MAS6041 2023–24
1 THE DISSERTATION PROCESS
1.1 THE DISSERTATION
The dissertation constitutes one-third of the entire MSc programme, i.e. 60 credits and so you should expect to spend between 400 and 600 hours working on it. Doing so will require careful planning to make sure that you use your time effectively and produce the best dissertation you can. The dissertation is also your opportunity to demonstrate how independent you have become as a statistician. While each project comes with a supervisor, they are intended for guidance and support, not to give you a recipe for how to complete your research. The final submission will be very much your piece of work and hopefully something you can look back on with pride and recognise your intellectual contribution. It is up to you to take responsibility for what you produce.
This document is intended to make some points about the process of working on and writing up an MSc dissertation for module MAS6041. Advice relating to supervision, the Dissertation Support Programme (DSP) and Dissertation Support Worker (DSW) will be given in a separate document on Blackboard. The latter two are relevant only for full-time students, part-time students will get all of their dissertation support from the Supervisor.
1.2 THE DISSERTATION PROJECT ALLOCATION
AVAILABLE TOPICS
In Semester 2 Week 1, you will be given a list of 10–12 research topics. These cover broad research areas that can accommodate several research questions through selection and combination of specific data sets, models and methods; each combination constitutes a dissertation project.
Each student will provide a list of 4 topics, ordered by preference by 12:00 Tuesday 27 February 2024. These will be used in the allocation process by the MSc academic staff, who will strive to give each student their first or second choice—although some exceptions may be unavoidable since supervision work must be spread fairly evenly between staff . Allocations will be announced by Week 8.
After their topic has been allocated, each student must look for and select at least one project —i.e. a dataset, model and method combination within the topic. Full- time students will initially meet with their supervisor in groups (determined by the allocated topic) to discuss their selection and begin the process of establishing their unique dissertation project.
STUDENT DISSERTATION PROPOSALS
Most students choose a dissertation topic from the list provided. Students are only likely to propose their own project if their daily job involves them in data analysis or if they have a strong interest in a particular topic derived from previous professional experience. To be considered, the project must have a substantial statistical/probabilistic component of suitable MSc level.
All students considering proposing their own project must submit a project proposal. This is a rather detailed description of the research to be conducted and the relevant personal background, comprising:
• Project Title
• Project description, including a chapter by chapter plan of the contents (350 words)
• Dataset(s) to be used and agreement of continued access to the data until completion of the dissertation (200 words)
• Work plan, including a Gantt chart and a contingency plan in case of change of status (200 words)
• Description of academic/professional background relevant to the project (300 words)
The proposal must be submitted using the template on Blackboard by the end of Week 4 (17:00 1 March 2024). There you will find a file with some examples of previously accepted student-proposed projects—bear in mind these are examples and do not conform to the template.
STUDENTS WITH ALLOCATED PROJECTS
Students who have a project allocated from previous years, but have made little or no progress, may consider switching to this model if they wish. You would need either to submit a list with your preferred topics or to produce a project proposal as described above.
1.3 DEADLINES
Fully understanding the goals of your project and then setting up a consistent and effective work-flow are your first goals and should be completed as soon as possible after the exams, in your first year of study. Alongside this you will need to undertake careful project planning and record keeping. For full-time students, we provide a recommended work schedule as part of the DSP (see separate document), with deadlines for various activities. Full-time students must meet all of the deadlines in the DSP schedule unless a prior arrangement has been made with the DSP lead (Prof. Caitlin Buck).
1.4 THE DISSERTATION DEADLINE
For full-time students the deadline for submitting your dissertation is Friday 20 Septem- ber 2024. Extensions to the time limit for submission of the dissertation are only given for exceptional reasons (i.e. certificatedmedical grounds or substantial extenuating circumstances). For full-time students, non-submission of a dissertation by the September deadline entails failure of the module (but re-submission may be allowed twelve months later). For those doing the course part-time, the year the dissertation is due for submission is determined in the first instance by the end-date of your registration, in some cases this can be changed to accommodate taking the course over a longer time, but this should be done well in advance of the September deadline, by formal request to the Course Director.
1.5 DISSERTATION REQUIREMENTS
DISSERTATION PREPARATION
You are required to produce your dissertation using either the LATEXor Bookdown template provided on Blackboard. You must not change any aspects of the file that will alter the geometry of the pages or the size of the font. Dissertations that do not conform precisely to the templates will, as a minimum, be penalised under Presentation during the examination process and, in extreme cases, may not be examined at all (although resubmission a year later may be permitted).
DISSERTATION LENGTH
Clear writing is usually concise, to the point, and avoids unnecessary repetition. It is thus a requirement that the main matter of your dissertation be not more than 70 pages, including all figures and tables. Those working on theoretical topics, with little need for figures and tables in their dissertation, should aim for considerably fewer than 70 pages (typically 30–50 pages). If you do submit something longer, examiners will read only the first 70 pages (or 50 pages of more theoretical material). Note too that, regardless of length, any material in appendices will only be inspected cursorily by examiners. You should use appendices judiciously, in the way outlined in the next section.
GENERAL ADVICE
A supervisor cannot say in advance that a dissertation will pass; the examiners determine this. However we do expect the following standards from any good dissertation.
