Tuesday, May 10, 2016

Answering a question from earlier today...

Chapter 12, number 1a in our fast track book asks us to find a 97% confidence interval for the true population slope.  To do so, we must understand both our formula for confidence intervals and the computer printout.

Formula

b - t (E) < B < b + t(E)

where b is our given slope value from the printout.  (Coefficient for Registrations) = 9.2472
 E is our standard error of the coefficient for the slope.  (SE coef for Registrations) = .2145
and "t" is our t-score.

Since B is the parameter we are trying to estimate, it would not make sense to use it in our formula to find "t".  Therefore "t" is taken straight from our student's t-distribution chart.  The degrees of freedom is seven and we look across for 97%.  However, we see there is only a value given for 95% and 98% in our Actual textbook.  The value at 95% IS 2.365 and the value at 98% is 2.998.  Therefore 97% falls somewhere in between.  A good estimate of this can be made around 2.75 however, the Fast Track book reveals its exact answer to be 2.715.

So, since
b - t (E) < B < b + t(E)
9.2472 - 2.715 (.2145) < B < 9.2472 + 2.715 (.2145)

Since we likely would use an estimate of "t" and not the exact value in this case, 2.7 or 2.75 would have sufficed.  However, I believe it likely that on the AP exam, they will give a value that is easily obtained from the chart or easily estimated.
Any more questions on this can be answered tomorrow.


No comments: