In mattblackwell/qsslearnr: Learnr tutorials based on Quantitative Social Science

library(gradethis)
library(learnr)
library(qsslearnr)
tutorial_options(exercise.checker = gradethis::grade_learnr)
knitr::opts_chunk$set(echo = FALSE)
tut_reptitle <- "QSS Tutorial 2: Output Report"
data(resume, package = "qss")

Conceptual Questions

question('Which of the following approaches to identifying causal relationships is considered the "gold standard" in many scientific disciplines?',
              answer("Randomized controlled trials"),
              answer("Randomized experiments"),
              answer("Either of these", correct = TRUE)
         )

For ethical and logistical reasons, social scientists are often unable to conduct randomized controlled trials (RCTs). Therefore, they must conduct observational studies in which data on naturally occurring events are collected and analyzed.

question("With observational studies, it is often hard to establish that changes in one variable caused changes in another variable. In other words, observational studies have less __________ compared to RCTS.",
         answer("Internal validity", correct = TRUE),
         answer("External validity"),
         answer("Generalizability"),
         answer("All of these"))

More Logicals in R

Complex relationals

In this exercise, you have the ages of a sample of 15 people, stored in the ages vector. We can use these relational operators to create a logical vector which indicates which ages fall within a specific range. In particular, we can find out which respondents are college-aged (18-22).

Exercise

• Create a logical vector, called college.aged, which indicates which observations in ages are greater than or equal to 18 and less than or equal to 22. Be sure to use parentheses to separate out the two logical statements.
• Take the sum of the college.aged vector to determine how many 18-22 year olds there are in the sample.

## check the value of the ages vector
ages <- c(31, 20, 43, 45, 41, 46, 28, 49, 61, 19, 39)

## create a logical vector called college.aged
## that is TRUE for someone between 18-22, inclusive


## find the number of college.aged respondents

college.aged <- (ages >= 18) & (ages <= 22)
sum(college.aged)

grade_result_strict(
  pass_if(~ identical(college.aged, (ages >= 18) & (ages <= 22))),
  pass_if(~ identical(.result, sum((ages >= 18) & (ages <= 22))))
)

Subsetting based on logicals

In the last exercise, you used logical statements to create a vector that told us whether each entry in the ages vector is in the 18-22 year-old range. We can now use that information to figure out what the actual ages of the respondents in that range are.

Exercise

• Use the brackets and the college.aged logical vector that has already been created to subset ages to the value only between 18 and 22, inclusive.
• Use the mean function to calculate the average age of the respondents in this subset. To do this, use the bracket subsetting and the college.aged logical vector that has already been created.

ages <- c(31, 20, 43, 45, 41, 46, 28, 49, 61, 19, 39)

## here's the college.aged logical vector
college.aged <- (ages >= 18) & (ages <= 22)

## calculate the average age among the college-aged in the sample

mean(ages[college.aged])

grade_result(
  pass_if(~ identical(.result, mean(ages[college.aged])))
)

Using simple conditional statements

For this exercise, we'll use the resume data once again.

What if we wanted to create a new vector that depends on whether a statement is true or false? For example, suppose you wanted to create an indicator variable for whether or not a specific resume had the name "Carrie." From, the last few examples, you know that resume$firstname == "Carrie" will give you a vector of TRUE and FALSE values based on whether or not the name for that unit is "Carrie." We can then use this to get create a new variable using the ifelse(X, Y, Z) command. This command takes a logical vector as X and returns a new vector of the same length as X that has the value Y if that value in X is TRUE and Z if that value in X is FALSE.

Exercise

• Use the ifelse function to create a new variable called carrie that is 1 if the resume name (firstname) is "Carrie" and 0 otherwise.
• Print the first six lines of resume using the head function to see the new variable.

## create a new variable called carrie
resume$carrie <-

## print the first 6 lines of the updated resume

## create a new variable called carrie
resume$carrie <- ifelse(???, 1, 0)

## print the first 6 lines of the updated resume
head(resume)

## create a new variable called carrie
resume$carrie <- ifelse(resume$firstname == "Carrie", 1, 0)

## print the first 6 lines of the updated resume
head(resume)

grade_result_strict(
  pass_if(~ identical(resume$carrie, ifelse(resume$firstname == "Carrie", 1, 0)))
)

Factors variables in R

Factor variables

You have seen that creating subsets can be helpful for calculating different quantities or statistics for specific subgroups in the data. When there is more than 1 or 2 subgroups of interest, however, this can be a cumbersome process. For that reason, it's helpful to know about factor variables. Basically, a factor variable is a categorical variable that takes a finite number of distinct values.

