RSiena (version 1.1-212)

sienaTimeTest: Functions to assess and account for time heterogeneity of parameters

Description

Takes a sienaFit object from a siena07 estimation and tests the addition of dummy parameters at waves m=2...(M-1) through the score type test of Schweinberger (2007b). Tests for joint significance, parameter-wise significance, individual significance, and one step estimation of the unrestricted model parameters are returned in a list.

After assessing time heterogeneity, effects objects can be modified via the timeDummy column. Simply type the periods which you would like to have time dummied separated by commas into this column. Using the timeDummy column within the effects object, you may specify which time dummy interacted parameters are to be estimated.

If you wish to use this function with sienaFit objects which use the finite differences method of derivative estimation, or which use maximum likelihood estimation, you must request the derivatives to be returned by wave using the byWave=TRUE option to siena07.

Usage

sienaTimeTest(sienaFit, effects=NULL, condition=FALSE)

Value

sienaTimeTest Returns a list containing many items, including the following:JointTestA chi^2 test for joint significance of the dummies.ParameterTestA chi^2 test for joint significance across dummies for each separate effect.IndividualTestA matrix displaying initial estimates, one step estimates, and a p-value for H0: the unrestricted parameters are equal to zero.

Details

This test follows the score type test of Schweinburger (2007b) as implemented by Lospinoso et. al. (2010) by using statistics already calculated at each wave to populate vectors of partitioned moment functions corresponding to a restricted model (the model that has been fit by sienaFit Object) and an unrestricted model (which contains dummies for m=2...(M-1)). A covariance matrix of these statistics is calculated, and a delta matrix is constructed through the score functions' outer products with these statistics. Through an orthogonalization and a Delta method of approximation, the variance-covariance structure of the dummy statistics is calculated, and appropriate statistical tests can be used.

If it is determined that a time heterogeneity occurs for any number of time periods or effects (or any combination therein), the timeDummy column provides facilities within a sienaFit object to quickly adjust the model and re-estimate in the usual way. The includeTimeDummy function can be used to add the desired dummies entries to the effects object.

References

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

Lospinoso, J.A., Schweinberger, M., Snijders, T.A.B, and Ripley, R.M. "Assessing and Accounting for Time Heterogeneity in Stochastic Actor Oriented Models". Advances in Data Analysis and Computation. Special Issue on Social Networks. Submitted. Available from http://www.stats.ox.ac.uk/~lospinos/.

M. Schweinberger and T.A.B. Snijders. Markov models for digraph panel data: Monte carlo-based derivative estimation. Comput. Stat. Data Anal., 51(9):4465-4483, 2007. ISSN 0167-9473.

M. Schweinberger. Chapter 4: Statistical Modeling of Network Panel Data: Goodness of Fit. PhD thesis, University of Groningen, 2007.

See Also

siena07, plot.sienaTimeTest, includeTimeDummy

Examples

## Estimate a restricted model
mymodel <- sienaModelCreate(fn=simstats0c, nsub=2, n3=100)
mynet1 <- sienaNet(array(c(s501, s502, s503), dim=c(50, 50, 3)))
mydata <- sienaDataCreate(mynet1)
myeff <- getEffects(mydata)
myeff <- includeEffects(myeff, transTrip, balance)
ans <- siena07(mymodel, data=mydata, effects=myeff, batch=TRUE)

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

## Suppose that we wish to include two time dummies.
## Add them in the following way:
myeff <- includeTimeDummy(myeff, recip, balance, timeDummy="2")
ans2 <- siena07(mymodel, data=mydata, effects=myeff, batch=TRUE)

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

## And so on..

## A demonstration of the plotting facilities, on a larger dataset:
mymodel <- sienaModelCreate(fn=simstats0c, nsub=4, n3=500)
mynet1 <- sienaNet(array(c(s501, s502, s503, s501, s503, s502), dim=c(50, 50, 6)))
mydata <- sienaDataCreate(mynet1)
myeff <- getEffects(mydata)
myeff <- includeEffects(myeff, transTrip, balance)
myeff <- includeTimeDummy(myeff, recip, timeDummy="2,3,5")
myeff <- includeTimeDummy(myeff, balance, timeDummy="4")
myeff <- includeTimeDummy(myeff, density, timeDummy="all")
ansp <- siena07(mymodel, data=mydata, effects=myeff, batch=FALSE)
ttp <- sienaTimeTest(ansp)

## Pairwise plots show
plot(ttp, pairwise=TRUE)

## Time test plots show
plot(ttp, effects=1:4, dims=c(2,2))
## A demonstration of RateX heterogeneity. Note that rate
## interactions are not implemented in general, just for
## Rate x cCovar.
mymodel <- sienaModelCreate(fn=simstats0c, nsub=2, n3=100)
mynet1 <- sienaNet(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")
ans <- siena07(mymodel, data=mydata, effects=myeff, batch=TRUE)