Proposal Writing Notes
Following are some thoughts on what should go into a PhD research
proposal in the outline I would expect one to follow. Before you go
further, you should have a mental picture of what you’re building.
First, you should have seen an instance of the thing you propose to
- I have read a dissertation and attended a dissertation defense.
- I have read a dissertation proposal and attended a proposal defense.
The latter can be a bit tricky as proposal defenses have traditionally
been closed exams. More recently they have been made open, but if you
can’t attend one then don’t worry too much about it. Second, you
should understand the process and outcomes of a dissertation:
- I know what an apprentice is
- I know what an apprentice piece is
The roles of student and advisor are very much the roles of apprentice
and master from the days of learning a craft or trade. You will learn
to do research by working with your advisor. Your dissertation is an
apprentice piece. In the same way that an apprentice piece is a
demonstration and not necessarily a practical artifact, your
dissertation is a demonstration that you know how to do and present
research. The practical artifacts are those papers and research
proposals that result.
- Hypothesis or Research Statement
- Tight, focused definition of what you will do
- May appear repeatedly in the document, but most always be the same
- Hypothesis works for experimental dissertations
- Research statement works for formal or tool oriented disserations
- Why are you doing this work?
- What problem are you attempting to solve?
- Why is the problem important?
- What real-world issue is motivating your work?
- What will you do that is unique?
- What new capabilities or knowledge will result?
- What is the intellectual contribution to the literature?
- Overview the entire proposal
- Exactly what it says - overview the entire proposal.
- The reviewer should learn about the complete proposal at a 10,000
- Many reviewers will make their decision reading the introduction
and use the remainder of the document to confirm their decision.
- The introduction should grab the reader and compel them to continue
- Do not write an introduction that says “In Chapter 1 we…, In
Chapter 2 we…,”
- Establish that this is a worthy problem and you know how to solve it
- Do not write a wishy-washy hypothesis or research statement
- Background work
- Prerequisite knowledge for reading the dissertation
- Where the research comes from
- Supporting information
- Related work
- Work simliar to yours worth comparing with
- Work that supports your work or puts it in contex
- Convince the committee you know the literature
- Do not be negative about other people’s work. They are your
- If you write this well, you can simply cut-and-paste into
- Building the solution
- Coding up a widget
- Writing up a mathematical theory
- Running verification tools
- Some combination of the two is most likely
- Describing the solution
- What does it look like intellectually?
- How is it constructed?
- Compare with related efforts
- Evaluation of the solution
- How will you know when you are done? Know precisely what “done” means.
- How will you know whether “it” worked? Identify what correct means
working from your hypothesis or research statement.
- Analytic evaluation involves mathematical formalism and proof
- Experimental evaluation involves performing experiments and
- Most dissertations that I advise involve both analytic and
experimental evaluation, but most lean towards experimental
- Understand how research is performed in your discipline. This is
- Automate verification activities wherever possible.
- Say “I developed…” or “We developed…” and never hide from your own work
- What kind of evaluation you do depends almost entirely on your
hypothesis or research statement. Hypotheses are almost always
supported experimentally while mathematical objectives are supported
- Publishing papers is astonishingly easier when you have solid
analytical or experimental results. I will review 30 papers in a
given year and 25 of them will have horrible or non-existent
evaluation sections. The other 5 get published.
- Work so far
- Your work, not related work
- Preliminary results
- Show that it is possible to construct the thing, mathematically or
experimentally, that you aim to construct
- Work you must do to finish
- Research and publication, not dissertation writing
- Provide a schedule for your work
- Identify where you plan to publish
- This is your contract with your committee
- Remember that no war is won without a plan, but no war goes
according to plan
- Plan on publishing before your defense, not after. It’s easy for
your advisor to defend a dissertation with peer-reviewed
publications from it
Conclusions and Future Work:
- Summarize the entire proposal in your conclusion
- Conclusions are almost a shorter introduction
- State your goal
- State how you intend to achieve it
- State how you intend to evaluate it
- Identify again what remains to be done
- What you have done thus far
- What you have left to do
- Outline your research plan
- For the dissertation (again)
- For the longer term
I can almost assure you that I will ask these questions. You might as
well answer (or avoid) them before I do:
- Which of these 5 dissertations do you intend to write? Most
students propose too much. Work with your advisor to select an
appropriate scope for your work
- What is your contribution? What did you write? What did you do?
Most students are unwilling to say what exactly they did. Make
sure I know what your contribution is.
- How will you know when you are done? Agree on what constitutes the
end of your dissertation. Treat the proposal like a contract.
- Where will you publish this work? Work is not done until published. Don’t make the mistake of not publishing your dissertation work.