Lab 3D

Directions: Follow along with the slides, completing
the questions in **blue** on your
computer, and answering the questions in **red** in your
journal.

Space, Click, Right Arrow or swipe left to move to
the next slide.

- Throughout the year, we’ve seen that:
- Means are used for describing the typical value in a sample or population, but we usually don’t know what they are, because we can’t see the entire population.
- Means of samples can be used to
*estimate*means of populations. - By including a margin of error with our estimate, we create an interval that increases our confidence that we’ve located the correct value of the population mean.

- Today, we’ll learn how we can calculate
*margins of error*by using a method called the*bootstrap*.- Which comes from the phrase,
*Picking yourself up by your own bootstraps*.

- Which comes from the phrase,

- Load the built-in
`atus`

(*American Time Use Survey*) data set, which is a survey of how a sample of Americans spent their day.**The United States has an estimated population of 327,350,075. How many people were surveyed for this particular data set?**

- The statistical question we wish to investigate is:

*What is the mean age of people older than 15 living in the United
States?*

**Why is it important that the ATUS is a random sample?****Use our**`atus`

data to calculate an estimate for the average age of people older than 15 living in the U.S.

- A
*bootstrapped*sample is when we take a random`sample()`

of our original data (`atus`

)*WITH*replacement.- The
`size`

of the sample should be the same size as the original data.

- The
- We can create a single
*bootstrapped*sample for the`mean`

in three steps:- Sample the number of the rows to use in our
*bootstrap*. `slice`

those rows from our original data into our*bootstrap*data.- Calculate the mean our our
*bootstrapped*data.

- Sample the number of the rows to use in our

- Fill in the blanks to
`sample`

the row numbers we’ll use in our*bootstrapped*sample.- Be sure to re-read what a
*bootstrapped*sample is from the previous slide to help you fill in the blanks. - Use
`set.seed(123)`

before taking the sample.

- Be sure to re-read what a

- Use the
`slice`

function to create a new data set that includes each row from our`sample`

.

- Look at the values of
`bs_rows`

and`bs_atus`

. **Write a paragraph that explains to someone that’s not familiar with**`R`

how you created`bs_rows`

and`bs_atus`

. Be sure to include an explanation of what the*values*of`bs_rows`

mean and how those values are used to create`bs_atus`

. Also, be sure to explain what each argument of each function does.

- Calculate the
`mean`

of the`age`

variable in your`bootstrapped`

data, then use a different value of`set.seed()`

to create your own, personal*bootstrapped*sample. Then calculate its`mean`

. **Compare this second***bootstrapped*sample with three other classmates and write a sentence about how similar or different the*bootstrapped*sample means were.

- To use
*bootstrapped*samples to create*confidence intervals*, we need to create many*bootstrapped*samples.- Normally, the more
*bootstrapped*samples we use, the better the*confidence interval*. - In this lab, we’ll
`do()`

500*bootstrapped*samples.

- Normally, the more
- To make
`do()`

-ing 500*bootstraps*easier, we’ll code our 3-step bootstrap method into a function.- Open a new R script (File -> New File -> R Script) to write your function into.

- Fill in the blank space below with the 3-steps
needed to create a
*bootstrapped*sample`mean`

for our`atus`

data.- Each step should be written on its own line between the curly braces.

- Highlight and
*Run*the code you write.

- Once your function is created, fill in the blanks
to create 500
*bootstrapped*sample means:

**Create a histogram for your bootstrapped samples and describe the***center*,*shape*and*spread*of its distribution.- These bootstrapped estimates no longer estimate the average age of people in the U.S.
- Instead, they estimate how much the estimate of the average age of people in the U.S. varies.

- In the next slide, we’ll look at how we can use these bootstrapped
means to create
*90% confidence intervals*.

- To create a 90% confidence interval, we need to decide between which
two
*ages*the middle 90% of our bootstrapped estimates are contained. **Using your histogram, fill in the statement below:**

The lowest 5% of our estimates are below _______ years and the highest 5% of our estimates are above_______ years.

- Use the quantile() function to check your estimates.
**Based on your bootstrapped estimates, between which two ages are we 90% confident the actual mean age of people living in the U.S. is contained?**

- Using your
*bootstrapped*sample means, create a 95% confidence interval for the mean age of people living in the U.S. **Why is the 95% confidence interval wider than the 90% interval?****Write down how you would explain what a 95% confidence interval means to someone not taking***Introduction to Data Science*.