How to write a user study proposal

===========================================================================

Outline for today:

Study outcomes:
 * should be actionable
   not just intellectually interesting

Related work (and contributions):
 * for each piece of related work:
    * what it accomplished (results)
    * how (techniques)
    * why (what was different and why did it work well)
    * contrast with your approach
 * One style is a systematic literature review, which just means that you
   give an explicit set of rules about which papers you discuss.

False positives/negatives

Study design practice

===========================================================================

Study outcomes:

How should others react if they believe your results?
Be sure to discuss the consequences if you have
 * positive result
 * negative result

Some possible outcomes:
 * change construction of future tools
 * adopt tool
 * more investigation in the same context
 * apply new designs to future research
    * measure/control new factors
    * stop unfruitful research
 * models/explanations => hypotheses/proxies

===========================================================================

Suppose that 1/10,000 people have cancer, and your medical test is 99% accurate.
In a population of 1,000,000 people:

                        total     test says "no cancer"     test says "cancer"

people with no cancer  999,900      989,901                   9,999

people with cancer         100      1                         99

In other words, if the 99%-accurate test says "cancer", then you have less
than a 1% chance of actually having cancer!

===========================================================================

How to structure your conclusion section

 * rephrase intro (easy to do, safe with readers)
 * the hypothesis is correct
 * how to apply the lessons in the future
 * open questions, future research
 * limitations

(Sometimes some of these go in their own sections.)

===========================================================================

The whole paper should answer the reader's question, "Why do I care?"

What are the concrete outputs of your research?
 * an implementation (not the important part!)
 * quantitative and qualitative results

Explain why your approach worked and why no one else tried it before.

Discuss consequences and extensions.

===========================================================================