Some Exercises to Help you Scope your Thesis or Dissertation

• Ask, “What is my halting criteria?” or “How will know that I've won?” This will help you know what work you need to do and when the work is done.
• Prepare the elevator pitch, that is, a one sentence description of what your contribution is or will be. Elaborate on this to create a one paragraph description of your research. Make it simple enough that your grandmother could get the gist of what you have done.
• Ask “How can this research fail?” If there's no risk in your work, then it's not research.

Tips on Writing

• Find a good model paper and copy style
• Read Strunk and White's Elements of Style
• Let what you write sit for a few days and then re-edit.
• Don't get your feelings hurt when you get negative feedback
• The PTAV pattern
  • State the Problem that you are solving, why it is important, and why your solution is a solution.
  • Present relevant Theory, including relevant literature and mathematical background, trying to show why your solution is a contribution to the field.
  • Present your Algorithm or approach to solving the problem, being clear to state what you are doing, why you are doing it, and what principles apply.
  • Validate your solution using one of the techniques below. Include experiment design, definition of your performance metrics, analyses of results, and limitations of the results. A separate limitations section is something that I've recently encountered and really like.

Tips on Presenting

• Power point is a visual medium, so try to use some visual that either literally or metaphorically represents the idea on the page
• Use minimal text on each power point slide; remember that you'll be speaking
• Plan on between one or two minutes per slide, meaning that a 25 minute presentation should not have any more than 25 slides
• Be careful with animations, they are often distracting and really get in the way if you have to skip slides

Tips on Conducting Experiments

• Understand different types of experimental validity
  • External validity: can cause and effect relationships based on a specific scientific study generalize to other populations and conditions?
  • Internal validity: do the inferences of a scientific study demonstrate a causal relation between two variables?
  • Ecological validity do the methods, materials and setting of the study must approximate the real-life situation that is under investigation?
  • Construct validity: does a scale measure or correlate with the theorized thing you are trying to measure?
  • Content validity: does a measure represent all facets of a given construct? (Situation awareness, intelligence quotient, autonomy)
• Always do a pilot study if you are working with human subjects, both to work out bugs in the design and to make sure that you are seeing the trends that you want
• All experiments must have IRB approval, and everyone conducting an experiment must go through BYU's online IRB training
• Record all the data you can; you can always find ways to process raw data if you need to measure something new, but if you don't record what you want then you have to do the study again.
• Use proper statistics, but remember that the effects that we are interesting in must be big to be interesting. BYU does statistical consultation for student work.
• Understand the right metrics for the task. See this paper for a good start.
• When performing a between subjects experiment design, give both groups a similar prior task to see if they are similar. This helps strengthen any conclusion about whether differences in measures were the result of the treatment or based on prior biases of the different groups. This prior experience should also be used to make sure that all subjects have the same comparison standard to anchor their subjective rankings (SA, workload).
• Find an online tutorial on how to design experiments. Pay close attention to training, counterbalancing, and ecological validity.

Latest writing tips document from Dr. McLain