Control Flow

Author

Affiliation

Julianne G. Clina

University of Kansas Medical Center

Published

September 2, 2026

Learning Objectives

By the end of this workshop, you should be able to:

Write if / else if / else statements to classify a single value
Use switch() to map a coded value to a label
Write a for loop to repeat an action across multiple patients
Use next to skip an iteration when data are missing

The core question to ask yourself throughout: what decision am I making, and for how many things at once? (if = one decision, one value at a time. for = repeat that decision across many values.)

## Creating example dataset for practice!
library(tidyverse)

patients <- tibble(
  id = sprintf("P%02d", 1:12),
  age = c(34, 58, 45, 62, 29, 51, 47, 39, 66, 55, 42, 60),
  sbp = c(118, 142, 135, 128, 122, 158, NA, 131, 145, 126, 138, 165),
  activity_code = c(
    "A",
    "C",
    "B",
    "A",
    "D",
    "B",
    "C",
    "A",
    "B",
    "D",
    "C",
    "A"
  ),
  steps = c(
    8200,
    3100,
    5400,
    9100,
    11200,
    4800,
    6200,
    8900,
    5100,
    12400,
    4300,
    6700
  )
)

patients

activity_code is a stand-in for how activity level was recorded in the field: "A" = sedentary, "B" = lightly active, "C" = moderately active, "D" = very active.

Activity 1: `if` / `else if` / `else`

Why control flow?

Subsetting (last workshop) answers “which rows meet a condition?” Control flow answers a different question: “given one value, which branch of logic should run?” This matters when the action you want to take is more complex than just keeping or dropping a row. e.g., assigning a category, printing a message, or triggering different calculations.

The basic shape:

# if (condition) {
# runs if condition is TRUE
#} else if (other_condition) {
# runs if condition is FALSE and other_condition is TRUE
#} else {
# runs if nothing above was TRUE
#}

A few things worth knowing:

if expects a single TRUE/FALSE — not a vector. Feeding it a vector of length > 1 will error (or has in recent R versions).
if/else returns a value, so you can assign the result directly: x <- if (a) 1 else 2.
This is different from ifelse(), which is vectorized and works element-by-element over a whole vector — useful later, but not what we’re practicing here.

Step 1: Classify one patient’s blood pressure

Standard categories (simplified for this exercise):

sbp < 120: “Normal”
sbp >= 120 & sbp < 130: “Elevated”
sbp >= 130 & sbp < 140: “High Stage 1”
sbp >= 140: “High Stage 2”

# Using patients$sbp[2], write an if / else if / else chain that prints the correct category

Step 2: Handle missing data

Patient 7 has a missing sbp value. Right now, running your code from Step 1 on patients$sbp[7] will likely throw an error or behave oddly, because if can’t evaluate NA as TRUE/FALSE.

# Add a check at the top of your if/else chain: if the value is NA,
# print "Missing" instead of evaluating the BP conditions

Why this matters for real data: any time you write if around a variable that might have missing values, you need to decide up front what should happen for NA — R won’t decide for you (it will error).

Activity 2: `switch()`

switch() is a cleaner way to write an if/else if chain when you’re matching one variable against a fixed set of known values, like a code.

switch(
  "B",
  A = "Sedentary",
  B = "Lightly active",
  C = "Moderately active",
  D = "Very active"
)

Compare this to what the equivalent if/else if chain would look like switch() says the same thing with a lot less repetition.

Step 1: Translate one code

# Use switch() to translate patients$activity_code[5] into its
# full label

Step 2: Add a fallback

switch() can take an unnamed final argument as a default, for values that don’t match anything listed:

switch(
  "Z",
  A = "Sedentary",
  B = "Lightly active",
  "Unknown code"
)

# Test your switch() from Step 1 on the value "Z" (a code that doesn't exist in our data) — add a default so it doesn't error

Activity 3: `for` loops

A for loop repeats the same block of code once per element of a sequence — useful when you want to apply an if/else decision (Activity 1) or a switch() (Activity 2) to every patient, not just one.

for (i in 1:5) {
  print(i)
}

i is just a variable name — it takes on each value in 1:5 in turn, and the code inside {} runs once per value.

Step 1: Loop through and classify every patient’s BP

Combine what you built in Activity 1 with a for loop: loop over all 12 patients, and print each patient’s id alongside their BP category. Don’t forget to handle NA (patient 7).

# for (i in 1:nrow(patients)) {
#   ... your if/else chain from Activity 1, using patients$sbp[i] ...
#   ... print(paste(patients$id[i], "-", category)) ...
# }

Step 2: Skip missing values with `next`

Instead of printing “Missing” for patient 7, use next to skip that iteration entirely — the loop moves on to the next patient without printing anything for the skipped one.

# Copy your loop from Step 1, but replace the "Missing" branch
# with: next

Step 3: Build a vector of results

Printing is fine for checking your work, but usually you want to save the result. Pre-allocate an empty character vector, then fill it in as the loop runs.

# bp_category <- vector("character", nrow(patients))
#
# for (i in 1:nrow(patients)) {
#   ... same if/else logic as before ...
#   bp_category[i] <- category   # or NA if missing
# }
#
# patients$bp_category <- bp_category
# patients

Key Takeaways

if/else makes a decision for a single value — it can’t take a vector.
switch() is a shortcut for matching one variable against a known list of codes, with an optional default.
for loops repeat a block of code once per element — this is how you scale a one-value decision (if/switch) up to a whole dataset.
next skips the rest of the current iteration and moves on — handy for skipping missing or bad data inside a loop.
Always ask: is this a single value (if/switch) or a whole vector/column at once (loop, or later: ifelse()/purrr::map())?