R help - Nov 2025 - Did behavior of "paired" argument in t.test() change?

Hi all,

Did the behavior of base t.test() subtly change with respect to the 
paired argument?

I think this code used to work with long form data, but now it seems not 
to? work:

 ? ? # Example long format data
 ? ? data_long <- data.frame(
 ? ? ? value = c(10, 12, 15, 8, 9, 11),
 ? ? ? time = factor(rep(c("before", "after"), each = 3))
 ? ? )

 ? ? t.test(value ~ time, data = data_long, paired = TRUE)

# Errorin t.test.formula(value ~ time, data = data_long, paired = TRUE) 
: # cannot use 'paired' in formula method

A quick Google seems to suggest that the above code should still work. 
What am I missing?

-------------------------------

Less sure about this one ...

Also, I thought it had always been possible to use x = "difference 
scores" in t.test(..., paired = TRUE) as long as y = NULL. My memory 
about this is less secure than the behavior above.

Control = c(10, 12, 15) Treat = c(8, 9, 11) data_wide <- data.frame( 
Control, Treat, Diff = Treat - Control) t.test(data_wide$Diff, NULL, 
"two.sided", 0, paired = FALSE) # works but y = NULL, mu = 0 mostly 
implies difference scores

t.test(data_wide$Diff, NULL, "two.sided", 0, paired = TRUE) # fails
but
if y = NULL the value of paired should not matter should it?

For y, help says "y an optional (non-empty) numeric vector of data 
values". I take this to imply that x = compared to a population mean 
hypothesis whenever y = NULL. Why should the value of paired influence 
the outcome?

Just bad memory? on my part?

R.version _ platform x86_64-w64-mingw32 arch x86_64 os mingw32 crt ucrt 
system x86_64, mingw32 status major 4 minor 5.1 year 2025 month 06 day 
13 svn rev 88306 language R version.string R version 4.5.1 (2025-06-13 
ucrt) nickname Great Square Root


-- 
---
Robert W. Baer, Ph.D.
Professor of Physiology
Kirksville College of Osteopathic Medicine
A.T. Still Univerisity of Health Sciences
800 W. Jefferson St.
Kirksville, MO 63501

	[[alternative HTML version deleted]]

Hallo

According to recent documentation this works for paired t.test.
t.test(Pair(value,time)~1, data=data_long)

It is the last example in help page.

And your second question seems to be also different from what help page
says:

If paired is TRUE then both x and y must be specified and they must be the
same length. Missing values are silently removed (in pairs if paired is TRUE
).

Cheers.
Petr


?t 4. 11. 2025 v 18:23 odes?latel Robert Baer via R-help <
r-help at r-project.org> napsal:
> Hi all,
>
> Did the behavior of base t.test() subtly change with respect to the
> paired argument?
>
> I think this code used to work with long form data, but now it seems not
> to  work:
>
>      # Example long format data
>      data_long <- data.frame(
>        value = c(10, 12, 15, 8, 9, 11),
>        time = factor(rep(c("before", "after"), each =
3))
>      )
>
>      t.test(value ~ time, data = data_long, paired = TRUE)
>
> # Errorin t.test.formula(value ~ time, data = data_long, paired = TRUE)
> : # cannot use 'paired' in formula method
>
> A quick Google seems to suggest that the above code should still work.
> What am I missing?
>
> -------------------------------
>
> Less sure about this one ...
>
> Also, I thought it had always been possible to use x = "difference
> scores" in t.test(..., paired = TRUE) as long as y = NULL. My memory
> about this is less secure than the behavior above.
>
> Control = c(10, 12, 15) Treat = c(8, 9, 11) data_wide <- data.frame(
> Control, Treat, Diff = Treat - Control) t.test(data_wide$Diff, NULL,
> "two.sided", 0, paired = FALSE) # works but y = NULL, mu = 0
mostly
> implies difference scores
>
> t.test(data_wide$Diff, NULL, "two.sided", 0, paired = TRUE) #
fails but
> if y = NULL the value of paired should not matter should it?
>
> For y, help says "y an optional (non-empty) numeric vector of data
> values". I take this to imply that x = compared to a population mean
> hypothesis whenever y = NULL. Why should the value of paired influence
> the outcome?
>
> Just bad memory  on my part?
>
> R.version _ platform x86_64-w64-mingw32 arch x86_64 os mingw32 crt ucrt
> system x86_64, mingw32 status major 4 minor 5.1 year 2025 month 06 day
> 13 svn rev 88306 language R version.string R version 4.5.1 (2025-06-13
> ucrt) nickname Great Square Root
>
>
> --
> ---
> Robert W. Baer, Ph.D.
> Professor of Physiology
> Kirksville College of Osteopathic Medicine
> A.T. Still Univerisity of Health Sciences
> 800 W. Jefferson St.
> Kirksville, MO 63501
>
>         [[alternative HTML version deleted]]
>
> ______________________________________________
> R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
> https://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>
	[[alternative HTML version deleted]]

? Tue, 4 Nov 2025 11:22:57 -0600
Robert Baer via R-help <r-help at r-project.org> ?????:
> Errorin t.test.formula(value ~ time, data = data_long, paired > TRUE) :
> cannot use 'paired' in formula method
This was changed in R-4.4.0:
https://bugs.r-project.org/show_bug.cgi?id=14359

The news item (see news(grepl('paired', Text))) recommends the use of
Pair(x1, x2) ~ 1 instead of the 'paired' argument. A further fix is
currently in R-devel (but not R-patched): without it, Pair(x1, x2) does
not handle the 'subset' argument.

-- 
Best regards,
Ivan