<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=us-ascii">
<meta name="Generator" content="Microsoft Exchange Server">
<!-- converted from rtf -->
<style><!-- .EmailQuote { margin-left: 1pt; padding-left: 4pt; border-left: #800000 2px solid; } --></style>
</head>
<body>
<font face="Times New Roman, serif" size="3">
<div>Dear Ruth,</div>
<div> </div>
<div>Across the three waves 22% of observations appear only in one wave. There is no opportunity for actors to enter between waves. Does this suggest Siena is not a good platform for these data?</div>
<div> </div>
<div> </div>
<div>My adjcommxxx objects are class matrix. I created them by following the example in the manual, namely:</div>
<div><font face="Courier New, monospace" size="2">#111TH COMMITTEES</font></div>
<div><font face="Courier New, monospace" size="2">temp1 <- as(comm111.full, "dgTMatrix")</font></div>
<div><font face="Courier New, monospace" size="2">edges111 <- cbind(temp1@i+1, temp1@j+1, temp1@x)</font></div>
<div><font face="Courier New, monospace" size="2">#CREATE EMPTY ADJACENCY MATRIX</font></div>
<div><font face="Courier New, monospace" size="2">adjcomm111 <- matrix(0, 555, 555)</font></div>
<div><font face="Courier New, monospace" size="2">#PUT EDGE VALUES IN DESIRED PLACES</font></div>
<div><font face="Courier New, monospace" size="2">adjcomm111[edges111[, 1:2]] <- edges111[,3]</font></div>
<div><font face="Courier New, monospace" size="2">#CHECK LENGTH</font></div>
<div><font face="Courier New, monospace" size="2">length(which(comm111.full !=adjcomm111))</font></div>
<div><font face="Courier New, monospace" size="2">#THIS RETURNS 0, WHICH IS CORRECT</font></div>
<div> </div>
<div>Best,</div>
<div>Jennifer</div>
<div>___________________________________________</div>
<div>Jennifer Nicoll Victor</div>
<div>Assistant Professor</div>
<div>Department of Political Science</div>
<div>University of Pittsburgh</div>
<div>4600 Wesley W. Posvar Hall</div>
<div>(412) 624-7204</div>
<div>E-mail: jnvictor@pitt.edu</div>
<div>Homepage: <a href="http://www.polisci.pitt.edu/person/jennifer-nicoll-victor">
http://www.polisci.pitt.edu/person/jennifer-nicoll-victor</a> </div>
<div> </div>
<div> </div>
<div>-----Original Message-----<br>
From: Ruth Ripley [<a href="mailto:ruth@stats.ox.ac.uk">mailto:ruth@stats.ox.ac.uk</a>]
<br>
Sent: Wednesday, March 21, 2012 4:29 PM<br>
To: Victor, Jennifer Nicoll<br>
Cc: rsiena-help@r-forge.wu-wien.ac.at<br>
Subject: Re: [Rsiena-help] What does this error mean?</div>
<div> </div>
<div>Dear Jennifer,</div>
<div> </div>
<div>10's will work correctly in RSiena, but they will not work in sna.</div>
<div> </div>
<div>There is little point having actors who are only present at time 3. If </div>
<div>you assume they arrive at time 2.5 they could be used in the simulation.</div>
<div> </div>
<div>Similarly you might want to assume that actor 3 leaves at some time </div>
<div>after 2.0.</div>
<div> </div>
<div>What format are your adjcommxxx? Try class(adjcommxxx) to find out. They </div>
<div>should be dgTMatrix.</div>
<div> </div>
<div>Regards,</div>
<div> </div>
<div>Ruth</div>
<div> </div>
<div>On 21/03/2012 20:14, Victor, Jennifer Nicoll wrote:</div>
<div>> Thanks, Ruth. As indicated in the manual, I have the structural zeroes in RSiena data set to "10," rather than "NA." This is correct, yes?</div>
