summary.ideal: summary of an ideal object

Description

Provides a summary of the output from ideal point estimation contained in an object of class ideal.

Usage

## S3 method for class 'ideal':
summary(object, prob=.95,
              burnin=NULL,
              sort=TRUE,
              include.beta=FALSE,...)

Arguments

object

an object of class ideal.

prob

scalar, a proportion between 0 and 1, the content of the highest posterior density (HPD) interval to compute for the parameters

burnin

of the recorded MCMC samples, how many to discard as burnin? Default is NULL, in which case the value of burnin in the ideal object is used.

sort

logical, default is TRUE, indicating that the summary of the ideal points be sorted by the estimated posterior means (lowest to highest)

include.beta

whether or not to calculate summary statistics of beta, if beta is available. If the item parameters were not stored in the ideal object, then include.beta is ignored.

...

further arguments passed to or from other functions

Value

An item of class summary.ideal with elements:
objectthe name of the ideal object as an unevaluated expression, produced by match.call()$object
xmn by d matrix of posterior means for the ideal points
xsdn by d matrix of posterior means for the ideal points
xHDRn by 2 by d array of HDRs for the ideal points
bmm by d+1 matrix of posterior means for the item parameters
bsdm by d+1 matrix of posterior standard deviation for the item parameters
bHDRm by 2 by d+1 array of HDRs for the item parameters
bSiga list of length d, each component a vector of length m, of mode logical, equal to TRUE if the corresponding discrimination parameter is distinguishible from zero; see Details. If store.item was set to FALSE when ideal was invoked, then bSig is a list of length zero.
party.quantif party information is available through the rollcall object that was used to run ideal, then party.quant gives the posterior mean of the legislators' ideal points by party, by dimension. If no party information is available, then party.quant=NULL.

Details

The test of whether a given discrimination parameter is distinguishible from zero first checks to see if the two most extreme quantiles are symmetric around .5 (e.g., as are the default value of .025 and .975). If so, the corresponding quantiles of the MCMC samples for each discrimination parameter are inspected to see if they have the same sign. If they do, then the corresponding discrimination parameter is flagged as distinguishible from zero; otherwise not.

Examples

Run this code

## fake example
set.seed(314159265)
fakeData <- matrix(sample(x=c(0,1),size=1000,replace=TRUE),
                   10,100)
rc <- rollcall(fakeData)
## short-run for demo purposes
idFake <- ideal(rc,maxiter=500,burnin=100,thin=10)
summary(idFake)

## Supreme Court Example
data(sc9497)
rc <- rollcall(data=sc9497$votes,
               legis.names=sc9497$legis.names,
               desc=sc9497$desc)
id1 <- ideal(rc)
summary(id1)

data(s109)
cl2 <- constrain.legis(s109,
                       x=list("KENNEDY (D MA)"=c(-1,0),
                         "ENZI (R WY)"=c(1,0),
                         "CHAFEE (R RI)"=c(0,-.5)),
                       d=2)
id2Constrained <- ideal(s109,
                        d=2,
                        priors=cl2,      ## priors (w constraints)
                        startvals=cl2,   ## start value (w constraints)
                        store.item=TRUE,
                        maxiter=5000,
                        burnin=500,
                        thin=25)

summary(id2Constrained,
        include.items=TRUE)