000 (Jones et al., 2010).
• Often it is useful to have a limitations/ future work section at the end of the discussion, although sometimes it works best to note these as you go through. Whilst it’s good to reflect on limitations, try to avoid shooting yourself in the foot by dwelling too much on issues. Equally, it’s not a great idea to start each section by listing what is wrong with your data, so better to save it until the end.
• You can bring new figures into the discussion BUT these should not display data that hasn’t been presented in the results. Examples of possible figures include: a conceptual diagram; annotated photos to illustrate a key process; or a comparison figure. You can also use figures from other papers, so long as these are referenced appropriately.
• Figures which draw together your findings and integrate them with existing conceptual or geographical frameworks (e.g. how your study site fits within the catchment-scale perspective) work very effectively in the discussion section, and can help you, and your reader, identify key concepts and their broader importance.

• A conclusions section is there to distil and summarise the key findings of your theses. It should only report material that has been described/interpreted earlier in your thesis. No new material should be introduced at this stage.
• Remember that your conclusions are the last thing your marker reads before deciding on your mark, so they are worth thinking about.
• As a general guide, each paragraph in your conclusion should relate to one main section in your results / discussion. This may vary if certain sections are longer than others and you need two paragraphs to cover one part.
• Within each paragraph, clearly state your key findings and use key results to illustrate this. E.g. ‘Glaciers retreated in the study region by x ma-1 between 1990 and 2019’. Avoid adding too much detail, as it will obscure the key message, so just use the main, headline results.
• Ensure that your signpost appropriately what the key message is in each paragraph and how your data relate to / contribute to this key point. E.g. “Our data showed that…”, “We attributed this to…”, “This is a key finding because…”.
• As well as following the principle of having a topic sentence and a summary sentence at the start and end of each paragraph respectively, and one effective structure for each paragraph is:
  • Summarise results/observation
  • Explain the process behind the results/observation
  • Statement of broader importance/implications
• At the end, it is useful to have a paragraph that ties the thesis together. This could include:
• Overarching findings
• Recommendations for future work
• Broader significance (why do these results matter?).

Structure

• Structure is absolutely key to writing a good thesis.
• Think about your overall structure (broad to specific) and your structure within paragraphs.
• Start the paragraph with a topic sentence that clearly states what the paragraph is about.
• Next comes you descriptive / explanatory material. Illustrate this with relevant examples e.g. your data of case studies, depending on the section.
• Add critical evaluation if appropriate (in the discussion and literature review).
• Sentence at the end to link back to your project: what does this paragraph tell us and why does it matter?
• Each sentence should have one main idea.
• Each paragraph should cover one main topic and be independent from previous paragraphs. Don’t refer directly to the previous paragraph: if you need to, the information should be in the same paragraph.
• Think about appropriate subheadings. These should reflect the content of each section.
• Generally speaking, if you only have one paragraph, it may not need a new sub heading. However, this is OK for certain parts of the thesis, e.g. if you have a paragraph for your study site. What you want to avoid is a sub-heading for every paragraph.
• Sometimes, adding in subheadings for each paragraph can help with the structure – you can think of this as the scaffolding, which gets remove towards the end of the job. Just make sure the sub headings you remove are reflected in the topic sentences of you paragraph.
• Also avoid having large sections with no sub-headings. It makes it hard for the reader to follow.
• Number your subheadings: it really helps the reader to understand the structure.