<div>></div>
<div>> Could my trouble have to do with the exogenous events file? I specified this file as the manual instructed and then used the command:</div>
<div>> compChange.us<-sienaCompositionChange(usexevents)</div>
<div>></div>
<div>> Where my "usexevents" file looks like this:</div>
<div>> V2 V3</div>
<div>> 1 1 3</div>
<div>> 2 1 3</div>
<div>> 3 1 2</div>
<div>> 4 3 3</div>
<div>></div>
<div>> (for the first four actors, where actors 1& 2 are present in all three time periods, and actor 2 is present only in the first two time periods, and actor 4 is present only in the 3 time period).</div>
<div>></div>
<div>> I received no errors with my sienaCompositionChange command, but I wonder if there is a problem?</div>
<div>></div>
<div>> As an aside, I also have trouble with the command:</div>
<div>> uscom<-varDyadCovar(list(adjcomm110, adjcomm111))</div>
<div>> Error in varDyadCovar(list(adjcomm110, adjcomm111)) :</div>
<div>> not a list of sparse triples matrices</div>
<div>></div>
<div>> My "adjcommxxx" files are sparse matrices, but I'm not certain about the "triples" part.</div>
<div>></div>
<div>> JNV</div>
<div>> ___________________________________________</div>
<div>> Jennifer Nicoll Victor</div>
<div>> Assistant Professor</div>
<div>> Department of Political Science</div>
<div>> University of Pittsburgh</div>
<div>> 4600 Wesley W. Posvar Hall</div>
<div>> (412) 624-7204</div>
<div>> E-mail: jnvictor@pitt.edu</div>
<div>> Homepage: <a href="http://www.polisci.pitt.edu/person/jennifer-nicoll-victor">
http://www.polisci.pitt.edu/person/jennifer-nicoll-victor</a></div>
<div>></div>
<div>></div>
<div>> -----Original Message-----</div>
<div>> From: Ruth Ripley [<a href="mailto:ruth@stats.ox.ac.uk">mailto:ruth@stats.ox.ac.uk</a>]</div>
<div>> Sent: Wednesday, March 21, 2012 3:14 PM</div>
<div>> To: Victor, Jennifer Nicoll</div>
<div>> Cc: rsiena-help@r-forge.wu-wien.ac.at</div>
<div>> Subject: Re: [Rsiena-help] What does this error mean?</div>
<div>></div>
<div>> Dear Jennifer,</div>
<div>></div>
<div>> The error you have indicates that the simulation process is failing.</div>
<div>> Either steps are happening so frequently that the cumulative time is</div>
<div>> increasing too slowly, or, more likely here, in conditional estimation,</div>
<div>> every step is choosing to do nothing so the target cannot be reached.</div>
<div>></div>
<div>> One point about sna: having seen some of your data, I wonder if you had</div>
<div>> any structural zeros in the matrix you used with sna. sna does not</div>
<div>> recognise these.</div>
<div>></div>
<div>> Regards,</div>
<div>></div>
<div>> Ruth</div>
<div>></div>
<div>> On 21/03/2012 15:38, Victor, Jennifer Nicoll wrote:</div>
<div>>> Thank you to Ruth Ripley and Nate Doogan for providing help with</div>
<div>>> this</div>
<div>> modeling problem. Based on their suggestions I have simplified the model</div>
<div>> and the effects; however, I still am unable to estimate the model. I'm</div>
<div>> seeking some general advice about fitting my data. Ruth wondered if</div>
<div>> Siena was the appropriate model for these data and I'm attempting to</div>
<div>> investigate this question and get more feedback. I appreciate further</div>
<div>> feedback from users of this list. I've also posted to the general</div>
