14: Joins

Content for Wednesday, May 13, 2026

Before class

📖 Reading:

During class

We’ll cover:

  • Why join data? Psychology data often lives in multiple tables
  • Keys — primary keys and foreign keys
  • Mutating joins (add columns from another table):
    • left_join() — keep all rows from the left table
    • inner_join() — keep only matching rows
    • right_join() and full_join()
  • Filtering joins (use one table to filter another):
    • semi_join() — keep rows that have a match
    • anti_join() — keep rows that don’t have a match
  • Common issues: many-to-many, missing keys, duplicates
TipAssignment 7 is assigned today

Assignment 7: Joins & Missing Data — due Sunday, May 24 at 11:59 PM.

Slides

View slides in new tab Download PDF

Embedded slides

After class

✅ Practice:

  1. Create two small tibbles with a shared key column. Practice left_join(), inner_join(), and full_join() — what’s different about the output?
  2. Use anti_join() to find participants who are in one dataset but not another (a common quality check)
  3. Try joining by multiple keys using by = c("id", "timepoint")
  4. What happens when a key isn’t unique? Try it and see
  5. Think about your final project: do you need to combine multiple tables?
NoteWhich join do you need?

Most of the time, you want left_join() — it keeps everything in your main table and adds columns from the second table where there’s a match.

Join Keeps
left_join(x, y) All rows from x, matching rows from y
inner_join(x, y) Only rows that match in both
full_join(x, y) All rows from both tables
anti_join(x, y) Rows in x with no match in y

Psychology application: combining data sources

# Participant demographics + survey responses
demographics <- tibble(
  participant_id = c("P01", "P02", "P03"),
  age = c(19, 22, 20),
  gender = c("Woman", "Man", "Woman")
)

survey_scores <- tibble(
  participant_id = c("P01", "P02", "P04"),
  anxiety_score = c(34, 28, 41)
)

# Keep all participants, add scores where available
demographics |>
  left_join(survey_scores, by = "participant_id")

# Find who's missing survey data
demographics |>
  anti_join(survey_scores, by = "participant_id")