Dear Allstaters,
I have a design and analysis problem that I hope someone will be able to
help with.
I have a series of 16 results (photographs of test strips) which need to
be interpreted by technicians as having either a positive or negative
score. Each result will be viewed 3 times (in randomised order) and the
median score for each of the 16 will be used as the definitive result.
All of the 16 result scores are theoretically negative, but some are
more likely than others to be viewed as positive by the technicians.
Ideally, this will show a range of result scores between the points
where 99% say the result is positive and 99% say the result is negative,
hence all of the results in between these bands will be the results
where most ambiguity in interpretation occurs, and hence these results
can be researched further in the lab.
I therefore have 2 questions:
(1) How can I calculate the sample size, i.e. how many technicians do I
need to read the results?
(2) How can I analyse these results?
Any help will be greatly appreciated!!
Thanks,
Susan
|