General Programming in R

Author

America Romero

Background

You’ve seen R before! We’ll cover the core building blocks — data structures, indexing, and functions.

Our Dataset: Simulated Survey

Throughout this tutorial, we’ll work with a simulated dataset from a fictional survey of One Direction fans. Participants reported their age, favorite member, how long they’ve been a fan, how many concerts they’ve attended, how much they spend on merch per year, and their self-identified stan level.

Why simulate data? You can use it to test your analysis code before collecting real data, and it helps you understand what your variables actually are.

set.seed(03252015)  # makes the random numbers reproducible, this is the date Mr. Zayn Malik left the group FYI. 

n <- 100  # number of participants

# Participant IDs
id <- 1:n

# Age: roughly 16-40
age <- round(rnorm(n, mean = 26, sd = 4))
age <- pmax(pmin(age, 40), 16)  # restrict to to [16, 40]

# Favorite member
favorite_member <- sample(
  c("Harry", "Niall", "Liam", "Louis", "Zayn"),
  size = n, replace = TRUE,
  prob = c(0.35, 0.20, 0.20, 0.15, 0.20)  # Harry most popular (sorry)
)

# Years as a fan
years_fan <- round(runif(n, min = 1, max = 13))

# Stan level: ordered — influences spending and concerts
stan_level <- sample(
  c("casual", "moderate", "hardcore"),
  size = n, replace = TRUE,
  prob = c(0.35, 0.45, 0.20)
)

# Concerts attended: hardcore fans go to more
concerts_attended <- ifelse(stan_level == "hardcore", sample(3:12, n, replace = TRUE),
                     ifelse(stan_level == "moderate", sample(1:5,  n, replace = TRUE),
                                                      sample(0:2,  n, replace = TRUE)))

# Merch spending per year (USD): correlated with stan level
merch_spending <- ifelse(stan_level == "hardcore", round(rnorm(n, mean = 300, sd = 80)),
                  ifelse(stan_level == "moderate", round(rnorm(n, mean = 120, sd = 40)),
                                                   round(rnorm(n, mean = 30,  sd = 15))))
merch_spending <- pmax(merch_spending, 0)  # no negative spending

# Still follows solo careers?
still_follows_solo <- ifelse(stan_level == "hardcore",
                       sample(c(TRUE, FALSE), n, replace = TRUE, prob = c(0.90, 0.10)),
                     ifelse(stan_level == "moderate",
                       sample(c(TRUE, FALSE), n, replace = TRUE, prob = c(0.65, 0.35)),
                       sample(c(TRUE, FALSE), n, replace = TRUE, prob = c(0.35, 0.65))))

# Put it all together
onedirection_fan_data <- data.frame(id, age, favorite_member, years_fan, stan_level,
                 concerts_attended, merch_spending, still_follows_solo)

Let’s take a quick look:

We can preview the first couple of rows of the data using head(). And we can look at the structure of the variables using str().

head(onedirection_fan_data)

  id age favorite_member years_fan stan_level concerts_attended merch_spending
1  1  25           Louis         4   hardcore                 7            407
2  2  27           Harry         7   moderate                 5             70
3  3  17           Harry         9   hardcore                10            336
4  4  22           Niall         7   hardcore                 9            256
5  5  25           Louis         5   hardcore                 4            271
6  6  29           Niall         2   moderate                 5            175
  still_follows_solo
1               TRUE
2              FALSE
3               TRUE
4               TRUE
5              FALSE
6              FALSE

str(onedirection_fan_data)

'data.frame':   100 obs. of  8 variables:
 $ id                : int  1 2 3 4 5 6 7 8 9 10 ...
 $ age               : num  25 27 17 22 25 29 23 29 22 31 ...
 $ favorite_member   : chr  "Louis" "Harry" "Harry" "Niall" ...
 $ years_fan         : num  4 7 9 7 5 2 1 6 7 6 ...
 $ stan_level        : chr  "hardcore" "moderate" "hardcore" "hardcore" ...
 $ concerts_attended : int  7 5 10 9 4 5 1 3 4 0 ...
 $ merch_spending    : num  407 70 336 256 271 175 142 157 226 36 ...
 $ still_follows_solo: logi  TRUE FALSE TRUE TRUE FALSE FALSE ...

