Dr.

Alex Liebrecht
[email protected]

Unit 6 Notes
Review of symbols (and Random Variables)

• 𝜇 = population mean
• 𝑋 = random variable
• 𝑥 = realization of random variable 𝑋
• 𝑋̅ = sample mean as a random variable (estimator of 𝜇)
• 𝑥̅ = realization of the sample mean 𝑋̅ (estimate of 𝜇)

The sample mean 𝑥̅ changes from one experiment to another as we tend to repeat the experiment. The
sampling distribution of 𝑋̅ describes the location and spread of where 𝑥̅ will fall.

Let’s say we have some population data --- 𝑁 = 1000 simulated from 𝑋~𝑈𝑛𝑖𝑓𝑜𝑟𝑚(−10, 10)

Now draw samples of size 𝑛 and take the mean.

1|Page
 Dr. Alex Liebrecht
[email protected]

What about an extreme example where the data is skewed?

Used population data 𝑁 = 1000 where 𝑋~Γ(𝑠ℎ𝑎𝑝𝑒 = 1, 𝑠𝑐𝑎𝑙𝑒 = 2)

Now draw samples of size 𝑛 and take the mean.

2|Page
 Dr. Alex Liebrecht
[email protected]
̅
Properties of the sample mean 𝑿

The mean of 𝑋̅ is equal to the mean of the population. To say it very strangely --- “the mean of means is
equal to the population mean.” Formally, we say the following:

𝜇𝑋̅ = 𝐸(𝑋̅) = 𝜇𝑋
The standard deviation of 𝑋̅ is the following:
𝜎
𝜎𝑋̅ =
√𝑛
The variance of 𝑋̅ is the following:

𝜎2
𝜎𝑋2̅ = 𝑉𝑎𝑟(𝑋̅) =
𝑛
Assumption and Conditions

• 𝑛 ≥ 30
̅ for Normal Populations
Sampling Distribution of 𝑿

Suppose a random sample of size 𝑛 is obtained from a normal population with mean 𝜇 and standard
deviation 𝜎. Then for any sample size 𝑛, the sample mean 𝑋̅ has an exact normal distribution with mean
𝜎
𝜇𝑋̅ = 𝜇𝑋 and standard deviation 𝜎𝑋̅ = 𝑋 .
√𝑛
𝜎
𝑋̅~𝑁(𝜇𝑋̅ = 𝜇𝑋 , 𝜎𝑋̅ = )
√𝑛
Example: At a certain airline, aircraft ages 𝑋 follow a normal distribution with mean 𝜇𝑋 = 10 years and
standard deviation of 𝜎𝑋 = 3 years.

a. Draw a bell curve based on randomly selecting one airplane. Use the empirical rule.

b. Draw a bell curve based on randomly selecting 9 airplanes and computing the mean for those 9
airplanes. Use the empirical rule.

c. In the first bell curve, shade 𝑃(𝑋 < 7) and in the second bell curve shade 𝑃(𝑋̅ < 7).

3|Page
 Dr. Alex Liebrecht
[email protected]

Example: Let’s look back at that IQ example. IQ score is normally distributed with mean equal to 100
and standard deviation 15.

a. When randomly selecting 1 person, what is the expected mean IQ score of that one person?

b. When randomly selecting 25 people, what is the expected mean IQ score for that group of 25?

c. When randomly selecting 1 person, what is the standard deviation for the sample mean? Draw a
bell curve using the empirical rule.

d. When randomly selecting 25 people, what is the standard deviation for the sample mean? Draw
a bell curve using the empirical rule.

e. When randomly selecting 225 people, what is the standard deviation for the sample mean?
Draw a bell curve using the empirical rule.

f. When randomly selecting 1 person, what is the probability that they have an IQ score above
105?

g. When randomly selecting 25 people, what is the probability the mean for the group of 25 is
above 105?

h. When randomly selecting 225 people, what is the probability the mean for the group of 225 is
above 105?

