Standardized income

Using the NLSY data and mutate(), make a standardized (centered at the mean, and divided by the standard deviation) version of income.

# replace the ... with your code
nlsy <- mutate(nlsy, income_stand = ...)

Standardized log(income)

Do the same thing, but using income on the log scale. Look at this variable using summary(). Can you figure out what happened? (Hint: look at your log(income) variable.)

nlsy <- mutate(nlsy, log_income_stand = ...)
summary(nlsy$log_income_stand)

Recode a factor

Turn the eyesight variable into a factor variable. The numbers 1-5 correspond to “excellent”, “very good”, “good”, “fair”, and “poor.” Make sure that categories are in an appropriate order.

nlsy <- mutate(nlsy, eyesight_fact = ...)

Combining factor levels

Use two different methods to combine the worst two categories of eyesight into one category.

Select centered variables

Create mean-centered versions of “age_bir”, “nsibs”, “income”, and the two sleep variables. Use the same ending (e.g., "_cent") for all of them. Then make a new dataset of just the centered variables using select() and a helper.

nlsy <- mutate(nlsy,
               age_bir_cent = ...,
               nsibs_cent = ...,
               ...)
nlsy_new <- select(nlsy, ...)

Go back to the beginning

You may have added a lot of variables to the original dataset by now. Create a dataset called nlsy_orig that contains only the variables we started off with, using the vector of names we originally used to name the columns and the all_of() helper. I’ll start you off with the variable names.

colnames_orig <- c("glasses", "eyesight", "sleep_wkdy", "sleep_wknd",
                   "id", "nsibs", "samp", "race_eth", "sex", "region", 
                   "income", "res_1980", "res_2002", "age_bir")

“Or” conditions

Create a dataset with all the observations that get over 7 hours of sleep on both weekends and weekdays or who have an income greater than/equal to 20,000 and less than/equal to 50,000.

nlsy_or <- filter(nlsy, ...)

Missing values

Create a dataset that consists only of the missing values in slp_cat_wkdy. Check how many rows it has (there should be 3!).

case_when()

Try again to make a new variable for standardized log-income. This time, if you are not able to calculate log(income) for an observation, replace it with a missing value (using case_when()). This time, when you standardize log(income), you’ll have to use na.rm = TRUE to remove missing values both when you take the mean and the standard deviation.

nlsy <- mutate(nlsy,
               log_income_stand = case_when(
                 ...
               ))

Relevel a factor

Make a new categorical income variable with at least 3 levels (you can choose the cutoffs). Make a bar graph with this new variable where the bars are in the correct order from low to high.

Rename variables

Look at help(rename). Looking at the examples to help, rename “age_bir” to “age_1st_birth” without making a new column.

Greater than/less than

Look up the between() function in help. Figure out how to use this to create a dataset with all the observations that get over 7 hours of sleep on both weekends and weekdays or who have an income greater than/equal to 20,000 and less than/equal to 50,000. Check to make sure you get the same number of rows as we did earlier.

Challenges

1. Figure out how to use the num_range() function to select the res_ variables from 1980 to 2000. There should only be one. Then try from 1980 to 2002. There should be two.

2. Download and save the full nlsy dataset into your 03-week folder. Then run the following lines to read it in.

nlsy_full <- read_rds("nlsy.rds")
colnames(nlsy_full) <- colnames_orig

You’ll see if you run levels(nlsy_full$res_1980) and levels(nlsy_full$res_2002), the participants’ residences in those years, what the levels of those factors are:

Imagine that you are doing an analysis with these data. Make a new factor variable for each year, that has the same, concise and readable levels, so that you could make a nice table comparing the distribution of residences across years.