R help - Dec 2024 - Weird Behavior of mean

Is the following a strange behavior for `mean` vs. `sd` ?

```
$ R --vanilla. ## 4.4.2
> x=c(NA,1,2,3)
> c( mean(x,na.rm=T), sd(x,na.rm=T) )
[1] 2 1
> T=20   ## bad idea for a parameter.  T is also used for TRUE
> c( mean(x,na.rm=T), sd(x,na.rm=T) )
[1] NA  1
>
```

This one was a baffler for me to track down for a few hours...

I've not checked the code, but I think that result would happen if mean
uses something like

if (na.rm == TRUE) {
  # do something to remove the NA's
}

And as uses something like

If (na.rm != FALSE) {
  # do something to remove the NA's
}

Or perhaps ever na.rm == T


If you ever see posts from Bert on here with T and F, he is hard core
thorough and uses full words for exactly this reason, someone can reassign
F as True if they want and your code will melt!

On Fri, 13 Dec 2024, 08:31 Ivo Welch, <ivo.welch at gmail.com> wrote:
> Is the following a strange behavior for `mean` vs. `sd` ?
>
> ```
> $ R --vanilla. ## 4.4.2
> > x=c(NA,1,2,3)
> > c( mean(x,na.rm=T), sd(x,na.rm=T) )
> [1] 2 1
> > T=20   ## bad idea for a parameter.  T is also used for TRUE
> > c( mean(x,na.rm=T), sd(x,na.rm=T) )
> [1] NA  1
> >
> ```
>
> This one was a baffler for me to track down for a few hours...
>
> ______________________________________________
> 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]]

Ivo,

To be a bit clearer, your problem was not with mean(0 but with any function
which is expecting a Boolean setting of TRUE/FALSE and instead got a
nonsense value of 20.

This gets even weirder in languages which automagcally transform many other
values to be considered true. Languages like C consider many non-zero values
to be true, including 20 as an integer. A language like python accepts a
very wide variety of things as truthy and the few that qualify as falsy are
empty strings, and zero as well as other empties like an empty list, empty
dictionary, empty set, empty range, or even a complex number with both a
zero real and imaginary part. 

Some see such designs as saving lots of labor and others suggest it would be
better and clearer programming practice to only accept explicit code that is
guaranteed to produce a Boolean such as is_empty(list) or whatever.

In this case, it is simply fairly poor design to alias T/F to TRUE/FALSE for
convenience and then have people inadvertently reuse the variables and
produce odd results. In languages which support constants that can be set
and protected from change, fine. But in R, I never use T/F as the longer
versions are clearer and less error prone.

I don't blame you for being teed off at why things where not working when T
was redefined.

-----Original Message-----
From: R-help <r-help-bounces at r-project.org> On Behalf Of Ivo Welch
Sent: Thursday, December 12, 2024 7:23 PM
To: r-help at r-project.org
Subject: [R] Weird Behavior of mean

Is the following a strange behavior for `mean` vs. `sd` ?

```
$ R --vanilla. ## 4.4.2
> x=c(NA,1,2,3)
> c( mean(x,na.rm=T), sd(x,na.rm=T) )
[1] 2 1
> T=20   ## bad idea for a parameter.  T is also used for TRUE
> c( mean(x,na.rm=T), sd(x,na.rm=T) )
[1] NA  1
>
```

This one was a baffler for me to track down for a few hours...

______________________________________________
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.