[Rsiena-help] Error in x$FRAN(zsmall, xsmall)

[Tom Snijders]

Dear Marco,

You write low Jaccard - but 0.8 is high. Turnover is low. High Jaccard should not be a problem.

You could try conditional estimation. If you have more than one dependent variable (network/behavior), try first to model variables one by one. If you have more than 2 waves, there might be nonconvergence due to time heterogeneity; if you have a converged estimation then you can study this by sienaTimeTest, else try estimation by successive pairs of waves.

Best,

Tom


============================================
Tom A.B. Snijders
Professor of Statistics in the Social Sciences
Department of Politics and Department of Statistics
Nuffield College
University of Oxford
tel. +44-01865-278599
http://stat.gamma.rug.nl/snijders/

________________________________
From: rsiena-help-bounces at lists.r-forge.r-project.org [mailto:rsiena-help-bounces at lists.r-forge.r-project.org] On Behalf Of marco tonellato
Sent: 05 May 2010 15:30
To: rsiena-help at lists.r-forge.r-project.org
Subject: [Rsiena-help] Error in x$FRAN(zsmall, xsmall)

Dear Ruth,

for the first time I've tried modelling some networks (83 x 577 nodes) with low Jaccard coefficient (around 0.8 - 1) and the estimation stops during phase 2 subphase 1 giving me this error message:

Error in x$FRAN(zsmall, xsmall) : Unlikely to terminate this epoch:

Is this likely to be correlated to the fact that there is a low Jaccard? Otherwise, what is it due to?

I'm using RSiena rev 83 on my machine (Linux  2.6.31-21-generic #59-Ubuntu SMP x86_64) using the patch you mentioned in the previous thread.

Thanks a lot.
Best,

Marco
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/rsiena-help/attachments/20100505/9b2bdfd7/attachment.htm>