Troubleshooting Common R Errors

1Objectives¶

By the end of this lesson you will be able to:

1. Recognize the eight R errors and warnings students learning R for statistics hit most often.

2. Diagnose the cause behind each message, not just the symptom.

3. Fix each one with a small, concrete change to your code.

4. Read any future R error the same systematic way: what does R say is wrong, and where.

2Errors are not a sign you did something wrong¶

Every R user — instructors included — sees these same eight messages constantly. An error is R telling you exactly where it got stuck, which makes it the single best debugging clue you have, not a signal to start over. Every message on this page is real R 4.5.2 output, produced by actually running the broken code shown — not retyped from memory — so what you see here is exactly what you’ll see on your own screen.

suppressMessages({library(mosaic); library(BSDA)})
set.seed(2200)
survey <- read.csv("data/survey_sim.csv")

3The eight errors, at a glance¶

Each is worked through below with the real broken code, the real message, and a real fix.

41. object '...' not found¶

mean(sleep_hors)

Error : object 'sleep_hors' not found

Cause: a typo — sleep_hors instead of sleep_hours — used as a bare name with no $ and no data = . R has no idea where to look for a column called sleep_hors, so it looks in your regular workspace, doesn’t find it there either, and stops.

Fix: spell the column name correctly and reach into the data frame with $ (or use it inside a formula with data = , Section 4’s territory):

mean(survey$sleep_hours)

[1] 6.7875

52. could not find function "..."¶

favstats(~sleep_hours, data = survey)

Error in favstats(~sleep_hours, data = survey) : 
  could not find function "favstats"

Cause: favstats() lives in the mosaic package (L05), and this session never ran library(mosaic) — maybe the kernel restarted, or the library() line further up the notebook was never executed.

Fix: run library(mosaic) (and library(BSDA)) once at the top of your session, before any line that uses a function from either package:

library(mosaic)
favstats(~sleep_hours, data = survey)

 min  Q1 median    Q3 max   mean       sd   n missing
 3.9 6.1    6.8 7.425 9.1 6.7875 1.012241 120       0

63. non-numeric argument to binary operator¶

survey$sleep_hours + survey$pet

Error in survey$sleep_hours + survey$pet : 
  non-numeric argument to binary operator

Cause: + (and every other arithmetic operator) only works between numbers. sleep_hours is numeric, but pet is text (“Cat person”, “Dog person”, “Neither”) — R has no defined meaning for 6.7 + "Cat person", so it refuses outright rather than guess.

Fix: this usually means you reached for the wrong tool — you likely wanted to compare sleep_hours across the groups in pet, which is exactly what the mosaic formula interface is for (L05–L07), not raw arithmetic between the two columns:

favstats(sleep_hours ~ pet, data = survey)

         pet min    Q1 median   Q3 max     mean        sd  n missing
1 Cat person 4.2 6.075   6.65 7.20 9.1 6.687500 1.0268667 40       0
2 Dog person 3.9 6.050   6.80 7.55 8.8 6.763636 1.0660352 55       0
3    Neither 5.3 6.600   7.00 7.50 8.6 7.000000 0.8631338 25       0

74. Wrong data = (right column name, wrong data frame)¶

mini <- survey[, c("student_id", "pet", "class_year")]
favstats(exam_score ~ pet, data = mini)

Error in eval(x, data, env) : object 'exam_score' not found

Cause: the same object '...' not found message as error #1, but from a different mistake — exam_score is spelled correctly, but mini (the data frame actually passed to data = ) never had that column to begin with; it was dropped by the [, c(...)] selection above.

Fix: point data = at a data frame that actually contains every column your formula names — here, the original survey:

favstats(exam_score ~ pet, data = survey)

         pet  min   Q1 median    Q3  max     mean        sd  n missing
1 Cat person 39.7 56.3  63.05 68.35 80.7 61.87000  9.356917 40       0
2 Dog person 40.0 58.0  62.20 68.45 86.1 63.17273  9.320357 55       0
3    Neither 42.2 49.8  55.60 60.20 84.1 57.47200 10.719084 25       0

85. $ vs. formula: passing a vector where a function wants ~¶

gf_histogram(survey$sleep_hours)

Error : Invalid formula type for gf_histogram.

Cause: every gf_* plotting function (L07) expects a formula (~sleep_hours) plus a separate data = argument — not a plain vector pulled out with $. survey$sleep_hours is valid R (it’s the exact syntax error #1’s fix used), just not the shape gf_histogram() is built to accept.

