R help - Sep 2005 - trimmed mean in R seems to round the trimming fraction

subject: trimmed mean in R seems to round the trimming fraction
to r-help at stat.math.ethz.ch.

Consider the following example of 10 numbers.  10% trimmed mean is correct
but you can see that the result is the same for many trimming fractions
till 0.20!

For example 13% trimmed mean should use interpolation of second and
eighth ordered observation.  R does not seem to do this.
The correct 13% trimmed mean is the average of 7.4 middle observations
(28+46+50+52+54+63+82+64.4)/7.4
Answer=(439.4)/7.4= 59.3784

> z=c(23, 40,     46,      50,      52,      54,    63,    82,92,98)
> mean(z, trim=0.10)
[1] 59.875
> mean(z, trim=0.13) # wrong result is  59.875
[1] 59.875
> mean(z, trim=0.15)
[1] 59.875
> mean(z, trim=0.18)
[1] 59.875
> mean(z, trim=0.1999999)
[1] 59.875
> mean(z, trim=0.20)
[1] 57.83333
> mean(z, trim=0.40)
[1] 53


Hrishikesh D. Vinod
Professor of Economics, Fordham University
E-Mail: Vinod at fordham.edu  Tel 718-817-4065,
Secretary 718-817-4048, Fax 718-817-3518
Web page:  http://www.fordham.edu/economics/vinod

VINOD at FORDHAM.EDU wrote:
> 
> 
> subject: trimmed mean in R seems to round the trimming fraction
> to r-help at stat.math.ethz.ch.
> 
> Consider the following example of 10 numbers.  10% trimmed mean is correct
> but you can see that the result is the same for many trimming fractions
> till 0.20!
> 
> For example 13% trimmed mean should use interpolation of second and
> eighth ordered observation.  R does not seem to do this.
> The correct 13% trimmed mean is the average of 7.4 middle observations
> (28+46+50+52+54+63+82+64.4)/7.4
> Answer=(439.4)/7.4= 59.3784
> 
> 
> 
>>z=c(23, 40,     46,      50,      52,      54,    63,    82,92,98)
>>mean(z, trim=0.10)
> 
> [1] 59.875
> 
>>mean(z, trim=0.13) # wrong result is  59.875
> 
> [1] 59.875
> 
>>mean(z, trim=0.15)
> 
> [1] 59.875
> 
>>mean(z, trim=0.18)
> 
> [1] 59.875
> 
>>mean(z, trim=0.1999999)
> 
> [1] 59.875
> 
>>mean(z, trim=0.20)
> 
> [1] 57.83333
> 
>>mean(z, trim=0.40)
> 
> [1] 53
> 
> 
Hi,

The "trim" argument works as documented. Did you read ?mean:

      If 'trim' is non-zero, a symmetrically trimmed mean is computed
      with a fraction of 'trim' observations deleted from each end
      before the mean is computed.

Also, I don't recall ever learning your method for trimmed mean, but R 
is a programming language and you can always write your own trimmed mean 
function.

HTH,

--sundar