getEffects: Function to create a Siena effects object

Description

Creates a basic list of effects for all dependent variables in the input siena object.

Usage

getEffects(x, nintn = 10, behNintn=4, getDocumentation=FALSE, onePeriodSde=FALSE)

Value

An object of class sienaEffects or sienaGroupEffects: this is a data frame of which the rows are the effects available for data set x.

The effects object consists of consecutive parts, each of which relates to one dependent variable in the input object. The columns are:

name: name of the dependent variable
effectName: name of the effect
functionName: name of the function
shortName: short name for the effect
interaction1: second variable to define the effect, if any
interaction2: third variable to define the effect, if any
type: "eval", "endow", "creation", "rate", or "gmm"
basicRate: boolean: whether a basic rate parameter
include: boolean: include in the model to be fitted or not
randomEffects: boolean: random or fixed effect. Currently not used.
fix: boolean: fix parameter value or not
test: boolean: test parameter value or not
timeDummy: comma separated list of periods, or "all", or "," for none -- which time dummy interacted parameters should be included?
initialValue: starting value for estimation, also used for fix and test.
parm: internal effect parameter values
functionType: "objective" or "rate"
period: period for basic rate parameters
rateType: "Structural", "covariate", "diffusion"
untrimmedValue: Used to store initial values which could be trimmed
effect1: Used to indicate effect number in user-specified interactions
effect2: Used to indicate effect number in user-specified interactions
effect3: Used to indicate effect number in user-specified interactions
interactionType: Defines "dyadic" or "ego" or "OK" effects, used in includeInteraction
local: whether a local effect; used for the option localML in sienaAlgorithmCreate
effectFn: here NULL, but could be replaced by a function later
statisticFn: here NULL, but could be replaced by a function later
netType: Type of dependent variable: "oneMode", "behavior", or "bipartite"
groupName: name of relevant group data object
group: sequential number of relevant group data object in total
effectNumber: a unique identifier of the row

The combination of name, shortName, interaction1,

interaction2, and type uniquely identifies any effect other than basic rate effects and user-specified interaction effects. For the latter, effect1, effect2 and effect3

are also required for the identification. The combination name,

shortName, period and group uniquely identifies a basic rate effect.

The columns not used for identifying the effect define how the effect is used for the estimation.

The columns initialValue and parm should not be confused:

initialValue gives the initial value for the parameter to be estimated, indicated in the manual by theta;

parm gives the internal value of the parameter defining the effect, indicated in the manual (Chapter 12) by p, and is fixed during the estimation.

A list of all effects in a given effects object (e.g., myeff), including their names of dependent variables, effect names, short names, and values of interaction1 and interaction2 (if any), is obtained by executing effectsDocumentation(myeff).

As from version 1.3.24, effects object have a "version" attribute. Effects objects including interaction effects created by

includeInteraction are not necessarily compatible between versions of RSiena. Therefore it is recommended, for effects objects including any interaction effects, to create them again when changing to a new version of RSiena. If an effects object including any interaction effects is used from an old version of RSiena, this will lead to a warning when running

siena07.

Arguments

x: an object of class `siena" or `sienaGroup"
nintn: Number of user-defined network interactions that can later be created.
behNintn: Number of user-defined behavior interactions that can later be created.
getDocumentation: Flag to allow documentation of internal functions, not for use by users.
onePeriodSde: Flag to indicate that the stochastic differential equation (SDE) model dZ(t) = [aZ(t) + b] dt + g dW(t) should be used, instead of the regular SDE with a scale parameter. This is only relevant in case the model includes a continuous dependent variable and one period is studied.

Author

Ruth Ripley

Details

Creates a data frame of effects for use in siena model estimation. The regular way of changing this object is by the functions includeEffects, setEffect, and includeInteraction.

Note that the class of the return object may be lost if the data.frame is edited using fix. See fix and edit.data.frame.

References

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

Examples

Run this code

mynet1 <- sienaDependent(array(c(s501, s502, s503), dim=c(50, 50, 3)))
mybeh <- sienaDependent(s50a, type="behavior")
mycovar <- coCovar(rnorm(50))
mydyadcovar <- coDyadCovar(matrix(as.numeric(rnorm(2500) > 2), nrow=50))
mydata <- sienaDataCreate(mynet1, mybeh, mycovar, mydyadcovar)
myeff <- getEffects(mydata)
myeff

Run the code above in your browser using DataLab