*All solutions and teacher notes in blue*

AP Statistics Handout Key: Lesson 8.1 (10.1 in our textbook)

Topics: review sampling distribution for 𝑥̅ , 95% confidence interval for 𝑥̅ , four step process

Lesson 8.1 Guided Notes

Students: “I’m gonna drop out and become a…”

YouTube Insta Influencer “TikToker”

Creator

Mr. Beast Kendall Jenner Josh Richards

2020: $24 Million 2019: $16 Million 2020: $1.5 Million Source:

Can you make this kind of money by making social media your full-time career? Probably not, but
maybe you could make a living…

Today’s Key Analysis: Do social media creators, on average, make a livable wage?

Searched “How much I make on YouTube.” Randomly sampled 35 from hundreds

The of results

Reliable data: They show their

private channel revenue pages in the
videos

Source:

Material adapted from the Skew The Script curriculum (skewthescript.org)

Lessons made available under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License
(https://creativecommons.org/licenses/by-nc-sa/4.0)
 2

Review: Sampling Distribution for 𝑥̅

The Population vs. The Sample Mean

𝜇 = population mean 𝑥̅ = sample mean

 Parameter  Statistic used to estimate µ
 Ex: mean salary among all YouTubers  Ex: mean salary in our sample

Estimation
In a world where…
1. The true mean yearly salary among YouTubers is $55,000
2. The true standard deviation of salaries is $29,500

We overestimated the
true mean by $5,000

In most real-life scenarios:

-You don’t know the true mean.
-You survey a random sample of 35 YouTubers. Among them, the average yearly earnings was
𝑥̅ = $60,000.
-Your estimate for the true mean (µ) is the sample mean (𝑥̅ )

The Sampling Distribution

Under certain conditions:

𝑥̅ ~ Normal 𝜇 ̅ = 𝜇, 𝜎 ̅ =
√

Centered at true mean (unbiased) Typical distance of our estimates

from true mean value
Calculations:
𝑥̅ ~ Normal 𝜇 ̅ = 𝜇, 𝜎 ̅ =
√
Means from repeated
𝑥̅ ~ Normal 𝜇 ̅ = 55000, 𝜎 ̅ = random samples of 35
√
YouTubers (in a world
where µ = $55,000).
𝑥̅ ~ Normal(𝜇 ̅ = 55000, 𝜎 ̅ = 4986)

Typical distance of estimates from true mean value.

Typically “off” by $4,986.

Material adapted from the Skew The Script curriculum (skewthescript.org)

Lessons made available under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License
(https://creativecommons.org/licenses/by-nc-sa/4.0)
 3

We overestimated
the true mean by
$5,000

1) Was our estimate’s error amount ($5,000) pretty typical? Justify using the sampling distribution
above.

This is a pretty typical estimation error in this situation. According to the sampling distribution, the
typical estimate error is 𝛔𝐱 = $4,986. Our actual error of $5,000 is pretty close to this.

There should be a way to report both:

a) Our estimate
b) How far off this estimate might be

One way to do this is with a confidence interval!

The t-distribution and interval for 𝑥̅

Sampled YouTuber Yearly Revenues ($)

1) Label the U.S. individual poverty line, the U.S. mean wage, the sample mean, and the
sample standard deviation on the dotplot above.

2) What concerns might you have about assuming the true mean matches the sample
mean above? Why?
There’s a chance that, due to random variation, we happened to select some unusually
high-earning or unusually low-earning YouTubers, which would lead to over or under-
estimating the true mean. Note: students may also point out selection bias in the
sampling method. This will be covered in the discussion question.

Material adapted from the Skew The Script curriculum (skewthescript.org)

Lessons made available under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License
(https://creativecommons.org/licenses/by-nc-sa/4.0)
 4

First Attempt: Creating the Confidence Interval for our Data

Under certain conditions:

𝑥̅ ~ Normal 𝜇 ̅ = 𝜇, 𝜎 ̅ =
√

1) What prevents us from directly calculating 𝜎 ̅ ? What can we do to provide a reasonable estimate?

We don’t know the true standard deviation of YouTuber salaries (𝝈). We plug in our sample standard
deviation (𝑺𝒙 = $𝟖𝟐, 𝟏𝟖𝟖) in its place to provide a reasonable estimate.

2) Using the YouTube data, what is 𝑆𝐸 ̅ ? Show your work.

𝑺𝒙
𝑺𝑬𝒙 =
√𝒏
𝟖𝟐𝟏𝟖𝟖
𝑺𝑬𝒙 = = $𝟏𝟑, 𝟖𝟗𝟐
√𝟑𝟓

3) Why can’t we directly calculate the confidence interval

using the method show on the diagram to the right?

Because we used an 𝑺𝒙 in place of 𝝈 when calculating

the standard error, there is some extra uncertainty
about the calculation of the standard error itself. So,
we need a distribution (besides the normal curve)
that reflects this greater uncertainty.

