Hi, Everyone --
I sometimes change questions and expand answers based on experience in section. I know there was some confusion about 8.6.A, although I had tried to head it off in lecture. Here's what I added to the key:
Some students wrote spectrum bias, and were misled by Table 5.2 in the text and a similar slide (#32) in the lecture, so they said no effect on specificity. Spectrum bias has generally been used to describe unrepresentative (separate) sampling of cases and controls, but as I mentioned in lecture last week,we can think of verification bias as a cause of spectrum bias. I also said in lecture that spectrum bias is the more general term, and that in general you should provide the most specific answer you can on problems.
For next year, I've changed slide 32 in Lecture 4 to make this more clear from the start.
Note also for 9.22 part b will be deleted in future years, if we use that problem again.
Have a good weekend!
Tom