Any variable can be turned into a factor by calling the as.factor() function like so:

mydata$myvar <- as.factor(mydata$myvar)

This will take the variable myvar and create a factor variable with levels that are observed in that variable. Most often, you will convert a character variable to a factor.

Exercise

• Finish the code below that creates the type character variable. Fill in the last values of race and sex and add the label WhiteMale to this last type.
• Convert the resume$type variable to a factor variable using as.factor().

resume$carrie <- ifelse(resume$firstname == "Carrie", 1, 0)

## fill in the last line of code to create a character vector for the type of
## application that was sent
resume$type <- NA
resume$type[resume$race == "black" & resume$sex == "female"] <- "BlackFemale"
resume$type[resume$race == "black" & resume$sex == "male"] <- "BlackMale"
resume$type[resume$race == "white" & resume$sex == "female"] <- "WhiteFemale"
resume$type[resume$race == ??? & resume$sex == ???] <-

## turn the character vector into a factor
resume$type <-

## fill in the last line of code to create a character vector for the type of
## application that was sent
resume$type <- NA
resume$type[resume$race == "black" & resume$sex == "female"] <- "BlackFemale"
resume$type[resume$race == "black" & resume$sex == "male"] <- "BlackMale"
resume$type[resume$race == "white" & resume$sex == "female"] <- "WhiteFemale"
resume$type[resume$race == ??? & resume$sex == ???] <- "WhiteMale"

## turn the character vector into a factor
resume$type <- as.factor(...)

## fill in the last line of code to create a character vector for the type of
## application that was sent
resume$type <- NA
resume$type[resume$race == "black" & resume$sex == "female"] <- "BlackFemale"
resume$type[resume$race == "black" & resume$sex == "male"] <- "BlackMale"
resume$type[resume$race == "white" & resume$sex == "female"] <- "WhiteFemale"
resume$type[resume$race == "white" & resume$sex == "male"] <- "WhiteMale"

## turn the character vector into a factor
resume$type <- as.factor(resume$type)

grade_code("Fantastic, you got that factor loaded up and ready to go. Now, let's see what you can do with it.")

Using factors

Imagine that you wanted to calculate the average callback for each level of type. You could create a subset for each level of type and then use mean on each one of those subsets. But that would take 8 lines of code!

A more efficient way to do this task would be to use the tapply(X, INDEX, FUN) function, which allows you to compute a function (FUN) on subsets of the data (X) defined by a factor variable (INDEX). For instance, suppose we had a grades data frame that had student exam grades out of 100 in the exam variable and a factor variable called section that reported which section they were enrolled in. Then we could calculate the average exam score within sections as:

tapply(grades$exam, grades$section, mean)

Exercise

• Use the table() function on the type variable to see how many fictitious applications were sent out with each type of name.

resume$carrie <- ifelse(resume$firstname == "Carrie", 1, 0)
resume$type <- NA
resume$type[resume$race == "black" & resume$sex == "female"] <- "BlackFemale"
resume$type[resume$race == "black" & resume$sex == "male"] <- "BlackMale"
resume$type[resume$race == "white" & resume$sex == "female"] <- "WhiteFemale"
resume$type[resume$race == "white" & resume$sex == "male"] <- "WhiteMale"

## turn the character vector into a factor
resume$type <- as.factor(resume$type)

## get the number of observations for each level of the type variable

table(resume$type)

grade_code()

• Use the tapply() function to calculate the mean of the call variable in each level of type.

## use the `tapply` function to calculate the mean in each level of type

## use the `tapply` function to calculate the mean in each level of type
tapply(resume$call, resume$type, mean)

grade_code("Great work, you have the skills you need to analyze experiments and observational data!")

Submit

submission_ui

submission_server()

mattblackwell/qsslearnr documentation built on Sept. 17, 2022, 6:25 p.m.