SECTION 6.4 Confidence Intervals 385
- Suppose that you go out and collect 50 samples of the random
variable 4×rand and compute the mean x. Compute the 95%
confidence interval so obtained. Does it contain the true meanμ?
(See Exercise 1, above.) - We can build on Exercise 2, as follows. The following simple TI
code can be used to count how many out of 100 95% confidence
intervals for the meanμof the random variable 4*randwill actually
contain the true mean (= 2):
PROGRAM: CONFINT
:0→C
:For(I,1,100)
:4*rand(50)→L 1
:mean(L 1 )→M
:M−. 32 →L
:M+. 32 →U
:C+ (L≤2)(2≤U)→C
:END
:Disp C
:Stop
(a) What is the number .32?
(b) What isC trying to compute?
(c) Run this a few times and explain what’s going on.
6.4.2 Confidence intervals for the mean; unknown variance
In this section we shall develop a method for finding confidence intervals
for the meanμof a population when we don’t already know the variance
σ^2 of the population. In the last section our method was based on the
fact that the statistic
X−μ
σ
was approximately normally distributed.
In the present section, since we don’t knowσ, we shall replaceσ^2 with