Writing style

• Write for your reader. You want to make points clear and to make it as easy as possible for the marker to give you marks. Don’t leave it for them to put things together and try to guess your main point.
• Write in terms of ideas NOT papers. Avoid saying ‘Author X’ or ‘Paper X said…' Instead, you want to synthesise the ideas and write in terms of those ideas. The only time papers should be directly referenced in this way is if they are really central to your work, e.g. a really key theory or if your project compares results directly to another case study. As such, this should only ever be a maximum of 1-2 references in your thesis.
• Describe what is said in papers in your thesis. It is not acceptable to just reference it and say it is explained in another paper. E.g. ‘We follow the methods of Author X (2019)’ has to be followed by an explanation of what the method actually is, not just state you followed their method. Your thesis should be a standalone document and shouldn’t require the reader to go and look things up in papers.
• Avoid adding words where you don’t need them. E.g. ’It has been observed that flood frequency has increased’ is unnecessarily wordy. Just state it directly: ‘flood frequency has increased’.
• Make sure you stick to the same tense throughout, unless you actually need to change tense. Often people discuss some of their results in past tense (which is usually correct, as the things they measured / observed happened in the past) but switch to present tense part way through.
• If you are doing something that involves prediction, then you may need to switch tense (e.g. glaciers will melt by 2050) and occasionally you may need the present tense, but be consistent.
• Be precise in your writing and be clear what you are referring to. A common issue is using ‘this’ too much, e.g. ‘this shows that…’. If you do this too much, the reader loses track of what ‘this’ is. It’s OK occasionally, but not too much. You shouldn’t refer to ‘this’ if the ‘this’ is in a previous paragraph: new paragraph, new topic.
• When discussing time frames or dates, it is best to give the actually time frame (e.g. 1970-2010 or the 2010s) rather than saying a number of years before present (e.g. over the past 20 years). The second format will not age as well as the first: if someone reads your thesis in 10 years, 20 years ago has a different meaning.
• Avoid overly long or overly short paragraphs. Same for sentences. Otherwise, your writing becomes hard to read. Each paragraph should fully explain one idea and be on one topic.
• Within sentences, avoid splitting clauses. If you do this, the reader has to remember what you were talking about, so it makes it hard to follow. E.g.
• This sentence splits the clause:
• The glaciers in Bhutan, which are located at high altitudes and receive their snowfall in the summer, are losing mass rapidly.
• A more direct, clearer way to write it:
• The glaciers in Bhutan are losing mass rapidly, are located at high altitudes and receive their snowfall in the summer.
• OR:
• The glaciers in Bhutan are located at high altitudes and receive their snowfall in the summer. They are losing mass rapidly.
• Use conjunctions (e.g. and, or, but) OR have separate sentences. Don’t just join to halves of sentences together with a comma.
• Define acronyms in full the first time you use them. Due to the thesis length, it can be useful to redefine specialist acronyms at the start of a major section, e.g. if you introduce the acronym in the literature review, you may want to remind the reader in the discussion.
• Avoid shortenings (don’t, can’t) and any slang / colloquialisms.
• When talking about numbers of things, write the number in full when there are less than 10.

Figures

• The primary aim of any figure should be to clearly convey its message to the reader, whether that is presenting your results, illustrating a method or summarising a concept.
• Figures should come after the point they are first referenced in the text.
• Generally, you should reference figures indirectly i.e. (Figure 1), rather than directly ‘Figure 1 shows…’. As with referencing papers, you should write in terms of ideas, not what each individual figure shows. If you write by figure, it’s much harder to synthesis information.
• Always make sure your figure is clearly legible (even for short-sighted academics!) and that you can read all of the text (e.g. axes labels, legends). If your reader can’t read it, they can’t give you marks!
• Make sure your figure is large enough to be clearly read: you can make full page figures and you can make them landscape. If you really need to, you could do an A3 pull out (e.g. for a really detailed geomorphological map).
• Avoid having page after page of figures. This makes it hard to follow the thesis narrative and you should instead think about how you can synthesise this information. For example, you could have a summary figure, then each individual one in the appendix.
• Think about the key message you want to covey from your figures and the most efficient way of doing this: more figures does not necessarily equal more marks!
• Think about colours. What do certain colours mean to people (e.g. red is usually warm temperatures, blue is cold)? Which colours show up well against each other? What if your reader is colour blind? Some useful online resources for this include:

www.colourblindawareness.org/colour-blindness/types-of-colour-blindness/

betterfigures.org/2015/06/23/picking-a-colour-scale-for-scientific-graphics/

venngage.com/blog/color-blind-friendly-palette/ 

• Add appropriate legend entries: usually all of the main features in the figure should be in the legend. Where the data are raster data, a classified symbology is usually better than stretched, as it always people to read off values more easily.
• Add appropriate captions. These should describe what the figure shows (e.g. location map of the study site, showing sampling pits (red dots) and meteorological stations (green triangles)). The caption should not analyse or present your data in any way, it just gives the reader the information they need to understand your figure. It should also state the source, if needed e.g. the source, date and type of background imagery, and/ or if the figure is from another paper (including noting if you have modified it).
• Make sure that your maps have a north arrow and a distance scale, and are correctly georeferenced.

References

• If more than one paper has contributed to an idea, you should reference both.
• If many papers have said it, pick the 3-4 most important and add e.g. to show you know there are more.
• With your references, you want to show that you know the literature in the field, so make sure references are as up-to-date as possible (e.g. use the newest IPCC report), include recent papers, but also don’t miss key older seminal papers. You should get a feeling for the important papers in your field from your reading: key ones will come up repeatedly in other people’s papers.
• In terms of the exact formatting (e.g. making journal names italic), make sure you are consistent. Bibliographic software will really help with this.