Function to set various columns in an effects object in a Siena model.
This function provides an interface to change various columns of a selected row of a Siena effects object.
setEffect(myeff, shortName, parameter = NULL, fix = FALSE, test = FALSE, random=FALSE, initialValue = 0, timeDummy = ",", include = TRUE, name = myeff$name[1], type = "eval", interaction1 = "", interaction2 = "", effect1=0, effect2=0, effect3=0, period=1, group=1, character=FALSE, verbose = TRUE)
myeff |
a Siena effects object as created by |
shortName |
A short name (all with or all without quotes) to identify the effect which should be changed. |
parameter |
Value of internal effect parameter. If NULL, no change is made. |
fix |
For fixing effects. Boolean required. Default FALSE. |
test |
For testing effects by score-type tests. Boolean required. Default FALSE. |
random |
For specifying that effects will vary randomly;
not relevant for |
initialValue |
Initial value required. Default 0. |
timeDummy |
string: Comma delimited string of which periods to
dummy. Alternatively, use |
include |
Boolean. default TRUE, but can be switched to FALSE to turn off an effect. |
name |
Name of dependent variable (network or behavior) for which effects are being modified. Defaults to the first in the effects object. |
type |
Character string indicating the type of the effect to be changed : currently "rate", "eval", "endow", or "creation". Default "eval". |
interaction1 |
Name of siena object where needed to completely identify the effect e.g. covariate name or behavior variable name. |
interaction2 |
Name of siena object where needed to completely identify the effect e.g. covariate name or behavior variable name. |
effect1 |
Only for |
effect2 |
Only for |
effect3 |
Only for |
period |
Number of period if basic rate. Use numbering within groups. |
group |
Number of group if basic rate.
Only relevant for |
character |
Boolean: whether the short name is a character string. |
verbose |
Boolean: should the print of altered effects be produced. |
Recall from the help page for getEffects
that a Siena effects object (class sienaEffects or
sienaGroupEffects) is a data.frame;
the rows in the data frame are the effects for this data set;
some of the columns/variables of the data frame
are used to identify the effect, other columns/variables define how this
effect is used in the estimation.
The arguments shortName, name, type,
interaction1, interaction2, effect1, effect2,
effect3, period, and group should identify one
effect completely. (Not all of them are needed;
see getEffects.)
The call of setEffect will set, for this effect,
the column elements of the resulting effects object for parameter,
fix, test, randomEffects, initialValue,
timeDummy, and include to the values requested.
The shortName must not be set between quotes,
unless you use character=TRUE.
An object of class sienaEffects or
sienaGroupEffects. This will be an updated version of the
input effects object, with one row updated.
Details of the row altered will be printed, unless verbose=FALSE.
Ruth Ripley
mynet <- sienaDependent(array(c(s501, s502, s503), dim=c(50, 50, 3)))
mybeh <- sienaDependent(s50a, type="behavior")
mydata <- sienaDataCreate(mynet, mybeh)
myeff <- getEffects(mydata)
# Set the initial rate parameter for one period:
myeff <- setEffect(myeff, Rate, initialValue=1.5, name="mybeh",
type="rate", period=2)
# Specify an effect parameter:
myeff <- setEffect(myeff, outTrunc, parameter=1)
myeff
# Now request the interaction between reciprocity and alcohol similarity
# to be tested without estimation.
myeff <- includeInteraction(myeff, recip, simX, interaction1=c("", "mybeh"))
myeff
# The following code uses the fact that the interaction effect is number 6.
# In practice one would just look up the numbers of effect1 and effect2
# from the response to \code{myeff} and fill them in.
(eff1 <- myeff[myeff$include,]$effect1[6])
(eff2 <- myeff[myeff$include,]$effect2[6])
myeff <- setEffect(myeff, unspInt, fix=TRUE, test=TRUE,
effect1=eff1, effect2=eff2)
myeffPlease choose more modern alternatives, such as Google Chrome or Mozilla Firefox.