search

AndrewLJackson / SIBER

ellipse and convex hull fitting package for stable isotope data
36 stars 14 forks source link

siber.example <- createSiberObject(my data) sample size n<5 Warning message #88

Closed elvitaeglite closed 1 year ago

elvitaeglite commented 2 years ago

Hi Andrew,

I am applying SIBER to a small sample size. Mostly less than 10 samples, but in some cases less than R.

When I run siber.example <- createSiberObject(my datat) - I receive this message ' At least one of your groups has less than 5 observations. The absolute minimum sample size for each group is 3 in order for the various ellipses and corresponding metrics to be calculated. More reasonably though, a minimum of 5 data points are required to calculate the two means and the 2x2 covariance matrix and not run out of degrees of freedom. Check the item named ‘sample.sizes’ in the object returned by this function in order to locate the offending group. Bear in mind that NAs in the sample.size matrix simply indicate groups that are not present in that community, and is an acceptable data structure for these analyses.'

Can you please suggest me how to proceed with it? Thanks, Elvita

AndrewLJackson commented 1 year ago

you can still proceed, but how much trust you can put in your estimated statistics is highly questionable. At the very least the uncertainties will be so huge as to render any comparison with those groups impossible or non-significant. More data in those groups is the only real solution if you want to proceed with this type of analysis.