`augment` error with `na.action = na.exclude` in `glm`

[gavinsimpson]

The problem

augment throws an error for a GLM fitted with glm() where there are NAs in the data when data is supplied and na.action = na.exclude, contrary to the documented behaviour.

Reproducible example

# from ?glm
library("broom")
clotting <- data.frame(
    u = c(5,10,15,20,30,40,60,80,100),
    lot1 = c(118,58,42,35,27,25,21,19,18),
    lot2 = c(69,35,26,21,18,16,13,12,12))
# make one value of `u` and `lot1` missing
take <- withr::with_seed(2, sample(nrow(clotting), 2))
clotting[take[1], "u"] <- NA
clotting[take[2], "lot1"] <- NA
# fit GLM
m <- glm(lot1 ~ log(u), data = clotting, family = Gamma, na.action = na.exclude)
# na.action has worked
identical(length(fitted(m)), nrow(clotting))
#> [1] TRUE
# augment fails
augment(m, data = clotting)
#> Error in `$<-`:
#> ! Assigned data `unname(predict(x, newdata, type = type.predict))` must
#>   be compatible with existing data.
#> ✖ Existing data has 9 rows.
#> ✖ Assigned data has 7 rows.
#> ℹ Only vectors of size 1 are recycled.
#> Caused by error in `vectbl_recycle_rhs_rows()`:
#> ! Can't recycle input of size 7 to size 9.

^{Created on 2026-04-14 with reprex v2.1.1}

Session info

sessioninfo::session_info()
#> ─ Session info ───────────────────────────────────────────────────────────────
#>  setting  value
#>  version  R version 4.5.3 (2026-03-11)
#>  os       macOS Tahoe 26.3
#>  system   aarch64, darwin20
#>  ui       X11
#>  language (EN)
#>  collate  C.UTF-8
#>  ctype    C.UTF-8
#>  tz       Europe/Copenhagen
#>  date     2026-04-14
#>  pandoc   3.9.0.2 @ /opt/homebrew/bin/ (via rmarkdown)
#>  quarto   1.9.36 @ /Applications/quarto/bin/quarto
#> 
#> ─ Packages ───────────────────────────────────────────────────────────────────
#>  package     * version date (UTC) lib source
#>  backports     1.5.0   2024-05-23 [1] CRAN (R 4.5.0)
#>  broom       * 1.0.12  2026-01-27 [1] CRAN (R 4.5.2)
#>  cli           3.6.5   2025-04-23 [1] CRAN (R 4.5.0)
#>  digest        0.6.39  2025-11-19 [1] CRAN (R 4.5.2)
#>  dplyr         1.2.0   2026-02-03 [1] CRAN (R 4.5.2)
#>  evaluate      1.0.5   2025-08-27 [1] CRAN (R 4.5.0)
#>  fastmap       1.2.0   2024-05-15 [1] CRAN (R 4.5.0)
#>  fs            2.0.1   2026-03-24 [1] CRAN (R 4.5.2)
#>  generics      0.1.4   2025-05-09 [1] CRAN (R 4.5.0)
#>  glue          1.8.0   2024-09-30 [1] CRAN (R 4.5.0)
#>  htmltools     0.5.9   2025-12-04 [1] CRAN (R 4.5.2)
#>  knitr         1.51    2025-12-20 [1] CRAN (R 4.5.2)
#>  lifecycle     1.0.5   2026-01-08 [1] CRAN (R 4.5.2)
#>  magrittr      2.0.4   2025-09-12 [1] CRAN (R 4.5.0)
#>  otel          0.2.0   2025-08-29 [1] CRAN (R 4.5.0)
#>  pillar        1.11.1  2025-09-17 [1] CRAN (R 4.5.0)
#>  pkgconfig     2.0.3   2019-09-22 [1] CRAN (R 4.5.0)
#>  purrr         1.2.1   2026-01-09 [1] CRAN (R 4.5.2)
#>  R6            2.6.1   2025-02-15 [1] CRAN (R 4.5.0)
#>  reprex        2.1.1   2024-07-06 [1] CRAN (R 4.5.0)
#>  rlang         1.1.7   2026-01-09 [1] CRAN (R 4.5.2)
#>  rmarkdown     2.30    2025-09-28 [1] CRAN (R 4.5.0)
#>  sessioninfo   1.2.3   2025-02-05 [1] CRAN (R 4.5.0)
#>  tibble        3.3.1   2026-01-11 [1] CRAN (R 4.5.2)
#>  tidyr         1.3.2   2025-12-19 [1] CRAN (R 4.5.2)
#>  tidyselect    1.2.1   2024-03-11 [1] CRAN (R 4.5.0)
#>  vctrs         0.7.2   2026-03-21 [1] CRAN (R 4.5.2)
#>  withr         3.0.2   2024-10-28 [1] CRAN (R 4.5.0)
#>  xfun          0.57    2026-03-20 [1] CRAN (R 4.5.2)
#>  yaml          2.3.12  2025-12-10 [1] CRAN (R 4.5.2)
#> 
#>  [1] /Library/Frameworks/R.framework/Versions/4.5-arm64/Resources/library
#>  * ── Packages attached to the search path.
#> 
#> ──────────────────────────────────────────────────────────────────────────────