The t-distribution to the Rescue! Note: we can’t do this to calculate the

interval
t distribution
 Wider than normal curve
 Width captures greater uncertainty when using 𝑆
to estimate σ

degrees of freedom (df) = n - 1

 df measures how precisely 𝑆 estimates σ
 The higher n, the more precisely we estimate σ, and
the more our t-curve approaches normal!

Material adapted from the Skew The Script curriculum (skewthescript.org)

Lessons made available under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License
(https://creativecommons.org/licenses/by-nc-sa/4.0)
 5

t critical values

𝑡 ∗ : the critical value of the t-interval

 Tells you how many standard errors you’re including in your interval.
 Determines the confidence level.

Normal (z) t (df = 34)

For normal curve, z* was For t-distribution, we need

1.96 for 95% confidence to go wider to capture 95%

Calculator steps to find t* t (df = 34)

t* = 2.03

2nd  VARS  4: InvT -Area: percent below

interval (2.5% = 0.025)
-df: n – 1 (35 – 1 = 34)

As expected, the t-critical value is slightly higher than 1.96 needed for a normal curve.

Confidence Interval for a Mean

Formula:
Using the formula and the info above, calculate and
 point estimate ± margin of error
interpret the confidence interval for the mean YouTube
 𝑥̅ ± 𝑡 ∗ (𝑆𝐸 ̅ )
salary:
𝑥̅ ± 𝑡 ∗ (𝑆𝐸 ̅ )
𝟕𝟎𝟏𝟎𝟔 ± 2.03(𝟏𝟑𝟖𝟗𝟐)  ($41905, $98307)

We are 95% confident the interval from $41,905 to $98,307 captures the true mean yearly income of
YouTubers.

Material adapted from the Skew The Script curriculum (skewthescript.org)

Lessons made available under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License
(https://creativecommons.org/licenses/by-nc-sa/4.0)
 6

The Four Steps Process

a) Construct and interpret a 95% confidence interval for the true mean salary of all YouTubers.

State:
We are estimating the true mean yearly YouTuber salary (µ), at 95% confidence.

Plan:
We will calculate a one-sample t-interval for µ, if all
conditions are met.
Conditions
Random: The sample of 35 YouTubers
was collected randomly

10%: 𝑛 ≤ 0.10𝑁 Normal/Large Sample:

35 ≤ 0.10 (all YouTubers) 𝑛 ≥ 30
It’s reasonable to assume 35 is less than 10% of all 35 ≥ 30
YouTubers The sample size is large

Do: For “DO” phase: Calculator steps

T-interval(𝑥̅ =
70106, 𝑠 = 82188, n
= 35, confidence =
0.95):
(𝟒𝟏𝟖𝟕𝟕, 𝟗𝟖𝟑𝟑𝟗)
Stats
𝑥̅ : sample mean
STAT  TESTS  𝑆 : sample stdev Output
8: TInterval n: sample size
Calculate

Conclude:
We are 95% confident the interval from $41,877 to $98,339 captures the true mean yearly income
of YouTubers.

Material adapted from the Skew The Script curriculum (skewthescript.org)

Lessons made available under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License
(https://creativecommons.org/licenses/by-nc-sa/4.0)
 7

Lesson 8.1 Discussion

We can directly estimate the true mean income of YouTubers from Ad Revenue:

Creator Share of 2019 YouTube Ad Revenue

$8,331,950,000 / # of Creators
18,000,000 = Average Yearly Pay Per Creator
$463

Confidence interval we estimated: $41,877 to $98,339

Check out our recommended
set of class discussion norms.
Discussion Question: Why was our confidence interval so far off?
Click this link.

We only sampled among people who made videos about their YouTube incomes. These people likely
have more to brag about than those who don’t make videos about their YouTube incomes.

Even though we sampled randomly among that group, the selection of that group itself was biased.
So, we overestimated.

Takeaway: When checking the random condition, check that sampling from whole population.

Minimum Sample Size Problem

Administrators at your school want to estimate how much time students spend on homework, on
average, during a typical week. They want to estimate  at the 90% confidence level with a
margin of error of at most 30 minutes. A pilot study indicated that the standard deviation of time
spent on homework per week is about 154 minutes. How many students need to be surveyed to
estimate the mean number of minutes spent on homework per week with 90% confidence and a
margin of error of at most 30 minutes?

Sources:
- Alphabet 4th quarter earnings release: https://abc.xyz/investor/static/pdf/2019Q4_alphabet_earnings_release.pdf?cache=05bd9fe
- “How YouTube Ad Revenue Works,” Investopedia, June 4, 2020: https://www.investopedia.com/articles/personal-finance/032615/how-
youtube-ad-revenue-works.asp#:~:text=Enabling%20ads%20on%20your%20YouTube,get%20the%20remaining%2055%20percent.
- “How Many YouTube Channels Are There?”, Tubics, https://www.tubics.com/blog/number-of-youtube-channels
- SocialBlade, data accessed 2019

Material adapted from the Skew The Script curriculum (skewthescript.org)

Lessons made available under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License
(https://creativecommons.org/licenses/by-nc-sa/4.0)