Assigned 2/9 Due 2/16 Returned 2/23
[Point number or BBG comment]
4.10
[1] kappa = .56
[2] Yes, predictions were better than random. [A kappa of 0 indicates
random agreement; many students did not address element of randomness choosing
instead to mechanically using rules of thumb.]
4.13
Q1
[3] Blood donors PVP = .0392
[4] Blood donor PVN = 1.00 [discuss rounding; some students answered
".99"; true answer was .999991 which is a lot closer to 1.00 than it
is to .99]
[5] IVDU PVP = .916
[6] IVDU PVN = .998
Q2 -- see p. 014
Q3
[7] The answer presented on p. 105 is still pertinent. Notice that the
PVP in the population was .916 meaning that the false positive rate = 1 - .916 =
.084. That is almost (but not quite) 1 in 10 positives are FPs. Another way to
look at this is there were 9 FPs. It would not be ethical in my opinion to scare
these 9 people when confirmatory testing is a much better option. No deduction
was taken, but in light of some problems with interpreting this problem and
4.10, I'm going to read you your "Miranda rights" -- You are graduate
students. You are expected to interpreted epidemiologic results in an informed
and intelligent way. You should not mechanically repeat interpretations you do
not fully understand. You have no right to an attorney. Anything you say could
be marked down. ;-)
Q4 -- see p. 105
Part II
Q5 - Q7 see p. 105
Part III
Q8
[8] Prev - .0377
Q9
[9] EIA - EIA PVP = .7935
[10] EIA - WB PVP = .9973
Q10 -- see pp. 107 -108