Manual for RSiena

1 Manual forRSienaRuth M. Ripley, Tom SnijdersZs ofia Boda, Andr as V or os, Paulina PreciadoUniversity of Oxford: Department of Statistics; Nuffield CollegeUniversity of Groningen: Department of SociologyDecember 15, 2021 AbstractSIENA(forSimulation Investigation for Empirical Network Analysis) is a computer pro-gram that carries out the statistical estimation of models for the evolution of socialnetworks according to the dynamic actor-oriented model of Snijders (2001, 2005),Snijders et al. (2007), and Snijders et al. (2010a). This is the Manual forRSiena, acontributed package to the statistical systemR. It complements, but does not replacethe help pages for theRSienafunctions!

2 It also contains contributions written by MarkHuisman, Michael Schweinberger, and Christian Manual is frequently updated, mostly only in a minor way. This version wasrenewed forRSienaversion General Acknowledgements .. Giving references to RSiena ..82 Getting started The logic of Stochastic Actor-Oriented Models .. of Stochastic Actor-Oriented Models .. , variables and effects .. of estimation procedure .. issues for longitudinal network modeling .. useful options inRSiena.. InstallingRandSIENA.. UsingSIENA withinR.. ExampleRscripts for getting started .. Steps for looking at results: ExecutingSIENA.

3 Getting help with problems ..213 Steps of modelling224 Input Data types .. data .. between matrix and edge list formats .. data .. covariates .. covariates .. Internal data treatment .. and dyadic transformations of covariates .. of behavioral variables .. dependent variables .. dependent variables with constraints .. Further data specification options .. determined values .. data .. change: joiners and leavers ..355 Model Definition of the model .. effects .. inSIENA.. specification .. Important structural effects for network dynamics:one-mode networks .. function.

4 Important structural effects for network dynamics:two-mode networks .. Effects for network dynamics associated with covariates .. Cross-network effects for dynamics of multiple networks .. Constraints between networks .. Effects on behavior evolution .. Model Type: non-directed networks .. Model Type: behavior .. Additional interaction effects .. Interaction effects for network dynamics .. Interaction effects for behavior dynamics .. Time heterogeneity in model parameters .. Limiting the maximum outdegree .. Goodness of fit with auxiliary statistics:sienaGOF.. Plots .. Composition change, missing data, and structural values insienaGOF.

5 606 An overview of the estimation procedure .. Parameters of the algorithm .. Results of the estimation .. Values .. Check .. estimation to obtain convergence .. Use of the algorithm parameters .. What to do if there are convergence problems .. Some important components of thesienaFitobject .. Algorithm .. Output .. check .. values and standard errors .. check .. Other estimation procedures .. Generalized Method of Moments estimation .. Using the functionsiena07().. Using the functionsienacpp().. Maximum Likelihood and Bayesian estimation .. Other remarks about the estimation algorithm.

6 Conditional and unconditional estimation .. Fixing parameters .. Automatic fixing of parameters .. Required changes from conditional to unconditional estimation .. Using multiple processes ..877 Standard Multicollinearity .. Precision of the finite differences method ..9038 Wald-type tests .. Wald tests .. of linear combinations .. Score-type tests .. Example: one-sided tests, two-sided tests, and one-step estimates .. tests .. Alternative application: convergence problems .. Testing differences between independent groups or periods .. Testing time heterogeneity in parameters.

7 1019 Accessing the generated networks .. Conditional and unconditional simulation .. Simulating with variable parameter values .. 10610 Getting Model choice .. 10811 Multilevel network FunctionsienaGroupCreate.. Multi-group Siena analysis .. Meta-analysis of Siena results .. Meta-analysis directed at the mean and variance of the parameters .. Meta-analysis directed at testing the parameters .. Contrast between the two kinds of meta-analysis .. Random coefficient multilevel Siena analysis .. Which data sets to use for sienaBayes .. Model specification.

8 How to enter your data in sienaBayes .. How to choose the parameter settings for sienaBayes .. Prior distributions .. Operation ofsienaBayes.. Assessing convergence .. Interpreting results of sienaBayes .. 12512 Formulas for Network evolution .. Network evaluation function .. Special effects for non-directed networks .. Multiple network effects .. Network creation and endowment functions .. Network rate function .. Behavioral evolution .. Behavioral evaluation function .. Behavioral creation function .. Behavioral endowment function.

9 Behavioral rate function .. Effects for estimation by Generalized Method of Moments .. 190413 Parameter Networks .. Behavior .. Ego alter selection tables .. Ego alter influence tables .. Average alter .. Average similarity .. Effect sizes and measures of fit .. Relative importance of effects .. Entropy / relative certainty .. Standard deviation of change statistics .. 20814 Error After updating .. During estimation .. As the result of a score-type test (including time test) .. In sienaGOF .. 21315 For programmers: Get the source code21416 For programmers: Other tools you need21417 For programmers: Building, installing and checking the package21418 For programmers: Understanding and adding an Example: adding the truncated out-degree effect.

10 Notes on effectGroups and two-mode networks .. 221A List of Functions in Order of Execution224B Changes compared to earlier versions234C References27351 General informationSIENA1, shorthand forSimulation Investigation for Empirical Network Analysis, is a set ofmethods implemented in a computer program that carries out the statistical estimation ofmodels for repeated measures of social networks according to the Stochastic Actor-orientedModel ( SAOM ) of Snijders and van Duijn (1997), Snijders (2001), Snijders et al. (2007),Snijders et al. (2010a), Snijders et al. (2013), and Greenan (2015); also see Steglich et al.