<div>> Statnet list and have estimated ERGMs with these data.</div>
<div>>></div>
<div>>> I have encountered numerous problems relating to degeneracy and</div>
<div>>> model</div>
<div>> specification. The data are from three waves (congresses) of US members</div>
<div>> of Congress and their co-membership in legislative organizations. I have</div>
<div>> a number of dyadic and nodal attribute variables as well about the</div>
<div>> actors. The hypothesis I seek to test is whether there are bridging-ties</div>
<div>> among the legislators through their memberships in legislative</div>
<div>> organizations. I suspect that the high degree of density in the data may</div>
<div>> be part of the problem.</div>
<div>>></div>
<div>>> One approach I've tried is to simply calculate betweenness (in the</div>
<div>>> SNA</div>
<div>> package) for each network of legislators in each time period, add this</div>
<div>> as a vertex attribute, and estimate an ERGM, in which betweenness is a</div>
<div>> nodal-covariate; however these models have difficulty with convergence</div>
<div>> and fit. I am unable to obtain MCMC standard errors, and even after</div>
<div>> numerous attempts at respecification the diagnostics suggest that fit</div>
<div>> may be problematic.</div>
<div>>></div>
<div>>> Another approach is to use Siena. Siena should allow me to take</div>
<div>> advantage of the longitudinal nature of the data. But the output that</div>
<div>> I've been able to obtain suggests that the model is problematic, as Ruth</div>
<div>> and Nate have pointed out. See output below.</div>
<div>>></div>
<div>>> So my various attempts at testing this hypothesis using these</div>
<div>>> advanced</div>
<div>> methods have left me frustrated. While we can (and have) tested the idea</div>
<div>> using a variety of less sophisticated methods, it seems to me that the</div>
<div>> advanced statistical methods, such as those available through ERGM and</div>
<div>> Siena, are designed for exactly this type of modeling; however, I have</div>
<div>> not been able to achieve satisfactory findings that are reportable. I</div>
<div>> can continue to try to reparameterize the models to search for fit and</div>
<div>> to avoid degeneracy, but I would like to hear thoughts of readers of</div>
<div>> this list regarding:</div>
<div>>> - the appropriate method for testing the hypothesis - best practices</div>
<div>>> for iterative modeling and specification for these</div>
<div>> types of models (given that each iteration is so computationally expensive)</div>
<div>>> - which architecture (ERGM, Siena, something else?) seems most</div>
<div>> appropriate for the data?</div>
<div>>></div>
<div>>> Siena verbose output:</div>
<div>>>> myeffus</div>
<div>>> effectName include fix test initialValue parm</div>
<div>>> 1 constant caucuses rate (period 1) TRUE FALSE FALSE 100.00000 0</div>
<div>>> 2 constant caucuses rate (period 2) TRUE FALSE FALSE 70.41182 0</div>
<div>>> 3 degree (density) TRUE FALSE FALSE 0.03424 0</div>
<div>>> 4 betweenness TRUE FALSE FALSE 0.00000 0</div>
<div>>> 5 same usparty TRUE FALSE FALSE 0.00000 0</div>
<div>>>> myusmodel<-sienaModelCreate(useStdInits=FALSE, projname='siena_us_02')</div>
<div>>>> usans02<-siena07(myusmodel, data=myusdata,effects=myeffus, batch=TRUE, verbose=TRUE)</div>
<div>>></div>
<div>>> Stochastic approximation algorithm.</div>
