December 2011 Semi-Statistical Predictions Forum

[SuchGr8Heights]

Anybody dying for an estimate of your score on today's test? It's a very crude estimate but maybe it might help calm you down a little bit? If you're interested reply to this post with every practice test you've taken in recent weeks (I used a sample size of n=10) and what date you've taken the test. You can also send me a PM with the same info and I can figure out the calculation for you as well. Again, it's a VERY crude estimate but it might give you some piece of mind. Happy waiting!

[SuchGr8Heights]

So, a couple people have been e-mailing me to elaborate on how I formulated my prediction. I will demonstrate it for you:

Date Score
11/9(1) 161
11/11(3) 155
11/15(7) 161
11/17(9) 165
11/20(12) 162
11/22(14) 159
11/23(15) 166
11/24(16) 164
11/27(19) 169
11/30(22) 165

From those numbers I arrived at a linear regression:
f(x)= 158.0097 + .3975x (where "x" is the day in parentheses above)

I then used this formula and plugged in x=25 (or December 3rd).
f(25)= 167.9472

A simple average of all the scores will low-ball where I actually stand because there was a positive correlation between the date of the PT and the score (I got better at taking the test each time).

Hope this makes a bit more sense for you all!

[soscurred]

Uhhh, as an economics major that specialized in econometrics, your model is completely wrong. You're basing this off of a progression of practice test scores which omits SIGNIFICANT variables such as incalculable factors (like stress). The values will definitely be biased.

[SuchGr8Heights]

"Semi-statistical"

[SuchGr8Heights]

And don't you think completely wrong is a little harsh? There is a definite upward trend!