Defending and writing your dissertation
- At the end of your fourth year, present a one-hour thesis proposal in the Wednesday research meeting. Plan on 30-35 minutes of talk time, followed by a half-hour of open discussion, primarily by the committee. Put the stuff you want feedback on first. Because it's unlikely you'll make it to the end of your deck, don't put anything critical at the end.
- Orals aka defense
- Who should I ask to be on my committee? Likely, a couple faculty will be familiar with your work. Ask them. To round out your committee, ask people who you may not know, but whose expertise will shine valuable light on your work. At least one person must be from outside the department. You can also (I think) have one research scientist who is not in the UC Academic Senate. This could be someone from e.g., an industrial research lab or faculty from a non-UC.
- Schedule your orals *at least* four months ahead of time — e.g., if you're defending in May, schedule in January. Coordinating 5 faculty schedules is essentially impossible. And you may have to drop one person you really want for scheduling reasons. Design students defend in the weekly public seminar, which makes coordinating at least Design faculty schedules easier. After the public defense, there is a closed-room session. Book a nearby room.
- Be judicious with faculty time. Faculty enjoy participating because they learn a lot: you'll put more time into your thesis talk than any other in your life. At the same time, life is short, so don't tap people unless they're really necessary. Don't ask more than the required three people to be on your reading committee; don't ask more than the required five people to be on your orals committee.
- Theses build theory, so articulate motivation and rationale for what you did
- Have a friend bring coffee.
- More info here
- written dissertation: serves two main purposes
- First, it's a comprehensive description of your research. That way, if someone else would like to replicate or extend your research, your dissertation provides everything needed to do so.
- Second — and, from a learning perspective, more importantly — a thesis is your chance to ask, as a whole, what your research means. This is hard. And this challenge is mostly reflected in the title, abstract, intro chapter, and concluding chapter. This is where you'll spend most of your time. It's also where I'll spend most of my time with you. (The internal chapters are likely mostly written up as conference papers already, so they'll get expanded with additional graphics/figures/explanation, but that's mostly straightforward for both you and I.)
- With the title, abstract, and intro chapter, you should expect a lot of Socratic discussion about why your title is what it is, what that means, how you set up your arguments, and what claims your introductory chapter makes. As I discuss with 'turf' above, you want to make broad conceptual claims… but not broader than you can support. This will require concentration, deep thought, and lots of iteration.
- Dissertation prose builds a logical argument, a bit like a mathematical proof. Every sentence should follow either from what you've written already or from a reference. For example, if you write, "X is increasing", the reader should be able to verify that assertion either from what you've written or from the reference accompanying that assertion.
- Download a dozen or so dissertations that you admire, and reverse engineer their argument. This will help you set up your intro chapter. Pick a couple from my group, a couple from HCI/Design elsewhere, and several from adjacent fields (e.g., other areas of computer science, the cognitive sciences, & learning.) ACM dissertation award winners are a great place to start.
- The point of a conclusions chapter is to build theory, think expansively, and tackle broader conceptual issues, not (primarily) to reiterate the literal outcomes of your work. As with any writing, begin with an outline of the broader themes you would like to reflect on. I suggest ~6 subsections with headers (order here random; optimize for narrative coherence):
- Shortcomings/Challenges. What were the challenges of this work? For example, the difficulty of gathering longitudinal data b/c people are tied to their email client. How would you recommend others address these challenges?
- Implications. If the world were to transform to rely heavily on your work, what broader technical and societal transformations would arise? How would your work scale? What would the social and technical challenges be? This might be a good place to connect your work to broader (relevant) writing on the role of technology in society.
- Systems/domains: where else could your techniques and insights apply? Think big.
- Patterns: What are the research issues for advancing and crystallizing patterns in this area?
- Methods: Reflecting on the methods you employed – system building, studies, etc. What worked best? What effort was wasted? What things would you do differently? And what would you recommend (or not) to others? (This might go adjacent challenges.)
- Behavior: what do you want to know about human behavior? For example, we know that most behavioral interventions (like exercising more) are tough to stick. Speculate beyond your data about behavior. Generalize. Make wild guesses. And think about what next steps would help you get to those.
- Your reading committee
- Different faculty provide different kinds of feedback on theses. Generally, it's the advisor's responsibility to go through with a fine-toothed comb. From others, you'll likely get a few specific requests for changes (e.g., add/change a graph/description), but mostly high-level reactions.
- Your advisor should complete all her/his changes first. Only send your dissertation to your committee after you have made all the changes that your advisor has requested and you believe the dissertation to be done. (Unless your committee request otherwise.) This serial process avoid wasted work by your reading committee.
- When you send your draft for feedback, to give your outside readers confidence that it's ready for them:
- it should look like a thesis: all the headers should be formatted corectly, all the figures should be legible, etc. It should signal that it has been carefully crafted as opposed to composed slapdash and half-complete. If the formatting is a mess, what does that say about the data analysis?
- cc me on the email, and explain that what you are sending incorporates all the changes I have requested.
- If you have a few small things that still need to be done, that's fine. This might be that you'd like to update a figure in the middle of Chapter 5 to include the most recent data, that you have a few stray importing typos in the references, or that you need to finish collating/formatting your experimental materials for the appendix. You can send a link rather than an attachment. This enables you to make those final edits, and the committee will always get the latest version. (Just to be clear, these should be small things, not "I'm still polishing the argument in chapter 1".)
- Plan on filing your thesis at least a month after you send it to your committee for review. You should give them 15 days at a bare minumum to give you comments, at least a week for you to address them, and then a few days for you to loop back to them with the final final version for approval.
- Job search: If you do an academic job search, plan on that being your full-time occupation from Thanksgiving through Easter. It'll take you most of a month to compose your research and teaching statements, cv, recommendations, etc. 1-1.5 Months of full-time work to create and revise your job talk. (If you don't believe me, ask a recent graduate who had a successful academic search.) And interviewing means really being prepared and having looked up the people you meet.
Here is good advice on writing your teaching statement.