John H. Robinson, IV wrote:
So let's take a look at something that is quantifiable: SAT scores.
http://www.stark.kent.edu/writing/review%2098-99/bodollo.htm
Mean Median Mode Minimum Maximum Range Standard Standard Sample
Error Deviation Variance
529 523 493 476 588 112 4.50 32.11 1030.88
Please note that Mean is Average. Please also note that Mean != Median.
I submit I have sufficiently proved my point, QED.
That looks ... wrong .. why is the range only 112 ... it should be 600.
In addition the standard error *does* allow significant overlap in the
mean and median. The standard error looks *way* too high for SAT
scores. Finally, SAT's are *scaled* so that 500 is *defined* to be the
mean.
*looks at data*
Hmmm. You can't calculate error or median like that from aggregate
statistics. Mean and median need to be weighted and sorted to work from
aggregate statistics.
*looks for SAT statistics from ETS*
Boy, ETS is doing a good job of hiding anything useful about SAT
statistics. Grumble. I'll keep looking.
-a
--
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list