<div>>> Initial value for gain parameter = 0.2.</div>
<div>>> Start of the algorithm.</div>
<div>>> Observed function values are</div>
<div>>> 1. 123349.0000 2. 5156392.0000 3. 123580.0000</div>
<div>>></div>
<div>>> Start phase 0</div>
<div>>> theta: 0.0342 0.0000 0.0000</div>
<div>>> Current parameter values:</div>
<div>>> 0.03424475 0.00000000 0.00000000</div>
<div>>></div>
<div>>> Start phase 1</div>
<div>>> Phase 1 Iteration 1 Progress: 0%</div>
<div>>> Phase 1 Iteration 2 Progress: 0%</div>
<div>>> Phase 1 Iteration 3 Progress: 0%</div>
<div>>> Phase 1 Iteration 4 Progress: 0%</div>
<div>>> Phase 1 Iteration 5 Progress: 0%</div>
<div>>> Phase 1 Iteration 10 Progress: 0%</div>
<div>>> Phase 1 Iteration 15 Progress: 1%</div>
<div>>> Time per iteration in phase 1 = 50.7582</div>
<div>>> Average deviations NR generated statistics and targets</div>
<div>>> after phase 1:</div>
<div>>> -39637.750000</div>
<div>>> 6787139.500000</div>
<div>>> -39249.250000</div>
<div>>></div>
<div>>> Diagonal values of derivative matrix :</div>
<div>>> 32772.3901 524963179.7438 38364.4002</div>
<div>>> dfra :</div>
<div>>> 32772.39 3087677.74 17924.11</div>
<div>>> 5095442.50 524963179.74 3357687.99</div>
<div>>> 39393.49 3556965.56 38364.40</div>
<div>>></div>
<div>>> inverse of dfra :</div>
<div>>> 0.00029660366044 -0.00000197940840 0.00003466438151</div>
<div>>> -0.00000228738847 0.00000001994553 -0.00000067696765</div>
<div>>> -0.00009248392723 0.00000018324891 0.00005323684804</div>
<div>>></div>
<div>>> Full Quasi-Newton-Raphson step after phase 1:</div>
<div>>> 1. 26.551774</div>
<div>>> 2. -0.252611</div>
<div>>> 3. -2.820084</div>
<div>>> This step is multiplied by the factor 0.10000.</div>
<div>>> Intervention 1.4.2: jump after phase 1 decreased by factor 26.5517736817607 .</div>
<div>>> Phase 1 achieved after 16 iterations.</div>
<div>>> theta: 1.03424 -0.00951 -0.10621</div>
<div>>> Current parameter values:</div>
<div>>> 1.034244749 -0.009513884 -0.106210769</div>
<div>>></div>
<div>>> Phase 2 has 4 subphases.</div>
<div>>> Each subphase can be repeated up to 4 times</div>
<div>>></div>
<div>>> Start phase 2.1</div>
<div>>> Phase 2 Subphase 1 Iteration 1 Progress: 3%</div>
<div>>> Phase 2 Subphase 1 Iteration 2 Progress: 3%</div>
<div>>> theta 1.0433 -0.0096 -0.0982</div>
<div>>> ac 1.009 0.985 1.014</div>
<div>>> Phase 2 Subphase 1 Iteration 3 Progress: 3%</div>
<div>>> Phase 2 Subphase 1 Iteration 4 Progress: 3%</div>
<div>>> theta 1.06134 -0.00977 -0.08233</div>
<div>>> ac 1.006 0.992 1.008</div>
<div>>> Phase 2 Subphase 1 Iteration 5 Progress: 3%</div>
<div>>> Phase 2 Subphase 1 Iteration 6 Progress: 3%</div>
<div>>> theta 1.07940 -0.00994 -0.06651</div>
<div>>> ac 1.01 0.99 1.01</div>
<div>>> Phase 2 Subphase 1 Iteration 7 Progress: 3%</div>
<div>>> Phase 2 Subphase 1 Iteration 8 Progress: 3%</div>
<div>>> theta 1.0975 -0.0101 -0.0508</div>
<div>>> ac 1.004 0.995 1.007</div>
<div>>> Phase 2 Subphase 1 Iteration 9 Progress: 3%</div>
<div>>> Phase 2 Subphase 1 Iteration 10 Progress: 3%</div>
<div>>> theta 1.1155 -0.0103 -0.0353</div>
<div>>> ac 1.004 0.996 1.006</div>
<div>>> Phase 2 Subphase 1 Iteration 20 Progress: 3%</div>
<div>>> theta 1.2058 -0.0112 0.0411</div>
<div>>> ac 1.002 0.999 1.006</div>
<div>>> Phase 2 Subphase 1 Iteration 30 Progress: 4%</div>
