Guidelines to interpret Structure Harvester

[may.jes...@gmail.com]

Dear All,
Can any body help me by providing steps how to interpret Structure Harvester results.
I have seen several email conversation asking about suggestions to know groups. Several persons have replied but I could not see any one replying stepwise advise to determine number of groups in harvester results.
I have attached a pdf,

https://dl.dropboxusercontent.com/u/52154965/Group%28K2to10%29.pdf

Can any body use it as a reference and explain steps to determine number of groups.
Regards
May

[may.jes...@gmail.com]

Pl Help.
May

[Andrea Schreier]

Hi, May! Its difficult to help you interpret your data without
knowing anything about it. Can you provide some details? Also, did
you run your data with K=1? I think its important to test the
possibility of no structure even if you do expect it.

Andrea

> --
> You received this message because you are subscribed to the Google Groups "structure-software" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to structure-softw...@googlegroups.com.
> To post to this group, send email to structure...@googlegroups.com.
> Visit this group at http://groups.google.com/group/structure-software.
> For more options, visit https://groups.google.com/d/optout.

[Jessica May]

Thanks Andrea.

I had prior idea that there will be more than 2 groups.

In any case, i do not want to know groups in this particular case only, i have other analysis also, where i have results from run fromm1 and on different burnin and iterations.

 I wanted to know general rules or steps to interpret Structure Harvester Results.

regards

May

You received this message because you are subscribed to a topic in the Google Groups "structure-software" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/structure-software/tSbvyj7-e6w/unsubscribe.
To unsubscribe from this group and all its topics, send an email to structure-softw...@googlegroups.com.

[Jessica May]

[Jessica May]

Dear All,
Can any body help me by providing steps how to interpret Structure Harvester results.
I have seen several email conversation asking about suggestions to know groups. Several persons have replied but I could not see any one replying stepwise advise to determine number of groups in harvester results.
I have attached a pdf,

https://dl.dropboxusercontent.com/u/52154965/Group%28K2to10%29.pdf

Can any body use it as a reference and explain steps to determine number of groups.
Regards
May

[Andrea Schreier]

Hi, Jessica. When I work with Structure Harvester output, I first
examine the L(K) plot to see what K has the highest likelihood and the
lowest variance (in likelihood among runs). Then I examine the Delta
K plot and see if there is agreement. If there isn't agreement, I
think about the possible reasons why. (Delta K can't consider K=1 and
also tends to underestimate structure as in my experience it seems to
detect only the highest level of structure.) Harvester also makes
files ready to input into CLUMPP to test for multimodality and prepare
to make plots in Distruct.

Hope this is helpful!

Andrea