[R] interpretation of stress in NMDS

[Thomas Petzoldt]

On 7/22/2010 2:21 PM, Graves, Gregory wrote:
> Among those users of Primer, stress values greater than 0.3 are
> interpreted as "questionable".  Using both isoMDS and metaMDS (vegan
> package), the stress values returned are much higher using my own data
> and using examples provided in R Help.  For example Rstress = 8.3, and
> the stressplot r2 = 0.99 indicating (to me) that the ordination is OK.
> I am guessing that the "stress" values across packages are not the same,
> and googling about has not returned a satisfactory answer ... thus this
> posting.  My concern being that reporting a stress value of 8 for what I
> consider a satisfactory result may raise a few Primer-user's eyebrows.

Dear Gregory,

as the help file of ?isoMDS tells us:

"stress   The final stress achieved (in percent)"

So 8.3 is in reality 0.083.

Thomas Petzoldt