Fix: switch back to the formula + data = pattern every mosaic and ggformula function shares:

gf_histogram(~sleep_hours, data = survey)

Plots without error — the identical picture as L07’s sleep-hours histogram, because it’s the identical underlying data.

96. Factor/character vs. numeric: a column that imported as text¶

Say one cell of a numeric-looking survey column got typed as "5 stars" instead of 5 before the file was saved. Simulate exactly that and re-read it, the same way any CSV comes in (L06):

tmp <- survey
tmp$rating_txt <- as.character(1:120 %% 5 + 1)   # a made-up 1-5 rating column
tmp$rating_txt[7] <- "5 stars"                    # one contaminated cell
write.csv(tmp, "data/contaminated_sim.csv", row.names = FALSE)
contaminated <- read.csv("data/contaminated_sim.csv")

read.csv() doesn’t know which cell is the “wrong” one — it just sees that the column isn’t consistently numeric, so it imports the entire column as text:

class(contaminated$rating_txt)

[1] "character"

mean(contaminated$rating_txt)

[1] NA
Warning message:
In mean.default(x, ..., na.rm = na.rm) :
  argument is not numeric or logical: returning NA

Cause: mean() needs numbers; a character column — even one that’s mostly digits — is not numeric to R until you explicitly convert it, and one stray non-numeric cell is enough to force the whole column to import as text.

Fix: find the offending cell(s), fix or remove them, then convert:

bad_rows <- contaminated[is.na(suppressWarnings(as.numeric(contaminated$rating_txt))), ]
bad_rows[, c("student_id", "rating_txt")]

  student_id rating_txt
7       S007    5 stars

contaminated$rating_txt[7] <- "5"
contaminated$rating <- as.numeric(contaminated$rating_txt)
mean(contaminated$rating)

[1] 3.016667

107. cannot open the connection / file not found¶

read.csv("data/servey_sim.csv")

Error in file(file, "rt") : cannot open the connection
In addition: Warning message:
In file(file, "rt") :
  cannot open file 'data/servey_sim.csv': No such file or directory

Cause: a typo (servey instead of survey) — or, just as often, a correctly spelled file that doesn’t exist relative to your current working directory, the rule L03 sets up and L06 applies to every CSV import.

Fix: list.files() shows you the truth about what’s actually there before you guess again:

list.files("data", pattern = "survey")

[1] "make_survey_sim.R" "survey_sim.csv"   

dim(read.csv("data/survey_sim.csv"))

[1] 120   9

118. there is no package called '...'¶

library(moasic)

Error in library(moasic) : there is no package called 'moasic'

Cause: a typo in the package name (moasic instead of mosaic) — or a correctly spelled package that genuinely isn’t installed on this machine yet. R can only load a package it can find by that exact name.

Fix: check the spelling first (this guide uses exactly two package names throughout, mosaic and BSDA — see L05); if it’s really missing, install it once with install.packages("mosaic") (L05), then load it:

library(mosaic)

Loads silently — on JupyterHub, mosaic and BSDA are pre-installed (see L05), so install.packages() is almost never needed there; it’s mainly a local-install step.

12A general strategy for any error you haven’t seen yet¶

1. Read the error text itself, top to bottom — R almost always names the exact function and often the exact argument that failed.

2. Check spelling of every variable, column, and package name involved — the single most common cause across all eight errors above.

3. Check types with class() — most of the errors in this lesson boil down to R having a different type of object than the code assumed.

4. Check data = — confirm the data frame you passed actually contains every column your formula names.

5. Restart and re-run top to bottom (L03) if something that worked before mysteriously stops working — a forgotten library() call after a kernel restart (error #2) is the most common reason.

13Summary¶

• All eight errors above are things everyone learning R this way hits, not a sign you’re doing something unusually wrong.

• object '...' not found can mean a typo’d bare name (#1) or a formula whose data = doesn’t hold that column (#4) — same message, different fix, so read the code, not just the text.

• could not find function (#2) and there is no package called (#8) both trace back to library() — a session that never ran it, or a package name that’s misspelled or genuinely not installed.

• non-numeric argument (#3) and the mean() warning in #6 both come from R’s type system: arithmetic needs numbers, and one contaminated cell in a CSV column is enough to make read.csv() import the whole column as text — class() catches both immediately.

• $ pulls a plain vector out of a data frame; gf_* (and every other mosaic function) wants a ~ formula plus data = instead (#5) — they are not interchangeable even though both are valid R.

• list.files() (#7) turns a guess about a file path into a fact.

• The full script that reproduces every error and every fix on this page — by actually running the broken code — is committed at data/make_L14_examples.R; run it yourself and you’ll see the exact same messages.