Playground: Chapter 2
Data Structures in R
This page practises the ideas of Chapter 2 in four parts. Practise the chapter works through the book’s exercises, with starter code, hints, and solutions. Go further goes beyond the book, with new questions and some of R’s built-in datasets. Check your understanding asks short questions whose answers open with a click. Do it yourself gives open tasks with an empty code space and no answers: you decide how to write the code, as you will in your own thesis.
In the exercises, replace each ______ with your own code and press Run Code. One exercise reads an SPSS file, which needs the haven package; it is marked as a task for the download project.
Practise the chapter
These are the exercises at the end of Chapter 2, with the same numbers.
Exercise 1: Ages over 30
Create a vector with the ages 24, 31, 28, 45, and 26. Use logical indexing to show only the ages over 30, and count how many there are.
A condition such as ages > 30 gives TRUE or FALSE for every value. Inside square brackets it keeps the TRUE values; inside sum() it counts them, because each TRUE counts as 1.
ages <- c(24, 31, 28, 45, 26)
ages[ages > 30]
sum(ages > 30)Two ages, 31 and 45, are over 30. The same condition did two jobs: it selected the values and it counted them. Selecting rows of a data frame, later in the chapter, works in exactly the same way.
Exercise 2: Levels in order
Create a factor from c("Agree", "Disagree", "Neutral", "Agree") whose levels run from "Disagree" through "Neutral" to "Agree". Check the order with table().
The levels argument takes a vector of the categories, written in the order you want.
answers <- factor(c("Agree", "Disagree", "Neutral", "Agree"),
levels = c("Disagree", "Neutral", "Agree"))
table(answers)The table now lists Disagree (1), Neutral (1), and Agree (2) in the order of the scale. Without the levels argument, R would sort the categories alphabetically, putting “Neutral” last, and every table and graph would show the scale in a meaningless order.
Exercise 3: Part-time students with children
Using students, count the students who are part-time and have children.
Both conditions must be true at the same time. The operator for “and” is a single symbol.
sum(students$study_mode == "Part-time" & students$has_children == "Yes")51 students are part-time and have children. With | (“or”) instead of &, the count would include every student who meets at least one of the two conditions, which is a different question.
Exercise 4: A new column
Add a column to students called over_30 that is TRUE for students older than 30. Count those students, and explain why the count might need na.rm = TRUE.
The new column is a comparison. Run the first sum() and look at the answer: one student’s age is missing.
students$over_30 <- students$age > 30
sum(students$over_30)
sum(students$over_30, na.rm = TRUE)205 students are older than 30. The first sum() returns NA because one student’s age is missing, and for that student R cannot tell whether the answer is TRUE or FALSE, so the comparison gives NA as well. The missing value passes from the age to the new column, and from there to anything calculated from it.
Exercise 5: A question behind a variable
Read the SPSS file with read_sav() and find the question behind support_3.
This exercise needs the haven package and the SPSS file, so it is done in the download project, where the file is included. The code is:
library(haven)
spss <- read_sav("wellbeing.sav")
attr(spss$support_3, "label")The label is “My supervisor cares about my wellbeing.” The SPSS file stores the wording of each question with its variable, which the CSV files cannot do; this is why the chapter keeps the labels when it imports SPSS data and uses them to start a codebook.
Exercise 6: Units of analysis
State the unit of analysis of each table in the package: students, semesters, questionnaire, and supervisors. For each, name one research question it could answer on its own. Write your answer first, then open the model answer.
In students, one row is a student (600 rows): for example, are part-time students more likely to consider dropping out than full-time students? In semesters, one row is one student in one semester (2,326 rows, up to four per student): does sleep change over the four semesters? In questionnaire, one row is again a student, with their answers to the 22 items: are stress and burnout scores related? In supervisors, one row is a supervisor (120 rows): do professors supervise more students than lecturers? A question about students should count each student once, so a question about students’ sleep needs one value per student, such as the first semester or an average, and not all 2,326 semester rows.
Go further
These exercises go beyond the book.
Exercise 7: A data frame of every kind
Build a small data frame of five invented students with one variable of each kind: a nominal category (faculty), an ordered category (how often they exercise), a number (age), and a logical value (whether they work). Then check the type of every column.
factor() has an argument that marks the levels as ordered, so that R knows Rarely < Sometimes < Often. The function that lists every column with its type is short for structure.
five <- data.frame(
faculty = factor(c("Science", "Education", "Science", "Engineering", "Education")),
exercise = factor(c("Rarely", "Often", "Sometimes", "Rarely", "Often"),
levels = c("Rarely", "Sometimes", "Often"),
ordered = TRUE),
age = c(26, 34, 29, 41, 25),
works = c(TRUE, FALSE, FALSE, TRUE, TRUE)
)
str(five)
five$exercise > "Rarely"str() shows a factor, an ordered factor (Ord.factor, with "Rarely" < "Sometimes" < "Often"), a number, and a logical column. Because the second factor is ordered, a comparison such as five$exercise > "Rarely" makes sense; for the faculty, which has no order, the same comparison would be meaningless, and R would warn you.