4|Page
 Dr. Alex Liebrecht
[email protected]

I am about to attempt to demonstrate that the standard deviation of the sample mean is a function of
the sample size. Let’s see how that looks when we scale for standard normal distribution (𝑍 distribution)

𝑍-score for realization of Random Variable 𝑋

𝑥 − 𝜇𝑥
𝑧=
𝜎𝑥
𝑍-score for realization of Random Variable 𝑋̅
𝑥̅ − 𝜇𝑥̅ 𝑥̅ − 𝜇𝑥
𝑧= = 𝜎
𝜎𝑥̅ 𝑥
√𝑛
Example: Let’s say that we are looking at a product with a mean manufacturing time of 40 minutes and
a standard deviation of 6 minutes. Assume the manufacturing time is normally distributed.

a. When randomly select 1 product and notice the manufacturing time of that product was 43
minutes. What is the corresponding 𝑧-score?

b. When randomly selecting 4 products, the sample mean of the manufacturing time was 43
minutes. What is the corresponding 𝑧-score?

c. When randomly selecting 9 products, the sample mean of the manufacturing time was 43
minutes. What is the corresponding 𝑧-score?

d. How many products would be randomly selected if a mean manufacturing time of 43 minutes
had a 𝑧 − 𝑠𝑐𝑜𝑟𝑒 of 5?

5|Page
 Dr. Alex Liebrecht
[email protected]

What does this look like in notation?

Example: Let 𝑋~𝑁(50, 7)

a. 𝑃(𝑋 < 48)?

b. 𝑃(𝑋̅ < 48)? Use 𝑛 = 30.

c. 𝑃(𝑋 < 𝑥) = 0.05. What is 𝑥?

d. 𝑃(𝑋̅ < 𝑥̅ ) = 0.05. What is 𝑥̅ ? Use 𝑛 = 30.

e. 𝑃(48 < 𝑋̅ < 𝑥̅ ) = 0.45. What is 𝑥̅ ?

f. *Review* 𝑃(𝑍 < 1)?

g. *Review* 𝑃(|𝑍| < 2.5)?

6|Page
 Dr. Alex Liebrecht
[email protected]

Practice: Let 𝑋~𝑁(200, 20)

a. 𝑃(𝑋 > 190)?

b. 𝑃(195 < 𝑋̅ < 205)? Use 𝑛 = 25.

c. 𝑃(𝑋̅ = 200)?

d. 𝑃(𝑋 > 𝑥) = 0.99. What is 𝑥?

e. 𝑃(|𝑋̅| < 𝑥̅ ) = 0.95. What is 𝑥̅ ? Use 𝑛 = 25.

f. 𝑃(|𝑋̅| > 𝑥̅ ) = 0.10. What is 𝑥̅ ? Use 𝑛 = 25.

g. What 𝑧-score corresponds to 𝑥 = 205 when 𝑛 = 1.

h. What 𝑧-score corresponds to 𝑥̅ = 205 when 𝑛 = 25.

7|Page
 Dr. Alex Liebrecht
[email protected]

Practice All Strategies Together: Let’s say that a bottling plant fills 12-ounce soda bottles. The
distribution of the amount of soda dispensed into each bottle follows a normal distribution with a mean
of 11.99 ounces and a standard deviation of 0.03 ounces.

a. Find the probability that a randomly selected bottle contains more than 12.1 ounces.

b. Find the probability that the mean amount of soda in 24 randomly selected bottles is less than
12.0 ounces.

c. Find the probability that the mean amount of soda in 12 randomly selected bottles is between
11.99 and 12.01 ounces?

d. When selecting 1 bottle, what 𝑧-score represents 12.1 ounces?

e. When randomly selecting 24 bottles, what 𝑧-score represents 12 ounces?

f. Suppose 1 bottle of soda is selected randomly. Determine the 70th percentile of that one soda
bottle.

g. Suppose that 6 bottles of soda are randomly selected. Determine the 70th percentile of the
sample mean.

8|Page
 Dr. Alex Liebrecht
[email protected]

Sample Proportion
The sample proportion 𝑝̂ is probably “close” to the population proportion 𝑝. We are essentially saying
that 𝑝 is our best guess for 𝑝̂ .

̂
Properties of the sample proportion 𝒑

𝑝(1 − 𝑝)
𝑝̂ ∼̇ 𝑁 (𝑝, √ )
𝑛

• 𝑝 = population proportion
• 𝑝̂ = sample proportion
• 𝑛 = sample size
• ∼̇= "approximately distributed as”

Another way that this is sometimes written is the following:

𝜇𝑝̂ = 𝑝

𝑝(1 − 𝑝)
𝜎𝑝̂ = √
𝑛

We essentially use the “approximately distributed as” because the distribution is not exactly normally
distributed but it is really close.

