includeTimeDummy {RSiena}

Function to include time dummy effects in a Siena model

Description

This function specifies time heterogeneity for selected effects in a Siena model, by interacting them with time dummies, without explicitly using time-dependent covariates.

Usage

includeTimeDummy(myeff, ..., timeDummy="all", name=myeff$name[1], type="eval",
              interaction1="", interaction2="", include=TRUE, character=FALSE)

Arguments

myeff

A Siena effects object as created by getEffects.

...

Short names to identify the effects for which interactions with time dummies should be included or excluded. This function cannot be used for regular interaction effects.

timeDummy

Character string. Either "all" or the periods for which to create dummies (from 1 to (number of waves - 1)), space delimited.

include

Boolean. default TRUE, but can be switched to FALSE to turn off an effect.

name

Name of dependent network or behavioral variable for which effects are being included. Defaults to the first in the effects object.

type

Type of dummy effects to be interacted.

interaction1

Name of variable where needed to completely identify the effects e.g. covariate name or behavior variable name.

interaction2

Name of variable where needed to completely identify the effects e.g. covariate name or behavior variable name.

character

Boolean: are the effect names character strings or not

Details

The arguments (..., name, interaction1, interaction2) should identify the effects completely. See includeEffects and effectsDocumentation for more information about this.

This function operates by setting the timeDummy column on selected rows of a Siena effects object, thereby specifying interactions of the specified effect or effects with dummy variables for the specified periods. The timeDummy column of myeff will be set to include the values requested if include=TRUE, and to exclude them for include=FALSE.

For an effects object in which the timeDummy column of some of the included effects includes some or all period numbers, interactions of those effects with ego effects of time dummies for the indicated periods will also be estimated by siena07. For the outdegree effect this is just the ego effect of the time dummies. If ... does not include the outdegree effect, then still this ego effect will be created, but its parameter will be fixed to 0.

An alternative to the use of includeTimeDummy is to define time-dependent actor covariates (dummy variables or other functions of wave number that are the same for all actors), include these in the data set through sienaDataCreate, and include interactions of other effects with ego effects of these time-dependent actor covariates by includeInteraction. This is illustrated in an example in the help file for sienaTimeTest. Using includeTimeDummy is easier; on the other hand, using self-defined interactions with time-dependent variables gives more control (e.g., it will allow to specify linear time dependence and test time heterogeneity for interaction effects).

Value

An updated version of myeff, with the timeDummy column for one or more rows updated. Details of the rows altered will be printed.

Author(s)

Josh Lospinoso

References

See https://www.stats.ox.ac.uk/~snijders/siena/ for general information on RSiena.

See Also

sienaTimeTest, getEffects, includeEffects, effectsDocumentation.

Examples

## Not run: 
## Estimate a restricted model
myalgorithm <- sienaAlgorithmCreate(nsub=4, n3=1000)
mynet1 <- sienaDependent(array(c(s501, s502, s503), dim=c(50, 50, 3)))
mydata <- sienaDataCreate(mynet1)
myeff <- getEffects(mydata)
myeff <- includeEffects(myeff, transTrip, balance)
myeff
(ans <- siena07(myalgorithm, data=mydata, effects=myeff))

## Conduct the score type test to assess whether heterogeneity is present.
tt <- sienaTimeTest(ans)
summary(tt)

## Suppose that we wish to include a time dummy.
## Since there are three waves, the number of periods is two.
## This means that only one time dummy can be included for
## the interactions. The default is for period 2;
## an equivalent model, but with different parameters
## (that can be transformed into each other) is obtained
## when the dummies are defined for period 1.
myeff <- includeTimeDummy(myeff, density, recip, timeDummy="2")
myeff       # Note the \code{timeDummy} column.
(ans2 <- siena07(myalgorithm, data=mydata, effects=myeff))

## Re-assess the time heterogeneity
tt2 <- sienaTimeTest(ans2)
summary(tt2)

## And so on..

## End(Not run)

## A demonstration of RateX heterogeneity.
## Note that rate interactions are not implemented in general,
## but they are for Rate x coCovar.
## Not run: 
myalgorithm <- sienaAlgorithmCreate(nsub=4, n3=1000)
mynet1 <- sienaDependent(array(c(s501, s502, s503), dim=c(50, 50, 3)))
myccov <- coCovar(s50a[,1])
mydata <- sienaDataCreate(mynet1, myccov)
myeff <- getEffects(mydata)
myeff <- includeEffects(myeff, transTrip, balance)
myeff <- includeTimeDummy(myeff, RateX, type="rate",
            interaction1="myccov")
myeff
(ans <- siena07(myalgorithm, data=mydata, effects=myeff))

## End(Not run)