Elements It is important that the dissertation shows how the time devoted to the project was spent.
It is fine to include ideas that did not work, or did not deliver what was expected. Research is often about trying new things and not everything always works out as initially hoped.
You should not describe absolutely everything you did. A dissertation is not simply a diary of your work. You will need to be concise and selective in what to include so that it is interesting to the reader. The write-up should be areflection on the analyses done, paying most attention to the parts that are helpful for understanding. That said, you must be absolutely sure to avoid all possible risk of accusations of plagiarism or collusion (see below).
If your dissertation involves developing your own code (ie writing routines/functions than are novel), then they should be included in the main text, either as pseudo-code or the code itself, with appropriate annotations and descriptions. Routine function calls need not be included in the main text (lm, glm, ar, bayesreg, stan, etc), but rather described, detailing in particular the choice of parameters. There is no need to include or describe the use of code for routine summaries (eg plotting, summary tables)
If the main objective of your dissertation is developing code (eg building a new package), then it is advisable to make it available electronically, detailing examples of its use in the main text.
In any case, it would make sense to have examples of the computational implementation in the appendix, eg a worked example with annotated code, for reference.
Writing style. You should write for another student on the MSc who knows nothing of the project, treating sketchily what would be common knowledge to you both, but with care material that would benewordifficult to such a person.
All dissertations should have a proper introduction and include a clear description of the problems that will be considered and the aims and objectives of the study.
The material should be properly ordered and organised. This involves dividing it into suitable chapters and sections, using satisfactory cross-referencing (ideally, including page numbers, provided they are right), and putting in appendices material that obscures the flow or is included only for the record.
Communication should be precise and effective.
All specialised terms and mathematical notation must be defined at first used and then be applied consistently.
Aim for clarity of writing and exposition; try to avoid patterned accounts of similar analyses.
Pay attention to: spelling, the right choice of words, numbering pages, constructing sentences properly, laying pages out well, producing titled and labelled graphs and tables that are intelligible and that add to the discussion, and soon.
Make sure the references and citations are handled properly, using BibTeX, as illustrated in dissertation templates.
Other aspects, including insight There are a variety of other aspects that contribute to the quality of adissertation. These are hard to describe exactly but are suggested by the list: difficulty of ideas used; novelty of techniques; originality; independent thinking; individual expression; critical writing (as opposed to regurgitation); appreciation of the meaning, context and significance of work.
1.6 PLAGIARISM AND COLLUSION
You must make clear the extent of any work in your dissertation that is not your own. Such work could be, for example, a program or library written by somebody else, or an account of theory in a particular book which you draw on extensively. If you are in any doubt about whether an explanation is needed then it almost certainly is. You will already know that plagiarism and collusion are very serious offences (see the guidance here). In the context of an MSc dissertation any significant plagiarism or collusion would undoubtedly lead to the degree not being awarded.
Unless agreed in writing with your supervisor at the start of your project, and acknowl- edged formally in your dissertation report, all use of generative AI in the dissertation research and writing process is forbidden. If we detect use of such tools, we will assume that you are seeking to gain an unfair advantage over other students on the dissertation module and you will be subject to the University’s disciplinary processes.
1.7 IF THINGS GO WRONG
If there are significant difficulties with supervision these can be discussed, in confidence, with the Course Director, Dr Kostas Triantafyllopoulos, or with the Dissertation Module Leader, Dr Miguel Juarez. Students should also make sure that their supervisor is informed of any special circumstances that should betaken into account in grading the dissertation.
2 SUPERVISION FOR PART-TIME STUDENTS
Due to the varying nature and needs of the projects it is not possible to give a widely applicable description of the supervision process. In addition, supervisors and support workers need to take time away from teaching for holidays, conferences and other research activities (especially in the summer) and so careful planning and flexibility will be needed by all concerned.
Part-time students should try to meet in-person with their supervisor if they are in Sheffield anyway (e.g. for exam weeks), and should expect around 10 hours of dedicated support from their supervisor in total. You should use some of these 10 hours of supervisor support to get feedback on drafts of the dissertation. We recommend that you seek feedback on one chapter of your writing. Do not however expect your supervisor to read and comment on the entire dissertation. The following is usually a good approach.
• Ask your supervisor to read one draft chapter (but not the introduction) and return formal feedback on the style, content and layout. This chapter must be sent to the supervisor by a pre-agreed date well before the September submission deadline. You should not expect them to correct basic linguistic issues, but they will highlight if there are serious concerns in this regard.
• Based on this feedback, you may choose to redraft and then use further support time allotted to you to resubmit to you supervisor for further comments. You should not expect your supervisor to read more than two drafts of a single chapter, however.
• Make sure that you leave sufficient time for your supervisor to read your draft chapter. Providing that you pre-agree a submission date, you should normally expect them to provide you with written feedback within two weeks from sending it to them — reading and giving detailed feedback is a time consuming process.
2.1 SCHEDULING SUPERVISOR TIME
It is crucial to realise that your supervisor may not be able to work on the sametime scale as you and they may be away for periods of the year (especially the summer). You cannot expect your supervisor to fit in supervision of an entire dissertation in the few weeks before the submission date. You must therefore carefully plan your work so as not to leave everything to the last minute.