Combining data is one of those everyday tasks in R that looks simple until the two data frames refuse to fit together. Learning how to combine two dataframes in R means understanding three genuinely different operations: stacking rows, attaching columns, and matching records by a key. This guide walks through each approach with working code, explains when to use which, and covers the errors that trip up most beginners And that's really what it comes down to..
Introduction: Three Operations, Not One
New R users often search for "merge two data frames" when what they actually need is to glue rows underneath each other, or to place columns side by side. These are not interchangeable. Confusing them is the single most common source of silently wrong results That alone is useful..
The three operations are:
- Row binding (union/append) — data frame B goes underneath data frame A. Both must describe the same variables.
- Column binding (attach) — data frame B goes beside data frame A. Both must describe the same observations, in the same order.
- Joining (relational merge) — records from B are matched to records in A using a shared identifier such as
customer_idordate.
Base R gives you rbind(), cbind(), and merge(). That's why the dplyr package gives you a clearer, faster family of verbs: bind_rows(), bind_cols(), and the *_join() functions. The data.table package gives you the fastest options for large data Not complicated — just consistent..
Method 1: Stacking Rows with rbind() and bind_rows()
Use row binding when you have two files with the same structure — January sales and February sales, or survey responses from two sites.
Base R: rbind()
jan <- data.frame(
id = 1:3,
month = "January",
sales = c(120, 95, 143)
)
feb <- data.frame(
id = 4:6,
month = "February",
sales = c(88, 201, 176)
)
q1 <- rbind(jan, feb)
rbind() requires the same number of columns, and the column names must match. If the names are identical but ordered differently, R matches them by name. If any name differs, you get an error like names do not match previous names. Critically, rbind() never fills gaps — a missing column is an error, not an NA.
dplyr: bind_rows()
bind_rows() is more forgiving and usually the better choice:
library(dplyr)
mar <- data.frame