Data Structures

R works with several types of objects. Knowing which type you’re dealing with is really really important!!

This table from Advanced R summarizes the basic data structures:

	Homogenous data	Heterogenous data
1-Dimensional	Atomic Vector	List
2-Dimensional	Matrix	Data frame
N-Dimensional	Array

Atomic Vectors

A vector is the simplest data structure in R: a sequence of values of the same type.

We can create simple vectors in this format, vector_name <- c(x, y, z).

logical_vec   <- c(TRUE, FALSE, TRUE)
integer_vec   <- c(1L, 2L, 3L)
double_vec    <- c(1.1, 2.5, 3.9)
character_vec <- c("Harry", "Niall", "Liam", "Louis", "Zayn")

Use typeof() to see the underlying type:

# Get type using typeof()
typeof(logical_vec)

[1] "logical"

typeof(onedirection_fan_data$still_follows_solo)

[1] "logical"

# integers are whole numbers, numeric vectors can hold decimals
typeof(integer_vec)

[1] "integer"

typeof(onedirection_fan_data$concerts_attended)

[1] "integer"

#does anyone use this?          
typeof(double_vec)

[1] "double"

typeof(character_vec)

[1] "character"

typeof(onedirection_fan_data$favorite_member)

[1] "character"

Use length() to count elements in our vectors:

length(character_vec)

[1] 5

length(merch_spending)

[1] 100

Relevance: Your condition labels, participant IDs, and response options are all vectors. Almost everything in R is built on top of them.

Factors

A factor is a special vector for categorical data. It stores data as integer codes with associated labels. Perfect for conditions, groups, or any nominal/ordinal variable.

# Convert favorite_member to a factor
typeof(onedirection_fan_data$favorite_member)

[1] "character"

onedirection_fan_data$favorite_member <- factor(onedirection_fan_data$favorite_member)

# Inspect it
levels(onedirection_fan_data$favorite_member)

[1] "Harry" "Liam"  "Louis" "Niall" "Zayn"

nlevels(onedirection_fan_data$favorite_member)

[1] 5

table(onedirection_fan_data$favorite_member)


Harry  Liam Louis Niall  Zayn 
   33    23    14    14    16

You can also create ordered factors for ordinal data (e.g., Likert scales):

# Convert stan_level to an ordered factor
onedirection_fan_data$stan_level <- factor(onedirection_fan_data$stan_level,
                                    levels = c("casual", "moderate", "hardcore"),
                                    ordered = TRUE)

head(onedirection_fan_data$stan_level)

[1] hardcore moderate hardcore hardcore hardcore moderate
Levels: casual < moderate < hardcore

levels(onedirection_fan_data$stan_level)

[1] "casual"   "moderate" "hardcore"

# Ordered factors support comparisons
onedirection_fan_data$stan_level[1] > onedirection_fan_data$stan_level[2]

[1] TRUE

Coercion: Converting Between Types

Sometimes R reads your data in as the wrong type. Numbers stored as characters (common in Qualtrics exports). You can force a conversion using as.numeric(), as.factor(), as.character(), etc.

# Character to numeric
x <- c("1", "2", "3")
class(x)

[1] "character"

x_num <- as.numeric(x)
class(x_num)

[1] "numeric"

# Character to factor
# Create a quick character vector and convert it
member_char <- c("Harry", "Niall", "Liam", "Louis", "Zayn")
class(member_char)

[1] "character"

member_factor <- as.factor(member_char)
class(member_factor)

[1] "factor"

levels(member_factor)

[1] "Harry" "Liam"  "Louis" "Niall" "Zayn"

# What happens when coercion fails?
as.numeric(c("1", "two", "3"))  # "two" becomes NA with a warning

Warning: NAs introduced by coercion

[1]  1 NA  3

Relevance: R needs to know a variable is a factor to treat it as categorical in statistical models. If you run a regression and your grouping column is stored as a character, R may not handle it the way you expect.

Lists

A list can hold elements of different types, even other lists. Unlike a vector where everything has to be the same type, a list can mix characters, numbers, logicals, vectors, and even other lists all in one object.

