Class 7 HIV Testing . . . continued
Discussion
Continue the discussion about drug testing and HIV tesing.
Homework for Monday, February 27
Read Chapter 15 and Chapter 3 of FPPA. Do the review exercises at the end of
each chapter.
Journal Assignment
- 1) It is estimated that about 1.5 million people in the United States are
infected with HIV. Suppose a randomly chosen person in the United States
tests positive for HIV in a single Elisa test. Find the probability that the
person is actually HIV positive.
- 2) How would you go about estimating the number of people in the United
States who are infected with HIV? What difficulties do you forsee and how
would you try to overcome them?
- 3) A lot of people had trouble with review exercise #12 from Chapter 2 of
FPPA. This exercise is a lot like the Berkeley graduate admissions story in
the chapter. Both are examples of "Simpson's paradox". Spell out the analogy between the two examples. Can you make up another scenerio where Simpson's paradox could occur?
Journal Comments
I (Linda) enjoyed reading your journals. Your answers to the three journal
questions we asked were thoughtful and interesting. In addition to answering
assigned questions, we would like you to put lots of other things in your
journals: for example, questions and comments related to class discussions,
and experiences that remind you of the class. (This time, I just gave extra
credit for people who included these extras, but in the future they should
be part of a complete journal.)
Here are some more specific comments:
Aids articles
Most people preferred the retonavir study because of its large size and
controlled, double-blind set-up. Some people pointed out that the 3-drug
therapy studies still serve an important purpose: it allows scientists to
get an idea of whether or not to test drug combinations further, without
having to go through a lengthy and expensive controlled study.
Some people were also concerned about the 15% drop-out rate in the retonavir
group and wanted to know more about the health of these patients.
Weather predictors
A lot of people objected to the idea of subtracting from the weather
predictor's salary and some people found the whole topic frivolous.
Survey data
People went wild on this one, and noticed, among other things:
-- One Princeton student had a pulse of about 15 beats per minutes and one
Grinnell woman was almost 9 feet tall. These "outliers" are probably
mistakes and can mess up calculations. One solution is to leave them out.
-- There are a lot of "round numbers" like 600 and 700 for SAT scores.
Possibly people were estimating, or as one student suggested, "rounding up".
-- There are more men than women in the Princeton class, and more women than
men in the Grinnell class. In fact, there are so few women in the Princeton
class (4) that comparisons of men and women probably don't mean much.
-- There is a high proportion of lefties in the combined Princeton/Grinnell
data (over 20%).
-- The Princeton students have higher SAT scores and gpa's than the Grinnell
students (almost everyone noticed this).
-- The Princeton students have a smaller mean family size than Grinnell
students.
-- Verbal SAT scores decrease with increasing shoe size.
-- The people who get more exercise have lower pulse rates.
Other comments
The tv show ER had an episode in which the chief doctor got in trouble for
hand-picking patients for a medical experiment instead of dividing them up
by lot.
The American Medical Association has a homepage for the latest aids news.