You are on page 1of 3

Lecture 2: Statistical Inference by Dr.

Javed Iqbal
Characteristics of the Sampling Distribution of Sample Mean (Weiss p-336 and
338)
(1) Mean: 𝝁𝒙̅ = 𝝁 (mean of sampling distribution equals population mean)
𝝈
(2) Standard deviation: 𝝈𝒙̅ =
√𝒏
(SD of sampling distribution equals Population SD divided by sq root of sample size)
(3) Probability Distribution:
a) Sampling from Normal Population: Key Fact 7.2 p-342, Example 7.8
p-342, which says that the sampling distribution of 𝑥̅ is normal regardless of the
sample size
b) Sampling from Non Normal Population: The CLT (p-343) Example 7.9
Fig 7.6 p- 345, (Anderson Fig 7.3 pdf p-366)
In this connection one very important theorem in statistics is called Central Limit
Theorem (CLT). This theorem says that the sampling distribution of 𝑥̅ is normal
regardless of the shape of parent population provided sample size if sufficiently
large.
How large the sample should be for CLT to applicable? A rule of thumb is 𝑛 ≥ 30
Refresher of Normal Probability:
Ex1: 𝑃(𝑍 < 2.13), 𝑃(𝑍 > 1.56), 𝑃(−1.35 ≤ 𝑍 ≤ −0.89)
Ex2: A Normal RW X has mean 𝜇 = 10 and SD 𝜎 = 2, find P(𝑃(𝑋 > 12.5), 𝑃(11.5 < 𝑋 < 13.6)
[Z= 1.25, Z =0.75, 1.80].

Anderson Ch 7, Ex 24 pdf p-372


Sol: a) As per CLT for large sample size ( n = 50>30):

𝑥̅ ~ 𝑁( 𝜇𝑥̅ = 𝜇 = 17.5, 𝜎𝑥̅ = 𝜎⁄ = 4⁄ = 0.5657)


√𝑛 √50
16.5−17.5 𝑥̅ −𝜇𝑥̅ 18.5−17.5
b) P(16.5 < 𝑥̅ < 18.5) = 𝑃 ( < < )
0.5657 𝜎𝑥̅ 0.5657
=P(−1.77 < Z <1.77) = P(Z <1.77) – P(Z <−1.77) = 0.9616 – 0.0384 =
0.9232
17.0−17.5 𝑥̅ −𝜇𝑥̅ 18.0−17.5
c) P(17.0 < 𝑥̅ < 18.0) = 𝑃 ( < < )
0.5657 𝜎𝑥̅ 0.5657
=P(−0.88< Z < 0.88) = P(Z <0.88) – P(Z <−0.88) = 0.8106 – 0.1894 =
0.6212

Do it: Anderson, Ex 25, 27 pdf p-372

Q-25 a) 𝑥̅ ~ 𝑁( 𝜇𝑥̅ = 𝜇 = 533, 𝜎𝑥̅ = 𝜎⁄ = 100⁄ = 10.5409)


√𝑛 √90
P(523 ≤ 𝑥̅ ≤ 543) = P(Z ≤0.95) − P(Z ≤ − 0.95) = 0.8289 −0.1711 = 0.6578

Q-27 a) 𝑥̅ ~ 𝑁( 𝜇𝑥̅ = 𝜇 = 37.39, 𝜎𝑥̅ = 𝜎⁄ = 4.6⁄ = 0.6505)


√𝑛 √50
P(37.39−1 < 𝑥̅ < 37.39 + 1) = 𝑃(𝑍 < 1.54) − 𝑃(𝑍 < −1.54)
= 0.9382 − 0.0618 = 0.8764
Weiss Example 7.10 p-345, Ex 7.69 p-347

[Ex 7.69 (d): 𝑃(|𝑥̅ − 𝜇| ≤ 1) = 𝑃(−1 ≤ 𝑥̅ − 𝜇 ≤ 1) = 𝑃(𝜇 − 1 ≤ 𝑥̅ ≤ 𝜇 + 1) = 𝑃(9.94 ≤


𝑥̅ ≤ 11.94)
= 𝑃(𝑍 ≤ 2.42) − 𝑃(𝑍 ≤ −2.42) = 0.9923 − 0.0078 = 0.9845]

Some Further Practice:


Sampling Distribution of mean: with 𝜇𝑥̅ = 48, 𝜎𝑥̅ =2.3

Computing three types of probabilities:


𝑋̅ −𝜇 50.5−48
(1) P(𝑋̅ < 50.5) = P( < )= P(𝑍 < 1.09) = 0.8621
𝜎𝑥̅ 2.3
𝑋−𝜇 ̅ 50.5−48
(2) P(𝑋̅ > 50.5) = P(𝑋̅ > 50.5) = P( > )
𝜎𝑥̅ 2.3
= P(𝑍 > 1.09) = 1 − P(𝑍 < 1.09) = 1 − 0.8621 = 0.1379
46.5−48 𝑋−𝜇 ̅
50.5−48
(3)P(46.5 < 𝑋̅ < 50.5) = 𝑃 ( < < ) = 𝑃(−0.65 < 𝑍 < 1.09)
2.3 𝜎𝑥̅ 2.3
=P(Z < 1.09)-P(Z < -0.65) = 0.8621-0.2578 = 0.6043

Use of Normal Table for unusual Z values e.g. P (Z < -5.36), or P (Z > 10.45) etc.

You might also like