This makes lists useful for storing information that belongs together but doesn’t fit neatly into a data frame, like metadata about your study.

survey_info <- list(
  name       = "1D Fan Survey",  # character
  n_subjects = n,                # numeric
  members    = c("Harry", "Niall", "Liam", "Louis", "Zayn"),  # vector
  pilot      = FALSE             # logical
)

You can access elements two ways — with $ by name, or with [[]] by name or position:

survey_info$name          # using $

[1] "1D Fan Survey"

survey_info[["name"]]     # using [[]] by name — same result

[1] "1D Fan Survey"

survey_info[[1]]          # using [[]] by position

[1] "1D Fan Survey"

# You can also check the structure of the whole list
str(survey_info)

List of 4
 $ name      : chr "1D Fan Survey"
 $ n_subjects: num 100
 $ members   : chr [1:5] "Harry" "Niall" "Liam" "Louis" ...
 $ pilot     : logi FALSE

Note: A single [] on a list returns another list. [[]] returns the actual element inside.

Data Frames

A data frame is the most common structure for research data. It’s like a spreadsheet: each column is a vector (all the same type), and each row is an observation.

nrow(onedirection_fan_data)      # number of rows (participants)

[1] 100

ncol(onedirection_fan_data)      # number of columns (variables)

[1] 8

dim(onedirection_fan_data)       # both at once

[1] 100   8

names(onedirection_fan_data)     # column names

[1] "id"                 "age"                "favorite_member"   
[4] "years_fan"          "stan_level"         "concerts_attended" 
[7] "merch_spending"     "still_follows_solo"

summary(onedirection_fan_data)   # quick descriptive stats for each column

       id              age        favorite_member   years_fan        stan_level
 Min.   :  1.00   Min.   :17.00   Harry:33        Min.   : 1.00   casual  :38  
 1st Qu.: 25.75   1st Qu.:23.00   Liam :23        1st Qu.: 3.00   moderate:39  
 Median : 50.50   Median :25.00   Louis:14        Median : 7.00   hardcore:23  
 Mean   : 50.50   Mean   :25.40   Niall:14        Mean   : 6.51                
 3rd Qu.: 75.25   3rd Qu.:27.25   Zayn :16        3rd Qu.: 9.00                
 Max.   :100.00   Max.   :32.00                   Max.   :13.00                
 concerts_attended merch_spending  still_follows_solo
 Min.   : 0.00     Min.   :  0.0   Mode :logical     
 1st Qu.: 1.00     1st Qu.: 32.0   FALSE:44          
 Median : 2.00     Median : 88.0   TRUE :56          
 Mean   : 3.12     Mean   :127.3                     
 3rd Qu.: 5.00     3rd Qu.:184.0                     
 Max.   :12.00     Max.   :421.0

Indexing & Subsetting Data

Sometimes we’re only working with part of the whole dataset. We can quickly look through our data by indexing them.

Indexing Vectors

merch_spending[1]                      # first element

[1] 407

merch_spending[1:5]                    # elements 1 through 5

[1] 407  70 336 256 271

merch_spending[c(1, 3, 5)]             # elements 1, 3, and 5

[1] 407 336 271

merch_spending[merch_spending > 300]   # elements greater than 300

 [1] 407 336 316 370 339 421 385 304 311 350 409

Data Frames

Data frames are indexed with [rows, columns]:

onedirection_fan_data[1, ]            # first row, all columns

  id age favorite_member years_fan stan_level concerts_attended merch_spending
1  1  25           Louis         4   hardcore                 7            407
  still_follows_solo
1               TRUE

onedirection_fan_data[, "age"]        # age column, all rows (returns a vector)

