|
T 5/1/07
|
Recommended
HW: Work at least 35 minutes’ worth
of AP review problems from the Barron’s book. Show all scratch work and
solutions on standard sheets of HW paper. I recommend that you do a mixture
of free-response and multiple-choice questions, with an emphasis on
multiple-choice questions to prepare you for Thursday.
In class and after school: Gather data for group mini-projects.
Group 1 (Denny, In Sung, Mike W.): LSRL t test, height vs. sleep, n
³ 40.
Methodology: Use script, ask hours of sleep first.
Ask approx. 10 per form.
Group 2 (Julian, Marcus, Peter): c2 test
for indep. of hair color and eye color, n
³ 150.
Methodology: 4 ´ 4 matrix, df = 9. Avoid double-counting of
subjects.
Group 3 (Ollie, James, Sam): 2-prop. z test of rock/paper/scissors, n ³ 30 (i.e., 30 pairs).
Methodology: See whether phuman wins against human differs from phuman wins against machine.
Group 4 (Alex, Kellie, Rick): 2-sample t test to compare mno practice quiz against mwith practice quiz.
Methodology: Teach words from a script, tell all subjects
when to take the graded quiz.
(Could be overnight or later the same day for the
graded quiz.) Goal of n1
= n2 ³ 10 but
will accept smaller sample sizes if data are gathered
carefully.
Group 5 (Matt, Nicholas, Mike R.): 1-prop. z test to see if pwilling
to take survey > .7 (n ³ 50).
Methodology: Question from a script, signature
roster, tally of refusers. Note:
This
group should also compute a C.I. for the true
probability of compliance.
|
|
|
W 5/2/07
|
HW due: Writeups for mini-projects. Each group leader will
make a short oral presentation stating the type of test, the null and
alternative hypotheses, and the conclusion. A one-page writeup (PHASTPC
style) is expected from each group. Present your raw data in a table or in a
similar suitable format. Group 5 must also submit a roster of signatures and
a computed confidence interval. Complete sentences are not required.
Remember, these are mini-projects.
If you are not quite finished at the beginning of class, we can use the first
20 minutes or so to finish the writeups. The remainder of the class will be
for review.
|
|
|
Th 5/3/07
|
AP Review
Test (Multiple Choice). Format is
22 questions in 50 minutes (note:
adjusted to 17 questions in 39 minutes), which is the same pace as the real
exam. On the real exam, you will have 40 questions in 90 minutes. Scoring
will be on an AP-style curve and will count 100 points.
Answer Key
Page 1: DEA
Page 2: EDD
Page 3: EBDA
Page 4: BCBC
Page 5: BEB
For almost all of the questions, the most popular answer selected by the
class was the correct answer. The only exception was #11, where more people
chose A than B. While A is a true answer in the sense that it satisfies the
requirements of the problem, it is not the best answer for a question to be asked on a statistics test. If
you disagree (and disagreement is certainly possible on this question),
please see the explanation at the top of p. 435 in your review book. I
realize that the question is somewhat unfair, but that’s going to happen
sometimes: You have to “get inside the head of the author of the question”
and figure out what is really being asked.
The statistics for the test scores, after curve but before bonus, were as
follows: n = 15,  = 77.6, median = 83. There were 7 students with an A (90 or
above).
This is an ideal example of how one could lie with statistics, however. More
of the story emerges when you look at the measures of dispersion: s = 22.6, IQR = 44. Do you see how
important it is to get more information than simply the measures of central
tendency? Laypeople rarely think to ask for the s.d. or the IQR.
There were no outliers. (Of course, how could there be any outliers when the
IQR is a whopping 44 points?)
How to interpret your score: Multiply the number of correct answers by 4,
subtract the number of wrong answers, and ignore the omissions. Divide by 68
to convert to a percentage. Add 20 percentage points, plus 9 points for
copier jam and 2 points for perfect attendance. If you had spare batteries,
add another point.
Example: Bill Bulldog had 10 right, 4 wrong, and 3 omissions. His raw score
is 10(4) – 4 = 36, which is 36/68 or 52.9%. His curved score is 72.9% (C),
but with the various bonuses including spare batteries, his final score is
84.9% (borderline B+).
|
|
|
F 5/4/07
|
HW due: As announced in class yesterday, I will be
collecting questions 18-22 from yesterday’s test. Show your work neatly.
Recommended additional work: Look over
the answer key and figure out where you made errors yesterday.
AP Review Test (Free Response).
Format is 2 questions (1 short, 1 long) in 38 minutes, which is the same pace
as the real exam. On the real exam, you will have 5 short questions and 1 long
question in 90 minutes. Scoring will be the same as yesterday.
|
|
|
M 5/7/07
|
No additional HW due. It goes without saying that AP
exam takers should be studying for tomorrow’s exam, and everyone (AP or not)
should be studying for the Must-Pass
Quiz.
Attendance
Rules for AP Weeks (May 7-11 and May 14-18)
Unless otherwise posted on this Web page, class will
be held each day at the normal time, and attendance will be taken. You are automatically
excused if you have a morning AP exam and if you actually attend the exam.
(If you fail to attend an exam for which you are scheduled, you will be
charged with a cut if you have any remaining, otherwise a point penalty of 1
point per minute of class missed.)
Note: If you have an afternoon exam
and wish to be excused from STAtistics class, you must submit a written
request in advance (e-mail preferred). I will normally approve such requests,
but approval is not automatic. You must make the request yourself. This
policy applies even for people who are planning to take the AP Statistics
exam.
|
|