An object of class emdi that represents point predictions of regional disaggregated indicators. Optionally, it also contains corresponding MSE estimates. Three different estimation approaches are implemented: direct estimation (class 'direct), the Fay-Herriot model (class "fh"), and the empirical best prediction (class "ebp"). Objects of these classes have methods for various generic functions. See Details for more information.
The following components are always included in an emdi object but not always filled and with different components depending on the estimation approach:
callthe function call that produced the object.
fixedfor details, see fixed in fh and
ebp. Not filled for class "direct".
frameworka list with components that describe the data setup, e.g., number of domains in the sample.
inddata frame containing estimates for indicators per domain.
methodcharacter returning the method for the estimation approach used to fit the linear mixed model and for the the optimal lambda (for class "ebp"), here "reml", or a list returning method for the estimation of the variance of the random effect and the applied MSE estimation (for class "fh"). Not filled for class "direct".
modellist containing a selection of model components. Not filled for class "direct".
MSEdata frame containing MSE estimates corresponding to the
point predictions in ind per indicator per domain if MSE is selected
in function call. If FALSE, MSE is NULL.
transformationcharacter or list containing information about applied transformation and, if appropriate, backtransformation. Not filled for class "direct".
transform_parama list with two elements, optimal_lambda
and shift_par, where the first contains the optimal parameter for a
transformation with transformation parameter or NULL for no and log
transformation and the second the potential shift parameter in the log or
Box-Cox transformation and NULL for no transformation. Not filled for
class "fh" and "direct".
successful_bootstrapsfor class "direct", a matrix with domains as rows and indicators as columns. The cells contain the number of successful bootstraps for each combination. For non-robust spatial Fay-Herriot, string with number of successful bootstraps. Not filled for other models.
Objects of class "emdi" have following methods:
compare_pred, estimators,
plot.emdi, predict.emdi,
qqnorm.emdi
Objects of class "direct", "ebp" and "fh" have methods for following generic
functions: compare_plot, getData,
getGroups, getGroupsFormula,
getResponse,
plot (for documentation, see plot.emdi),
print, qqnorm (for documentation, see
qqnorm.emdi) and summary (for documentation,
see emdi_summaries).
Objects of class "ebp" and "fh" additionally have methods for following
generic functions: coef (for default documentation,
see coef), confint (for default documentation,
see confint), family (for default
documentation, see family), fitted (for default
documentation, see fitted.values),
fixef, formula (for default documentation, see
formula), getVarCov,
intervals, logLik (for default documentation,
see logLik), nobs (for default documentation,
see nobs), ranef, residuals
(for default documentation, see residuals), terms
(for default documentation, see terms), vcov (for
default documentation, see vcov)
Objects of class "ebp" have additionally methods for following generic
functions: sigma (for default documentation, see
sigma)
Objects of class "fh" have additionally methods for following generic
functions: compare, extractAIC (for default
documentation, see extractAIC) and
step.
Alfons, A. and Templ, M. (2013). Estimation of Social Exclusion Indicators
from Complex Surveys: The R Package laeken. Journal of
Statistical Software, 54(15), 1-25.
Fay R.E., Herriot R.A. (1979) Estimates of income for small places: An
application of James–Stein procedures to census data. Journal of the
American Statistical Association, Vol. 74, 269–277.
Molina, I. and Rao, J.N.K. (2010). Small area estimation of poverty
indicators. The Canadian Journal of Statistics, Vol. 38, No.3, 369-385.