  [1] 25 27 17 22 25 29 23 29 22 31 22 22 30 24 26 24 23 27 30 28 31 26 28 31 23
 [26] 22 25 27 26 24 27 26 25 22 26 21 28 31 26 27 21 26 32 22 30 26 30 23 22 21
 [51] 25 22 26 27 31 26 22 25 18 24 27 21 26 23 30 21 27 26 22 24 30 18 24 27 24
 [76] 23 22 25 32 29 30 26 25 26 24 21 28 29 18 21 32 25 29 23 32 25 23 25 23 27

onedirection_fan_data$merch_spending  # same as above using $ notation

  [1] 407  70 336 256 271 175 142 157 226  36 298 256 222 106  39 316 370 196
 [19] 134  40   9  32  30  19  22 142  88 135  31  29  44  19  62 154  95 339
 [37]  85  14  32   0 180 421 123 289  35  31  19 272  67  58   4 138  32 279
 [55] 385   4  23 304 311  24 350  42  54 107 178  22 141  17  33  14  86  27
 [73]  57 285  76 108  53  88 107 409  33 213  50   4  79 274 146  23  56 139
 [91] 298 135  16 143  90  49 162 116  18   0

onedirection_fan_data[1:5, c("id", "favorite_member", "stan_level", "merch_spending")]

  id favorite_member stan_level merch_spending
1  1           Louis   hardcore            407
2  2           Harry   moderate             70
3  3           Harry   hardcore            336
4  4           Niall   hardcore            256
5  5           Louis   hardcore            271

Tip: Leaving the row or column slot empty means “give me everything.”

Logical Subsetting

This is how you filter data, selecting rows that meet a condition:

# Only hardcore fans
hardcore <- onedirection_fan_data[onedirection_fan_data$stan_level == "hardcore", ]
nrow(hardcore)

[1] 23

# Fans who spend more than $200 on merch
big_spenders <- onedirection_fan_data[onedirection_fan_data$merch_spending > 200, ]
head(big_spenders)

   id age favorite_member years_fan stan_level concerts_attended merch_spending
1   1  25           Louis         4   hardcore                 7            407
3   3  17           Harry         9   hardcore                10            336
4   4  22           Niall         7   hardcore                 9            256
5   5  25           Louis         5   hardcore                 4            271
9   9  22            Zayn         7   hardcore                 4            226
11 11  22            Zayn        11   hardcore                 5            298
   still_follows_solo
1                TRUE
3                TRUE
4                TRUE
5               FALSE
9                TRUE
11               TRUE

# Harry fans who still follow solo careers
harry_solo <- onedirection_fan_data[onedirection_fan_data$favorite_member == "Harry" & onedirection_fan_data$still_follows_solo == TRUE, ]
nrow(harry_solo)

[1] 16

Working with Data

Reading in Data

Sometimes you’ll load in data from a CSV (e.g., exported from Qualtrics or a spreadsheet):

# Read a CSV from your working directory
my_data <- read.csv("https://www.dropbox.com/scl/fi/ol2shg9pbof0x2595in38/sim_onedirection_fan_data.csv?rlkey=9uuz318hhyfw253omdv696vao&dl=1")

write.csv(my_data, "example_data.csv", row.names = FALSE)

# Check where R is looking
getwd()

[1] "/Users/rchavez/Library/CloudStorage/GoogleDrive-robert.s.chavez@gmail.com/My Drive/Teaching/PSY_607_data_science/2026s/class_tutorials"

# Set a new working directory if needed
# setwd("/path/to/your/folder")
# Session > Set Working Directory > Choose Directory

Inspecting and Cleaning

We can’t have dirty data!

# Check for missing values
sum(is.na(my_data))

[1] 5

colSums(is.na(my_data))

                id                age    favorite_member          years_fan 
                 0                  2                  0                  1 
        stan_level  concerts_attended     merch_spending still_follows_solo 
                 0                  0                  1                  1

sum(is.na(onedirection_fan_data))

[1] 0

colSums(is.na(onedirection_fan_data))

                id                age    favorite_member          years_fan 
                 0                  0                  0                  0 
        stan_level  concerts_attended     merch_spending still_follows_solo 
                 0                  0                  0                  0

# Rename a column (base R approach)
names(onedirection_fan_data)[names(onedirection_fan_data) == "merch_spending"] <- "merch_usd"
names(onedirection_fan_data)

[1] "id"                 "age"                "favorite_member"   
[4] "years_fan"          "stan_level"         "concerts_attended" 
[7] "merch_usd"          "still_follows_solo"

# Rename back
names(onedirection_fan_data)[names(onedirection_fan_data) == "merch_usd"] <- "merch_spending"

Descriptive Statistics

Base R has all the tools you need for quick summaries.

