obpFHbenchmark: Benchmarked observed best predictor for Fay-Herriot model.

In rohosen/OBPSAE: Observed best predictors for small area estimation

Description

This function computes the benchmarked Observed Best Predictor (OBP) for Fay-Herriot model. Depending on the method specified by the user it computes the Adjusted OBP or Augmented OBP or both.

Usage

obpFHbenchmark(formula, data, errorvar, weight, method = c("adjusted",
  "augmented"), randvar = NULL, maxiter = 100, precision = 1e-04)

Arguments

formula	an object of class formula (or one that can be coerced to that class): a symbolic description of the model to be fitted. The variables included in formula must have a length equal to the number of small areas. More about the model specification are given under Details.
data	optional data frame containing the variable names in formula.
errorvar	vector containing the variances of the random errors for all small areas.
weight	vector containing the sampling weights of small areas. If sum of the weights is not 1, the weights are normalized.
method	string specifying the benchmarking method. Options are "adjusted" and "augmented". Computes both if not specified. See Details for more usage information.
randvar	variance of the random effect. If not supplied, BPE is estimated.
maxiter	maximum number of iterations used in estimating randvar.
precision	covergence tolerance limit for estimating randvar.

Details

If method is set to "adjusted", only obpAdjusted is returned.

If method is set to "augmented", obpAugmented, A.BPE.aug and beta.BPE.aug are returned.

The variance of the random effect can be specified by the user. Otherwise the function will calculate its Best Predictive Estimator (BPE). In the process of of computing OBP it also calculates the BPE of the regression coefficients of the fixed effect.

formula is specified in the form response ~ predictors where the predictors are separated by +. formula has an implied intercept term. To remove the intercept term, use either y ~ x - 1 or y ~ 0 + x.

Value

The function will return a list with all the following objects by default.

obpAdjusted	a vector of adjusted OBP values.
obpAugmented	a vector of augmented OBP values.
A.BPE.aug	BPE of variance component of random effects under the augmented model (if not provided by the user).
beta.BPE.aug	BPE of fixed effects regression coefficients under the augmented model.

References

Bandyopadhyay R, Jiang J (2017) "Benchmarking the Observed Best Predictor"

Jiang J, Nguyen T, and Rao J. S. (2011), "Best Predictive Small Area Estimation", Journal of the American Statistical Association.

rohosen/OBPSAE documentation built on May 17, 2019, 2:22 p.m.