as_covariate_rsiena: Function to create a covariate object

In RSiena: Siena - Simulation Investigation for Empirical Network Analysis

Function to create a covariate object

Description

This function creates a monadic or dyadic covariate object

Usage

as_covariate_rsiena(val, type="monadic", centered=TRUE, 
                                nodeSet="Actors", warn=TRUE, 
                                imputationValues=NULL)

Arguments

val	Covariate values, given as a vector, or matrix (may be sparse, of type "TsparseMatrix"), or 3-dimensional array, or list of 3 sparse matrices of type "TsparseMatrix"; which of these structures determines the kind of covariate (see Details).
type	"monadic", "oneMode" or "bipartite": whether the values refer to a monadic covariate, and in case of a dyadic covariate, whether it is for a one-mode or a bipartite (two-mode) network (see Details).
centered	Boolean: if TRUE, then the mean value is subtracted.
nodeSet	Name of node set: character string; for type="bipartite", referring to bipartite (two-mode) dyadic covariates, a vector of two node set names (character strings). If the entire data set contains more than one node set, then the node sets must be specified in all data objects.
warn	Logical: determines whether a warning given if all values are NA, or all non-missing values are the same.
imputationValues	For monadic covariates only: Vector or matrix of covariate values of same length or dimensions as val, to be used for imputation of NA values (if any) in val. Must not contain any NA.

Details

This function determines the kind of covariate from the kind of arguments given:

• if val is a vector, it will be a time-constant monadic (actor) covariate;

• if val is a matrix and type="monadic", it will be a changing monadic (actor) covariate;

• if val is a matrix and type="oneMode" or "bipartite", it will be a time-constant dyadic covariate;

• if val is a 3-dimensional array or list of sparse matrices of type "TsparseMatrix", it will be a changing dyadic covariate.

When part of a siena data object, the created covariate is associated with the node set nodeSet of this data object. In practice, the node set needs to be specified only in the case of the use of the covariate with a two-mode network.

If there are any NA values in val, and imputationValues is given, and the covariate is monadic, then the corresponding elements of imputationValues are used for imputation. If imputationValues is NULL, imputation is by the mean value. In both cases, cases with imputed values are not used for calculating target statistics (see the manual).

For changing covariates (monadic or dyadic), the values for wave m are supposed in the simulations to be valid in the whole period from wave m to wave m+1. If the data set has M waves, this means that the values, if any, for wave M will not be used. These values may be given or omitted as part of val.

The function replaces coCovar, varCovar, coDyadCovar, varDyadCovar, which now are obsolete.

Value

The covariate is returned as an object of class coCovar, varCovar, coDyadCovar, or
varDyadCovar, as the case may be. This object can be used as an argument to make_data_rsiena.

Author(s)

Tom A.B. Snijders

References

See https://www.stats.ox.ac.uk/~snijders/siena/

See Also

make_data_rsiena, as_nodeset_rsiena; this function replaces the functions coCovar, varCovar, coDyadCovar, varDyadCovar.

Examples

# constant monadic covariate:
myconstcovar <- as_covariate_rsiena(s50a[,1])
senders <- as_nodeset_rsiena(50, nodeSetName="senders")
receivers <- as_nodeset_rsiena(30, nodeSetName="receivers")
senders.attribute <- as_covariate_rsiena(rep(1:10, each=5), 
                                         nodeSet="senders")
receivers.attribute <- as_covariate_rsiena(rep(1:5, each=6), 
                                         nodeSet="receivers")
# varying monadic covariate:
myvarcovar <- as_covariate_rsiena(s50a, type="monadic")
senders.covariate <- as_covariate_rsiena(s50a, nodeSet="senders", 
                                               type="monadic")
receivers.covariate <- as_covariate_rsiena(s50s[1:30,], 
                            nodeSet="receivers", type="monadic")
# constant dyadic covariate:
mydyadvar <- as_covariate_rsiena(s503, type="oneMode")
# varying dyadic covariate:
mydyadvar <- as_covariate_rsiena(array(c(s501, s502), dim=c(50, 50, 2)), 
                                type="oneMode")
# using sparse matrices for a constant dyadic covariate:
require(Matrix)
sps501 <- as(s501,"TsparseMatrix")
sps502 <- as(s502,"TsparseMatrix")
sps502r <- as(s502[1:30,],"TsparseMatrix")
spcDyadvar <- as_covariate_rsiena(sps501, type="oneMode")
spvDyadvar <- as_covariate_rsiena(list(sps501,sps502), type="oneMode")
spvDyadBipvar <- as_covariate_rsiena(list(sps501,sps502), 
                   type="bipartite", nodeSet=c("senders","receivers"))

RSiena documentation built on April 20, 2026, 5:07 p.m.