mean(onedirection_fan_data$merch_spending)

[1] 127.31

median(onedirection_fan_data$merch_spending)

[1] 88

sd(onedirection_fan_data$merch_spending)

[1] 115.7802

var(onedirection_fan_data$merch_spending)

[1] 13405.04

range(onedirection_fan_data$merch_spending)

[1]   0 421

min(onedirection_fan_data$merch_spending)

[1] 0

max(onedirection_fan_data$merch_spending)

[1] 421

summary() gives you everything at once:

summary(onedirection_fan_data$merch_spending)

   Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
    0.0    32.0    88.0   127.3   184.0   421.0

summary(onedirection_fan_data$concerts_attended)

   Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
   0.00    1.00    2.00    3.12    5.00   12.00

Frequency tables with table():

table(onedirection_fan_data$favorite_member)


Harry  Liam Louis Niall  Zayn 
   33    23    14    14    16

table(onedirection_fan_data$favorite_member, onedirection_fan_data$stan_level)  # cross-tabulation

       
        casual moderate hardcore
  Harry     14       13        6
  Liam       9        8        6
  Louis      6        5        3
  Niall      4        7        3
  Zayn       5        6        5

Relevance: You’ll run these before any inferential statistics. They help you catch data entry errors, check distributions, and understand your sample.

The `apply` Family

The apply functions let you apply a function across rows, columns, or groups without writing a loop.

`apply()` — over rows or columns of a matrix or data frame

# Create a numeric-only subset
numeric_onedirection_fan_data <- onedirection_fan_data[, c("age", "years_fan", "concerts_attended", "merch_spending")]

# Column means
apply(numeric_onedirection_fan_data, MARGIN = 2, FUN = mean)

              age         years_fan concerts_attended    merch_spending 
            25.40              6.51              3.12            127.31

`tapply()` — apply a function by group

Summarize a variable by group.

# Mean merch spending by stan level
tapply(onedirection_fan_data$merch_spending, onedirection_fan_data$stan_level, mean)

   casual  moderate  hardcore 
 28.28947 114.92308 311.91304

# Mean concerts attended by favorite member
tapply(onedirection_fan_data$concerts_attended, onedirection_fan_data$favorite_member, mean)

   Harry     Liam    Louis    Niall     Zayn 
3.333333 2.391304 2.571429 4.214286 3.250000

`sapply()` — apply over a list or vector, simplify result

# Get the class of each column
sapply(onedirection_fan_data, class)

$id
[1] "integer"

$age
[1] "numeric"

$favorite_member
[1] "factor"

$years_fan
[1] "numeric"

$stan_level
[1] "ordered" "factor" 

$concerts_attended
[1] "integer"

$merch_spending
[1] "numeric"

$still_follows_solo
[1] "logical"

# Get descriptive stats for numeric columns
sapply(numeric_onedirection_fan_data, function(x) c(mean = mean(x), sd = sd(x)))

           age years_fan concerts_attended merch_spending
mean 25.400000  6.510000          3.120000       127.3100
sd    3.452418  3.534805          2.944881       115.7802

Conditional Logic & Loops

General form:

if (condition){ Do something } else { Do something different }

`if` / `else`

avg_spending <- mean(onedirection_fan_data$merch_spending, na.rm = TRUE)
avg_spending

[1] 127.31

if (avg_spending >= 200) {
  print("This is a high-spending fanbase")
} else if (avg_spending >= 75) {
  print("This is a moderate-spending fanbase")
} else {
  print("This is a low-spending fanbase")
}

[1] "This is a moderate-spending fanbase"

# if/else only works on one value at a time
# To apply logic across a whole column, use ifelse()

onedirection_fan_data$spender_type <- ifelse(
  onedirection_fan_data$merch_spending > 150, "big spender", "small spender"
)

table(onedirection_fan_data$spender_type)


  big spender small spender 
           31            69

`for` Loops

# Let's loop through each One Direction member
# and print how many fans in our survey picked them as their favorite

for (member in levels(onedirection_fan_data$favorite_member)) {
  
  # subset to just fans of this member
  fan_subset <- onedirection_fan_data[onedirection_fan_data$favorite_member == member, ]
  
