lavaSearch2 (version 1.5.6)

leverage2: Extract Leverage Values

Description

Extract leverage values from a Gaussian linear model.

Usage

leverage2(object, ...)
# S3 method for lm
leverage2(object, param = NULL, data = NULL, ...)
# S3 method for gls
leverage2(object, param = NULL, data = NULL, ...)
# S3 method for lme
leverage2(object, param = NULL, data = NULL, ...)
# S3 method for lvmfit
leverage2(object, param = NULL, data = NULL, ...)
# S3 method for lm2
leverage2(object, param = NULL, data = NULL, ...)
# S3 method for gls2
leverage2(object, param = NULL, data = NULL, ...)
# S3 method for lme2
leverage2(object, param = NULL, data = NULL, ...)
# S3 method for lvmfit2
leverage2(object, param = NULL, data = NULL, ...)

Value

a matrix containing the leverage relative to each sample (in rows) and each endogenous variable (in column).

Arguments

object

a lm2, gls2, lme2, or lvmfit2 object.

...

arguments to be passed to sCorrect.

param

[optional] the fitted parameters.

data

[optional] the data set.

Details

The leverage are defined as the partial derivative of the fitted values with respect to the observations. $$ leverage_i = \frac{\partial \hat{Y}_i}{\partial Y_i} $$ See Wei et al. (1998).

If argument p or data is not null, then the small sample size correction is recomputed to correct the residuals.

References

Bo-Cheng Wei et al., Generalized Leverage and its applications (1998), Scandinavian Journal of Statistics 25:1:25-37.

See Also

sCorrect to obtain lm2, gls2, lme2, or lvmfit2 objects.

Examples

## simulate data
set.seed(10)
m <- lvm(Y1~eta,Y2~eta,Y3~eta)
latent(m) <- ~eta
d <- lava::sim(m,20, latent = FALSE)

## standard linear model
e.lm <- lm(Y1~Y2, data = d)

sCorrect(e.lm) <- TRUE
range(as.double(leverage2(e.lm)) - influence(e.lm)$hat)

## latent variable model
e.lvm <- estimate(m, data = d)
sCorrect(e.lvm) <- TRUE
leverage2(e.lvm)