print.sienaBayesFit: Methods and functions for processing sienaBayes objects

In RSienaTest: Siena - Simulation Investigation for Empirical Network Analysis

Description

print and summary methods for sienaBayesFit objects, and further functions for interpretation of results.

Usage

## S3 method for class 'sienaBayesFit'
print(x, nfirst=NULL, ...)

## S3 method for class 'sienaBayesFit'
summary(object, nfirst=NULL, ...)

## S3 method for class 'summary.sienaBayesFit'
print(x, nfirst=NULL, ...)

shortBayesResults(x, nfirst=NULL)

plotPostMeansMDS(x, pmonly=1, excludeRates=TRUE, nfirst=NULL, ...)

Arguments

object	An object of class sienaBayesFit.
x	An object of class sienaBayesFit or summary.sienaBayesFit as appropriate.
nfirst	The first element of the MCMC chain used for calculating properties of the chain; i.e., the first element for which it is assumed that convergence has occurred. If NULL, the value x$nwarm+1 will be used.
pmonly	if pmonly=0 a plot is made for posterior means and standard deviations, if pmonly=1 only for the posterior means, if pmonly=2 only for the posterior standard deviations.
excludeRates	Logical: whether to exclude the rate parameters for determining the coordinates.
...	For extra arguments. At present used only for further specification of the plot made by plotPostMeansMDS, allowing further arguments for plot and text.

Value

The function print.sienaBayesFit prints results (estimated posterior means, posterior standard deviations, and percentiles) of a Bayesian analysis, as resulting from function sienaBayes.
The function summary.sienaBayesFit prints more extensive results, including groupwise results.
The columns "cred. from" and "cred. to" give the left and right end points of the estimated 95 percent credibility interval for the parameters, and the column "p" gives the estimated posterior probability that the parameter is greater than 0.
The function shortBayesResults produces a data frame containing posterior means and standard deviations, and 95 percent percentiles. This is meant to be a component for use in other functions.
Of this data frame, variables "name" to "effectName" are as in a sienaEffects object; variables "postMeanGlobal", "postSdGlobal", "cFrom", "cTo" are the posterior mean, posterior standard deviation, and lower and upper boundaries of the 95 percent posterior interval for the global parameters (mu and eta); variables "postSdBetween", "cSdFrom", "cSdTo" are the posterior mean, and similar percentiles, for the global between-group standard deviations (sigma) of the parameters that are randomly varying between groups.

The function plotPostMeansMDS makes an MDS plot of the posterior means. This can be useful for detection of outliers. Its return value is a list with two components: corpm, the similarities between the groups (correlation matrix of the transposed matrix of posterior means), used as input for the MDS; points, the matrix of two-dimensional MDS coordinates.

It is advisable to employ the first parameter; looking at trace plots will be helpful to estimate the number of runs from which convergence has occurred; this may well be different from the default value x$nwarm+1.

Function siena.table can also be used to make html or LaTeX tables of posterior means and standard deviations of a sienaBayesFit object.

Author(s)

Tom Snijders

References

See also http://www.stats.ox.ac.uk/~snijders/siena/
For the MDS method used in plotPostMeansMDS: see p. 308 of
W.N. Venables and B.D. Ripley (2002), Modern Applied Statistics with S. Fourth edition. Springer.

See Also

siena.table, sienaBayes, extract.sienaBayes, extract.posteriorMeans.

Examples

  Group1 <- sienaDependent(array(c(N3401, HN3401), dim=c(45, 45, 2)))
  Group3 <- sienaDependent(array(c(N3403, HN3403), dim=c(37, 37, 2)))
  Group4 <- sienaDependent(array(c(N3404, HN3404), dim=c(33, 33, 2)))
  Group6 <- sienaDependent(array(c(N3406, HN3406), dim=c(36, 36, 2)))
  dataset.1 <- sienaDataCreate(Friends = Group1)
  dataset.3 <- sienaDataCreate(Friends = Group3)
  dataset.4 <- sienaDataCreate(Friends = Group4)
  dataset.6 <- sienaDataCreate(Friends = Group6)
  FourGroups <- sienaGroupCreate(
        list(dataset.1, dataset.3, dataset.4, dataset.6))
  FourEffects <- getEffects(FourGroups)
  FourEffects <- includeEffects(FourEffects, transTrip)
  FourEffects <- setEffect(FourEffects, density, random=TRUE)
  FourEffects <- setEffect(FourEffects, recip, random=TRUE)
  print(FourEffects, includeRandoms=TRUE)
# Note this also shows the "randomEffects" column
# and the dimensions of the objects for specifying the priors.
  print(FourEffects, includeRandoms=TRUE, dropRates=TRUE)
# Note this does not show the rate effects.
  FourAlgo <- sienaAlgorithmCreate(projname = "FourGroups", maxlike=TRUE,
                                   seed=123)
## Not run: 
  bayes.model <- sienaBayes(FourAlgo, data = FourGroups,
       effects = FourEffects, nprewarm=10, nwarm=10, nmain=25, nrunMHBatches=10)
  bayes.model
# If one would assume that converge was reached by iteration 20:
  print(bayes.model, nfirst=20)
  summary(bayes.model)
  shortBayesResults(bayes.model)
# In the following, "main" is used as ... :
  plotPostMeansMDS(bayes.model, main="MDS of posterior means")
  siena.table(bayes.model, type="html", d=3)

## End(Not run)

RSienaTest documentation built on July 14, 2021, 3 a.m.