Assumptions and Conditions for “Approximate Normality”

• 𝑛 ≥ 30
• 𝑛𝑝̂ ≥ 5
• 𝑛(1 − 𝑝̂ ) ≥ 5
If the above three conditions all hold, it is safe to assume normality for the distribution of 𝑝̂ .

9|Page
 Dr. Alex Liebrecht
[email protected]

Example: Let’s say we are flipping a fair coin 100 times. It is safe to assume that the population
proportion of heads is 0.5. Let’s say we are about to flip this coin 100 times and record the number of
heads.

a. Is it safe to assume the distribution of the sample proportion is approximately normal? Why or
why not?

b. What is the mean of the sample proportion?

c. What is the standard deviation of the sample proportion?

d. Draw a bell curve representing the distribution of the sample proportion.

Example 2: Let’s say that we repeat this experiment but this time, we flip the coin 1000 times. It is again
safe to assume the population proportion of heads is 0.5.

a. What is the mean of the sample proportion?

b. What is the standard deviation of the sample proportion?

c. Draw a bell curve representing the distribution of the sample proportion.

10 | P a g e
 Dr. Alex Liebrecht
[email protected]

I am about to attempt to demonstrate that the standard deviation of the sample proportion is a function
of the sample size. Let’s see how that looks when we scale for standard normal distribution (𝑍
distribution)

𝑍-score for realization of Random Variable 𝑋

𝑥 − 𝜇𝑥
𝑧=
𝜎𝑥

𝑍-score for realization of Random Variable 𝑋̅

𝑥̅ − 𝜇𝑥̅ 𝑥̅ − 𝜇𝑥
𝑧= = 𝜎
𝜎𝑥̅ 𝑥
√𝑛
𝑍-score for realization of random variable 𝑝̂
𝑝̂ − 𝜇𝑝̂ 𝑝̂ − 𝑝
𝑧= =
𝜎𝑝̂
√𝑝(1 − 𝑝)
𝑛
Example: Let’s think back to the last problem where we flipped a fair coin 100 times and the population
proportion of heads is 0.5.

a. What 𝑧-score corresponds to getting heads 55 times?

b. What z-score corresponds to getting heads 60 times?

Example 2: Now let’s increase the number of flips to 1000 times. Still assume the population proportion
is 𝑝 = 0.5.

a. What 𝑧-score corresponds to getting heads 550 times?

b. What 𝑧-score corresponds to getting heads 600 times?

11 | P a g e
 Dr. Alex Liebrecht
[email protected]

Example Problem: Let’s say that we are going to flip a fair coin (𝑝 = 0.5) 50 times. Assume normality
for this problem. I know that we can do this with binomial distribution but I want to start building up
normality.

a. Is it safe to assume normality?

b. What is the expected sample proportion?

c. What is the standard deviation of the sample proportion?

d. What is the probability we get heads less than 30 times? (assuming normality)

e. What is the probability we get heads more than 27 times? (assuming normality)

f. What is the probability we get heads between 24 and 27 times? (assuming normality)

g. What is the 10 percentile for the proportion of heads? (assuming normality)

h. What is the 10 percentile for the number of heads when flipping a coin 50 times? (assuming
normality)

*Reminder to Alex: Emphasize the ease of answering (g) and (h) when assuming normality

12 | P a g e
 Dr. Alex Liebrecht
[email protected]

Practice: The population proportion for people that experience Dyscalculia (a learning disorder in
comprehending arithmetic and understanding numbers) is 0.045. Answer the following questions based
on randomly selecting 320 students.

a. Is it safe to assume normality? Why or why not?

b. What is the expected sample proportion?

c. What is the standard deviation for the sample proportion?

d. Draw a bell curve based on the empirical rule.

e. What is the probability that the sample proportion of students with Dyscalculia will be more
than 5%?

f. What is the probability that less than 10 students out of the 320 will have Dyscalculia?

g. What is the middle 95 percentile cutoffs for the proportion of students with Dyscalculia?

h. What is the middle 95 percentile cutoffs for the number of students with Dyscalculia?

i. What 𝑧 −score corresponds to 5 students having Dyscalculia?

13 | P a g e