Search All of the Math Forum:
Views expressed in these public forums are not endorsed by
Drexel University or The Math Forum.
|
|
|
|
[ap-stat] Re: Newsweek
Posted:
May 11, 2005 4:52 PM
|
|
Well,
Since I am a teacher at the #21 ranked school (2nd in California) I think it is a great system!!!!!!
Us at Troy did a search on the #19 school (#1 in the state of California: Eastern Sierra Academy) The have a ratio of 4.25 to our ratio of 4.165. Funny thing, we will graduate about 550 students this June, adn they will graduate 5. The total number of students there is only 22, with 3 full time teachers!
Obviously, since this is a statistic, it must be true.
A not so serious (or overly excited) teacher,
Paul Rodriguez
Troy High School
Fullerton, Ca.
----- Original Message -----
From: Dennis Roberts
To: for Teachers of AP Statistics
Sent: Wednesday, May 11, 2005 10:35 AM
Subject: [ap-stat] Newsweek
IN reading the current Newsweek article ... make sure to read the FAQs .. where Matthews "explains" all his rationales about why he did what he did ... what schools were NOT included ... etc. That link is:
http://www.msnbc.msn.com/id/7760504/site/newsweek
If anyone knows of any link to a spreadsheet type of file for these data (w/o going to Matthews directly) ... it would be very helpful. I can't find any such link at the Newsweek site ... sometimes if a list is rather uniform up and down the page, it's cut and pasteable but this one zig zags all over the place.
IF FOR NO OTHER REASON ... the data are "useful" in showing the problems one has when working with ranks and making some inference to the underlying variable that has been ranked. In this case, it's the challenge ratio ... which is the # of AP or IB TESTS that are given in May ... for all students ... / # of graduating seniors in June. A CR of 3 for example would mean that at that particular school ... 3 AP and/or IB TESTS were given PER senior who graduated. It's possible (but doubtful) than not a single senior took ANY AP or IB test. It does NOT mean the number of tests taken BY graduating seniors ... and that distinction is important to keep in mind. (What would have made more sense is to know how many AP and/or IB tests each of those graduating seniors had taken).
I took the list of 1000 ranked schools ... ranks of 1, 50, ..... 1000 .. then rounded the CR to the first decimal place (strictly for convenience) ... and listed them out. Here are the data. The diff value simply is the running difference between a CR value and the one right below it.
MTB > prin c1-c3
Row rank challratio diff
1 1 10.8 *
2 50 3.3 7.5
3 100 2.8 0.5
4 150 2.5 0.3
5 200 2.3 0.2
6 250 2.1 0.2
7 300 2.0 0.1
8 350 1.8 0.2
9 400 1.7 0.1
10 450 1.7 0.0
11 500 1.6 0.1
12 550 1.5 0.1
13 600 1.4 0.1
14 650 1.4 0.0
15 700 1.3 0.1
16 750 1.3 0.0
17 800 1.2 0.1
18 850 1.2 0.0
19 900 1.1 0.1
20 950 1.1 0.0
21 1000 1.0 0.1
One might inquire if there are outliers in this data set?
When you go from 1 to about 100 as a rank ... the CR goes from nearly 11 down to a little less than 3. It then takes a change in ranks from about 100 down to 300 to go from a CR of 3 down to about 2. And then it required a change in ranks from about 300 to 1000 to drop one more point on the CR scale. The notion that one can interpret the ranks like ratio scores ... that is ... a school with a rank of 200 versus one with a rank of 400 ... has TWICE the CR is just not the case.
The graph is below (with obvious interpolation) since I only had 21 data points.
Dennis Roberts
dmr@psu.edu
http://www.personal.psu.edu/users/d/m/dmr/droberts.htm
|
|
|
|
