Chronology Current Month Current Thread Current Date
[Year List] [Month List (current year)] [Date Index] [Thread Index] [Thread Prev] [Thread Next] [Date Prev] [Date Next]

[Phys-L] Re: Math SAT



While the statistical analysis given is correct, it does not account
for systematic errors. For example if only the top students take the
SAT and a larger proportion of the lower students that year take the
ACT, the SAT would show improvement. Are more students taking SAT
prep courses? Or what if fewer students decide to go to college?
Statistics like this need to be carefully analyzed to remove such
possible errors.

But there are many imponderables such as:
1. Is this for exactly the same test (I thought they changed some
questions each year)
2. What changes have occurred in the population that takes the SAT?
3. What actual statistical changes occur from one year to the next?

Finally can one find a causal link between this rise and what is going
on in schools and will the rise make a significant difference in the
quality of students in college?

One of the problems is that we are expending a large effort to whip
the teachers, administrators, and students to see a miniscule
improvement in some test scores. Meanwhile solutions which work well
are being totally ignored. Methods of cognitive enhancement are not
being generally used, and research based teaching is rare. The fact
that childhood poverty is one of the largest predictors of failure is
being ignored.

Even if the result is statistically significant, it is insignificant
in a practical sense. The effect size is extremely low. The elephant
has labored and given birth to a mouse. Next year the SAT scores will
be incomparable with previous years because of the changes in the
test.


John M. Clement
Houston, TX



I am seeing all this wonderful news of an increase in the average SAT
from 518 to 520. Am I wrong or is this nothing
hype. Isn't a change of 2 points in the noise?

Let s = the standard deviation of student scores. Assuming a very
large
population of students, of which only N take the test, the standard
deviation of the mean of the scores is s/sqrt(N), where N = the
number
of students taking the SAT.

As a quick estimate: Let s = 100 pt
N = 50 states x (100 high
schools/state)
x (100 students taking SAT/high school) = 5 x 10^5

So standard deviation of the mean is 100/sqrt(5x10^5) = 0.14 pt.

If my estimates of s and N are any where close to reality, the 2
point
rise appears highly significant.

Don Polvani
Northrop Grumman Corp.

-----Original Message-----
From: Forum for Physics Educators [mailto:PHYS-L@list1.ucc.nau.edu]
On
Behalf Of jbellina
Sent: Wednesday, August 31, 2005 2:30 PM
To: PHYS-L@LISTS.NAU.EDU
Subject: Math SAT

I am seeing all this wonderful news of an increase in the average
SAT
from 518 to 520. Am I wrong or is this nothing hype. Isn't a
change of
2 points in the noise?

joe