Exercise 8: The student with no age
Find the students whose age is missing, and show their ID, faculty, and programme.
The rows go before the comma, the columns after it. The function that answers “is this value missing?” with TRUE or FALSE is the one from Chapter 1.
students[is.na(students$age), c("student_id", "faculty", "programme")]One student, S0007, a Master’s student in Health Sciences, has no age. Looking at the rows with missing values, rather than only counting them, is the first step in deciding what to do about them: here, one student out of 600 changes nothing, but a pattern (for example, all missing ages in one faculty) would matter.
Exercise 9: Rows and students
The semesters table has more rows than there are students. Show the rows of student S0001, count all the rows, and count the different students.
unique() keeps each different value once, so the length of its result is the number of different students.
semesters[semesters$student_id == "S0001", ]
nrow(semesters)
length(unique(semesters$student_id))
table(semesters$semester)Student S0001 has four rows, one for each semester. The table has 2,326 rows but only 600 students, and the last line shows why the total is not 4 × 600 = 2,400: 37 students have no records for semesters 3 and 4, because they left the programme. Counting rows here would count semesters, not students, and would give students who stayed more weight than students who left.
Exercise 10: A codebook for a built-in dataset
R’s built-in ToothGrowth data records the length of tooth-growing cells in 60 guinea pigs given vitamin C in two ways and three doses. Build its codebook as a data frame, with the name, meaning, type, and possible values of each variable.
The chapter’s SPSS codebook used a function that applies another function, here class, to every column in turn and collects the results.
codebook <- data.frame(
variable = names(ToothGrowth),
meaning = c("Length of the tooth-growing cells (odontoblasts)",
"How vitamin C was given: orange juice (OJ) or ascorbic acid (VC)",
"Dose of vitamin C in milligrams per day"),
type = sapply(ToothGrowth, class),
values = c("4.2 to 33.9", "OJ, VC", "0.5, 1, 2"),
row.names = NULL
)
codebooklen and dose are numeric, and supp is a factor. The meanings come from the help page, ?ToothGrowth, and the values from range() and table(). Notice that dose is stored as a number although it takes only three values: whether to treat it as a number or as three groups is an analysis decision, and a codebook is a good place to record it.
Check your understanding
Answer each question in your own words first, then click to see a model answer.
1. Why does c(6.5, "seven") become a vector of text?
A vector holds values of one type only. Text cannot be turned into a number, but a number can always be written as text, so R converts 6.5 to "6.5" and the whole vector becomes character. One word in a column of numbers has the same effect when a file is read, which is why a numeric column that R reads as text usually hides a stray word or symbol.
2. What is the difference between x[2] and x[-2]?
x[2] keeps only the second value; x[-2] keeps everything except the second value. For x <- c(10, 20, 30), the first gives 20 and the second gives 10 and 30.
3. What are a factor’s levels for?
The levels list the possible categories and fix their order. The order controls how tables, graphs, and models show the categories, and it decides which category is the reference in a regression (Chapter 8). The levels also record categories that may have no cases in the data, which a plain text vector cannot do.
4. Why should a question about students count each student once, even when a table has several rows per student?
Because the unit of the question is the student. Counting rows would count semesters, giving students with more semesters more weight, and would treat four measurements of the same person as four independent people. Rows from the same student are related, and the methods of Chapter 10 are needed to analyse them together.
5. What does a codebook add to a data file?
The meaning of every value. A data file holds names and numbers; the codebook says what each variable measures, the question behind it, its units or answer scale, which numbers stand for which categories, and how missing values are coded. Without it, even the researcher cannot be sure what the data means a year later.
Do it yourself
These tasks have no starter code and no answers. Each code space below is empty and ready to run: write your own code, as you would for a thesis. The files students.csv and semesters.csv are available to read, the data frames students and semesters are already loaded, and R’s built-in datasets are always available.
1. Import the semester file, semesters.csv, yourself. Check its structure, and report how many rows, how many different students, and how many semesters it contains.
2. Choose three variables from students (for example, gender, employment, and financial_worry). Decide whether each should be a factor, an ordered factor, or a number, convert it, check the result with str(), and justify each choice in one sentence.
3. R’s built-in esoph data comes from a study of cancer of the oesophagus. Look at it with str() and head(), and read its help page with ?esoph. What does one row represent? How many people does the data describe? Which variables are ordered factors, and why?
4. This task is for the download project. Read the SPSS file, wellbeing.sav, build its codebook with the variable name, label, and type of every variable, add a column for notes, and save it as a CSV file with write.csv().
Work on your own computer
The project contains the data (as CSV, Excel, and SPSS files) and all four parts of this page as an R script: the exercises with blanks, the questions to check your understanding (answers in solutions.R), and the open tasks, each with space to write your code.
- Download chapter02.zip, unzip it, and double-click
chapter02.Rproj. - Or type this one line in RStudio’s Console:
usethis::use_course("https://polla-fattah.github.io/data2thesis_r/playground/chapter02.zip")