<div>>> theta 1.2961 -0.0121 0.1146</div>
<div>>> ac 1.00 1.00 1.01</div>
<div>>> Phase 2 Subphase 1 Iteration 40 Progress: 4%</div>
<div>>> theta 1.386 -0.013 0.185</div>
<div>>> ac 1.00 1.00 1.01</div>
<div>>> Phase 2 Subphase 1 Iteration 50 Progress: 5%</div>
<div>>> theta 1.4767 -0.0139 0.2526</div>
<div>>> ac 1.00 1.00 1.01</div>
<div>>> Phase 2 Subphase 1 Iteration 60 Progress: 5%</div>
<div>>> theta 1.5670 -0.0148 0.3167</div>
<div>>> ac 1.00 1.00 1.01</div>
<div>>> Phase 2 Subphase 1 Iteration 70 Progress: 5%</div>
<div>>> theta 1.6574 -0.0157 0.3777</div>
<div>>> ac 1.00 1.00 1.01</div>
<div>>> Phase 2 Subphase 1 Iteration 80 Progress: 6%</div>
<div>>> theta 1.7477 -0.0166 0.4350</div>
<div>>> ac 1.00 1.00 1.01</div>
<div>>> Phase 2 Subphase 1 Iteration 90 Progress: 6%</div>
<div>>> theta 1.8380 -0.0174 0.4888</div>
<div>>> ac 1.00 1.00 1.01</div>
<div>>> Phase 2 Subphase 1 Iteration 100 Progress: 7%</div>
<div>>> theta 1.9283 -0.0182 0.5386</div>
<div>>> ac 1.00 1.00 1.01</div>
<div>>> Phase 2 Subphase 1 Iteration 110 Progress: 7%</div>
<div>>> theta 2.019 -0.019 0.585</div>
<div>>> ac 1.00 1.00 1.01</div>
<div>>> Error in x$FRAN(zsmall, xsmall) :</div>
<div>>> Unlikely to terminate this epoch: more than 1000000 steps</div>
<div>>> Calls: siena07 ... proc2subphase -> doIterations -> <Anonymous> -> .Call</div>
<div>>> Execution halted</div>
<div>>></div>
<div>>></div>
<div>>> If you got to the end of this post, I appreciate your attention and any advice you might have!</div>
<div>>> Best,</div>
<div>>> Jennifer Victor</div>
<div>>> ___________________________________________</div>
<div>>> Jennifer Nicoll Victor</div>
<div>>> Assistant Professor</div>
<div>>> Department of Political Science</div>
<div>>> University of Pittsburgh</div>
<div>>> 4600 Wesley W. Posvar Hall</div>
<div>>> (412) 624-7204</div>
<div>>> E-mail: jnvictor@pitt.edu</div>
<div>>> Homepage: <a href="http://www.polisci.pitt.edu/person/jennifer-nicoll-victor">
http://www.polisci.pitt.edu/person/jennifer-nicoll-victor</a></div>
<div>>></div>
<div>>></div>
<div>>> -----Original Message-----</div>
<div>>> From: rsiena-help-bounces@r-forge.wu-wien.ac.at [<a href="mailto:rsiena-help-bounces@r-forge.wu-wien.ac.at">mailto:rsiena-help-bounces@r-forge.wu-wien.ac.at</a>] On Behalf Of Ruth Ripley</div>
<div>>> Sent: Friday, March 09, 2012 8:39 AM</div>
<div>>> To: rsiena-help@r-forge.wu-wien.ac.at</div>
<div>>> Subject: Re: [Rsiena-help] What does this error mean?</div>
<div>>></div>
<div>>> Dear Jennifer,</div>
<div>>></div>
<div>>> The errors both indicate that you are fitting a model which is too</div>
<div>>> complicated for your data. I suggest you start with just the default</div>
<div>>> effects and add others one by one to see which one is giving the problem.</div>
<div>>></div>
<div>>> Regards,</div>
<div>>></div>
<div>>> Ruth</div>
<div>>></div>
<div>>> On 09/03/2012 02:45, Victor, Jennifer Nicoll wrote:</div>
<div>>>> After letting my RSiena model for 12 hours on a high memory machine I received the following output. What advice can you provide to help me overcome this error?</div>
<div>>>></div>
<div>>>>> myusmodel<-sienaModelCreate(useStdInits=FALSE, projname='siena_us_01')</div>
<div>>>>> usans01<-siena07(myusmodel, data=myusdata,effects=myeffus)</div>
<div>>>> No X11 device available, forcing use of batch mode</div>
