I want you to practice doing a quantitative expert elicitation that produces a CDF of an expert’s estimate of an unknown number. You should follow the protocol described in chapter 9 of TPPA. You can choose your own topic for this. You only need to conduct the interview with one person and you only need to elicit one quantity (with uncertainty). I want you to do a good job eliciting the uncertainty/range of possible values for that quantity, so that you can build it up into a CDF. The quantity should be one that they don't know exactly or can easily look up (so don’t ask how long they have worked at their job or the price of gasoline today in Dubai) - it should be something that they have some reasonable uncertainty about, which makes the elicitation more interesting.
Note: you want to make sure the interviewee is reasonably "expert" at the topic. This is most easily done by finding your expert first then figuring out an interesting topic that you think they'll know about. [If you need a good topic, I suggest that most adults are reasonably expert at the following uncertain question: “How many years from now until you retire?”]
Your write-up should be a brief experimental report that includes each of the following:
- A description of your experience in the interview, and the results. Did everything go as planned? If no, what happened and how did you accommodate the change? What were the results from your interview?
- Include a table or graph showing the elicited CDF of the expert’s response.
- Some discussion on what you learned from the process. Did the interviewee give a “good” answer, or was their response confusing or odd in some way? Did the interviewee come close to the true value (if you know it)?