[EmilHvitfeldt]

this appears to be an issue regardless of the the value of the argument na.action

# from ?glm
library("broom")
clotting <- data.frame(
    u = c(5,10,15,20,30,40,60,80,100),
    lot1 = c(118,58,42,35,27,25,21,19,18),
    lot2 = c(69,35,26,21,18,16,13,12,12))
# make one value of `u` and `lot1` missing
take <- withr::with_seed(2, sample(nrow(clotting), 2))
clotting[take[1], "u"] <- NA
clotting[take[2], "lot1"] <- NA
# fit GLM
m <- glm(lot1 ~ log(u), data = clotting, family = Gamma)

augment(m, data = clotting)
#> Error in `$<-`:
#> ! Assigned data `unname(predict(x, newdata, type = type.predict))` must
#>   be compatible with existing data.
#> ✖ Existing data has 9 rows.
#> ✖ Assigned data has 7 rows.
#> ℹ Only vectors of size 1 are recycled.
#> Caused by error in `vectbl_recycle_rhs_rows()`:
#> ! Can't recycle input of size 7 to size 9.

^{Created on 2026-04-14 with reprex v2.1.1}

[gavinsimpson]

But I believe your example is as per documented behaviour; with na.action = na.omit (the default), internal to the model, fitted values and residuals components will have length 7 and information about where the NAs are is lost. This is incompatible with data, and documented behaviour. I think it would be quite complex to reliably infer where the NAs are if a user supplies data but fitted the model with na.omit, and such functionality could cause much confusion to casual users.

Ignore the rest of that (now deleted); I'd forgotten that predict() uses the $fitted.values component internally if newdata is missing.

[gavinsimpson]

The problem is in

broom/R/stats-glm.R

Lines 98 to 104 in 54e201b

	if (se_fit) {
	pred_obj <- predict(x, newdata, type = type.predict, se.fit = TRUE)
	df$.fitted <- pred_obj$fit |> unname()
	df$.se.fit <- pred_obj$se.fit |> unname()
	} else {
	df$.fitted <- predict(x, newdata, type = type.predict) |> unname()
	}

The default for newdata is NULl. predict() checks for missingness of newdata to trigger the use of object$fitted.values. Internally, predict() ends up calling predict.lm() with newdata = NULL, and again, while that function does know that there is no data, it still decides to napredict() or not depending on missingness of newdata. I don't know if this is a bug or infelicity in predict.lm, but absent R Core changing behaviour of predict in presence of missing newdata, passing newdata = NULL is going to be incompatible with using predict for models where na.exclude is used.