<div>>>> Start phase 0</div>
<div>>>> theta: 0.0342 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000</div>
<div>>>></div>
<div>>>> Start phase 1</div>
<div>>>> Phase 1 Iteration 1 Progress: 0%</div>
<div>>>> Phase 1 Iteration 2 Progress: 0%</div>
<div>>>> Phase 1 Iteration 3 Progress: 0%</div>
<div>>>> Phase 1 Iteration 4 Progress: 0%</div>
<div>>>> Phase 1 Iteration 5 Progress: 0%</div>
<div>>>> Phase 1 Iteration 10 Progress: 0%</div>
<div>>>> Phase 1 Iteration 15 Progress: 0%</div>
<div>>>> Phase 1 Iteration 20 Progress: 1%</div>
<div>>>> Phase 1 Iteration 25 Progress: 1%</div>
<div>>>> Phase 1 Iteration 30 Progress: 1%</div>
<div>>>> Phase 1 Iteration 35 Progress: 1%</div>
<div>>>> Phase 1 Iteration 40 Progress: 1%</div>
<div>>>> Error in solve.default(z$dfra) :</div>
<div>>>> system is computationally singular: reciprocal condition number = 4.58224e-20</div>
<div>>>> theta: 0.03416 0.01123 -0.01203 -0.00364 -0.17691 -0.40977 0.23908 -0.14365 0.25761 1.00000 0.21070</div>
<div>>>></div>
<div>>>> Start phase 2.1</div>
<div>>>> Phase 2 Subphase 1 Iteration 1 Progress: 14%</div>
<div>>>> Phase 2 Subphase 1 Iteration 2 Progress: 14%</div>
<div>>>> theta 0.04034 0.01130 -0.01206 -0.00342 -0.17012 -0.40569 0.24425 -0.13712 0.25683 0.99919 0.20914</div>
<div>>>> ac 1.063 1.049 1.006 1.044 1.025 1.063 1.067 1.065 0.651 0.954 0.866</div>
<div>>>> Phase 2 Subphase 1 Iteration 3 Progress: 14%</div>
<div>>>> Phase 2 Subphase 1 Iteration 4 Progress: 14%</div>
<div>>>> theta 0.05223 0.01144 -0.01213 -0.00299 -0.15670 -0.39786 0.25405 -0.12452 0.25377 0.99722 0.20489</div>
<div>>>> ac 1.136 1.112 1.013 1.096 1.105 1.134 1.161 1.137 0.690 0.984 0.881</div>
<div>>>> Phase 2 Subphase 1 Iteration 5 Progress: 14%</div>
<div>>>> Phase 2 Subphase 1 Iteration 6 Progress: 14%</div>
<div>>>> theta 0.06332 0.01158 -0.01221 -0.00256 -0.14449 -0.39056 0.26230 -0.11261 0.24601 0.99419 0.19779</div>
<div>>>> ac 1.137 1.140 0.936 1.145 1.104 1.136 1.164 1.140 0.863 0.962 0.967</div>
<div>>>> Phase 2 Subphase 1 Iteration 7 Progress: 14%</div>
<div>>>> Error in x$FRAN(zsmall, xsmall) :</div>
<div>>>> Unlikely to terminate this epoch: more than 1000000 steps</div>
<div>>>> Calls: siena07 ... proc2subphase -> doIterations -> <Anonymous> -> .Call</div>
<div>>>> Execution halted</div>
<div>>>></div>
<div>>>></div>
<div>>>></div>
<div>>>> Jennifer N. Victor</div>
<div>>>> Assistant Professor of Political Science</div>
<div>>>> University of Pittsburgh</div>
<div>>>> 4600 Posvar Hall</div>
<div>>>> jnvictor@pitt.edu</div>
<div>>>> (412) 624-7204</div>
<div>>>> _______________________________________________</div>
<div>>>> Rsiena-help mailing list</div>
<div>>>> Rsiena-help@lists.r-forge.r-project.org</div>
<div>>>> <a href="https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rsiena-help">
https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rsiena-help</a></div>
<div>>></div>
<div>>></div>
<div>>> _______________________________________________</div>
<div>>> Rsiena-help mailing list</div>
<div>>> Rsiena-help@lists.r-forge.r-project.org</div>
<div>>> <a href="https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rsiena-help">
https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rsiena-help</a></div>
<div>></div>
<div>></div>
<div>></div>
<div> </div>
<div> </div>
</font>
</body>
</html>