Dear list,
Indeed, there is no need adding error bars to boxplot. Points with error
bars as suggested will be quite better.
Many thanks to those who have advised me on this.
Regards
Ogbos
On 27 January 2011 12:29, Jim Lemon <jim@bitwrit.com.au> wrote:
> On 01/27/2011 05:04 AM, ogbos okike wrote:
>
>> Dear all,
>> I am trying to add error bars on a boxplot but have encountered an
error
>> as
>> indicated below. Is there a package I need to install or a library I
have
>> to
>> load before this goes please.
>> Thanks for any idea.
>> Ogbos
>>
>> x<-replicate(20,rnorm(50))
>> boxplot(x,notch=TRUE,main="Notched boxplot with error bars")
>> error.bars(x,add=TRUE)
>> Error: could not find function "error.bars"
>>
>
> Hi Ogbos,
> Adding "error bars" is easy, but as Josh pointed out, may not
mean much. If
> you are seeking to illustrate a confidence interval for the median, it may
> be better to use points or symbols and error bars rather than boxplots.
>
> Jim
>
>
[[alternative HTML version deleted]]