[Rsiena-help] Warnings "standard deviation is zero" in RSiena 1.1.-232 but not in RSiena 1.1-212

[Tom Snijders]

Dear Mark,

The error message is produced by a comparison of a variance with 0, and this of course may differ between systems depending on machine accuracy. I attach a file in which this comparison is replaced by a comparison with e-12 (scientific notation for 10^{-12}).
You can use this by putting this file in your working directory and then, after
library(RSienaTest)
giving the command
insertSource("phase3.r", package='RSienaTest')

Does the error then still occur?

Best wishes,
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

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 Mark Manger
Sent: 27 June 2014 16:55
To: rsiena-help at lists.r-forge.r-project.org
Subject: Re: [Rsiena-help] Warnings "standard deviation is zero" in RSiena 1.1.-232 but not in RSiena 1.1-212

Thanks for the quick reply. Tried that, but doesn't make any difference - I still get 16 warnings and phase 3 doesn't complete.

I wonder if I could produce the estimates with RSiena 1.1-212, and then update to 1.1-276 to use sienaRI()? Inconvenient, but better than nothing.

-Mark

On Jun 26, 2014, at 5:26 PM, Ruth Ripley <ruth at stats.ox.ac.uk<mailto:ruth at stats.ox.ac.uk>> wrote:


Dear Mark,

I am just guessing, but it could be that the message is coming from the
new Dolby option. Try setting Dolby off by dolby=FALSE in
sienaAlgorithmCreate (or sienaModelCreate).

Regards,

Ruth

On 26/06/2014 22:01, Mark Manger wrote:

Dear Tom, fellow RSiena users,

A small RSiena problem to resolve between World Cup matches:  I may have
stumbled upon a bug/version conflict/problem of sorts with RSiena. I'm
estimating model with two sets of one-mode networks.

The model can be estimated fine with R 2.15.3 and RSiena RSiena_1.1-212
on my Mac running Mavericks, including using the parallel package. I've
built RSiena_1.1-212 from source when downgrading to identify the
potential problem, but it definitely worked back in April last year with
the binary as well.

In the meantime I've upgraded to R 3.1.0 and RSiena 1.1-232. Alas, the
identical model now estimates fine until Phase 3 Iteration 1, then stops
with the following warnings (any number from 16 to 25 warnings are
generated):

Warning messages:
1: In cor(z$sf[, i], scores[, i]) : the standard deviation is zero
2: In cor(z$sf[, i], scores[, i]) : the standard deviation is zero
3: In cor(z$sf[, i], scores[, i]) : the standard deviation is zero
4: In cor(z$sf[, i], scores[, i]) : the standard deviation is zero
5: In cor(z$sf[, i], scores[, i]) : the standard deviation is zero
6: In cor(z$sf[, i], scores[, i]) : the standard deviation is zero
7: In cor(z$sf[, i], scores[, i]) : the standard deviation is zero
8: In cor(z$sf[, i], scores[, i]) : the standard deviation is zero
9: In cor(z$sf[, i], scores[, i]) : the standard deviation is zero
10: In cor(z$sf[, i], scores[, i]) : the standard deviation is zero
11: In cor(z$sf[, i], scores[, i]) : the standard deviation is zero
12: In cor(z$sf[, i], scores[, i]) : the standard deviation is zero
13: In cor(z$sf[, i], scores[, i]) : the standard deviation is zero
14: In cor(z$sf[, i], scores[, i]) : the standard deviation is zero
15: In cor(z$sf[, i], scores[, i]) : the standard deviation is zero
16: In cor(z$sf[, i], scores[, i]) : the standard deviation is zero

etc.

Updating RSiena to 1.1.276 changes nothing, i.e. I still get the
warnings. I also get this when estimating the model without any
covariates and three network effects.

First, I wonder if this should lead me to question the results obtained
with Siena 1.1-212. If not, I'd be fine just using the old version. The
only advantage of course would be that with 1.1-276 I could use sienaRI,
which looks extremely promising. Second, what do these warning tell me?
The .out file looks fine to me.

-Mark


*Mark S. Manger*
Assistant Professor of Political Economy and Global Affairs

Munk School of Global Affairs & Department of Political Science |
University of Toronto
Observatory Site | 315 Bloor Street West | Room 212
Toronto, ON   M5S 0A7
Phone: 416-946-8927 | Fax: 416-946-8877

mark.manger at utoronto.ca<mailto:mark.manger at utoronto.ca> <mailto:mark.manger at utoronto.ca>
*www.munkschool.utoronto.ca/mga<http://www.munkschool.utoronto.ca/mga><http://www.munkschool.utoronto.ca/mga>*

JOIN THE GLOBAL CONVERSATION






_______________________________________________
Rsiena-help mailing list
Rsiena-help at lists.r-forge.r-project.org<mailto:Rsiena-help at lists.r-forge.r-project.org>
https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rsiena-help
_______________________________________________
Rsiena-help mailing list
Rsiena-help at lists.r-forge.r-project.org<mailto:Rsiena-help at lists.r-forge.r-project.org>
https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rsiena-help

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/rsiena-help/attachments/20140707/ac40cd7f/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: phase3.r
Type: application/octet-stream
Size: 22289 bytes
Desc: phase3.r
URL: <http://lists.r-forge.r-project.org/pipermail/rsiena-help/attachments/20140707/ac40cd7f/attachment-0001.obj>