  # count them
  n_fans <- nrow(fan_subset)
  
  cat(member, "has", n_fans, "fans in our sample\n")
}

Harry has 33 fans in our sample
Liam has 23 fans in our sample
Louis has 14 fans in our sample
Niall has 14 fans in our sample
Zayn has 16 fans in our sample

# Now let's add a little more — mean merch spending per member

for (member in levels(onedirection_fan_data$favorite_member)) {
  fan_subset <- onedirection_fan_data[onedirection_fan_data$favorite_member == member, ]
  avg_spend  <- round(mean(fan_subset$merch_spending, na.rm = TRUE), 2)
  cat(member, "fans spend an average of $", avg_spend, "on merch\n")
}

Harry fans spend an average of $ 122.73 on merch
Liam fans spend an average of $ 132.3 on merch
Louis fans spend an average of $ 122.43 on merch
Niall fans spend an average of $ 140.14 on merch
Zayn fans spend an average of $ 122.62 on merch

Tip: In R, apply functions can sometimes be faster for working with data frames. But there’s a million ways you can do the same thing in R. How cool!

Mini-hacks

Mini-hack 1:

Create a new column in onedirection_fan_data called concert_goer that evaluates if a participant has gone to 3 or more concerts.

onedirection_fan_data$concert_goer <- ifelse(
  onedirection_fan_data$concerts_attended >= 3, "TRUE", "FALSE"
)

table(onedirection_fan_data$concert_goer)


FALSE  TRUE 
   54    46

Mini-hack 2:

Use tapply() to find the mean number of concerts attended for each favorite member.

tapply(onedirection_fan_data$concerts_attended, onedirection_fan_data$favorite_member, mean)

   Harry     Liam    Louis    Niall     Zayn 
3.333333 2.391304 2.571429 4.214286 3.250000

Mini-hack 3:

Subset the data frame to include only fans aged 18–22. How many participants is that?

gen_z <- onedirection_fan_data[onedirection_fan_data$age >= 18 & onedirection_fan_data$age <= 22, ]
head(gen_z)

   id age favorite_member years_fan stan_level concerts_attended merch_spending
4   4  22           Niall         7   hardcore                 9            256
9   9  22            Zayn         7   hardcore                 4            226
11 11  22            Zayn        11   hardcore                 5            298
12 12  22            Zayn         5   hardcore                 6            256
26 26  22            Zayn         7   moderate                 3            142
34 34  22            Liam         8   moderate                 5            154
   still_follows_solo  spender_type concert_goer
4                TRUE   big spender         TRUE
9                TRUE   big spender         TRUE
11               TRUE   big spender         TRUE
12               TRUE   big spender         TRUE
26              FALSE small spender         TRUE
34               TRUE   big spender         TRUE

nrow(gen_z)

[1] 22

Mini-hack 4:

Use a for loop to print the proportion of fans who still follow solo careers for each stan level.

for (level in levels(onedirection_fan_data$stan_level)) {
  level_subset <- onedirection_fan_data[onedirection_fan_data$stan_level == level, ]
  prop <- round(mean(level_subset$still_follows_solo, na.rm = TRUE), 2)
  cat(level, "fans:", prop * 100, "% still follow solo careers\n")
}

casual fans: 24 % still follow solo careers
moderate fans: 64 % still follow solo careers
hardcore fans: 96 % still follow solo careers

Background

Our Dataset: Simulated Survey

Data Structures

Atomic Vectors

Factors

Coercion: Converting Between Types

Lists

Data Frames

Indexing & Subsetting Data

Indexing Vectors

Data Frames

Logical Subsetting

Working with Data

Reading in Data

Inspecting and Cleaning

Descriptive Statistics

The apply Family

apply() — over rows or columns of a matrix or data frame

tapply() — apply a function by group

sapply() — apply over a list or vector, simplify result

Conditional Logic & Loops

if / else

for Loops

Mini-hacks

Mini-hack 1:

Mini-hack 2:

Mini-hack 3:

Mini-hack 4:

The `apply` Family

`apply()` — over rows or columns of a matrix or data frame

`tapply()` — apply a function by group

`sapply()` — apply over a list or vector, simplify result

`if` / `else`

`for` Loops