Indicator analysis result problem

[ann.berlin]

Hey there,

I am writing my bachelor thesis atm and today I found a problem in my indicator analysis.
So I'm trying to find out which species is predominantely found in one area. The areas are devided in 1 and 2.

Whenever I run the indicator analysis the results differ, even though I am using the same data. So sometimes the significant indicator distrubution is 1= 5 & 2 = 9, then it's 1 = 4 & 2 = 9 or 1 = 5 & 2 = 11

So my question is: is that normal? If so, why?

Thank you so much for reading and helping me!

All the Best from Berlin, Germany!

Here is my code:
Note: Arten is basically all vegetation data from each patch and TRArten are dry grassland species from each patch
Arten<-read.table("Artenliste.txt", header = TRUE, stringsAsFactors = TRUE, dec = ",", row.names = 1)
attach(Arten)
TRArten<-read.table("Trockenrasenarten.txt", header = TRUE, stringsAsFactors = TRUE, dec = ",", row.names = 1)
attach(TRArten)

library(labdsv)
Arten_on<-summary(indval(Arten, Age, numitr= 1000))
Trockenrasen_on<-summary(indval(TRArten, Age, numitr= 1000))

I'm not sure how to upload my data, so follow here: WeTransfer - Send Large Files & Share Photos Online - Up to 2GB Free

[nirgrahamuk]

@ann.berlin

So my question is: is that normal? If so, why?

You are using a function that involves randomizations. Perhaps a larger number might cause a stabilization. Regardless the classic way of controlling randomization in r for reproducibilty is to use set.seed()

numitr
the number of randomizations to iterate to calculate probabilities

[ann.berlin]

Hey!

Thank you for explaining! That must be it.
I used a permutation of 10.000 and it is more stable now.

Thank you from Berlin,
Ann