Search

[Abstract]: This paper introduces an area-level Poisson mixed model with SAR(1) spatially correlated random effects. Small area predictors of proportions and counts are derived from the new model and the corresponding mean squared errors are estimated by parametric bootstrap. The behaviour of the introduced predictors is empirically investigated by running model-based simulation experiments. An application to real data from the Spanish living conditions survey of Galicia (Spain) is given. The target is the estimation of domain proportions of women under the poverty line.

[Abstract]: Today, generalized linear mixed models (GLMM) are broadly used in many fields. However, the development of tools for performing simultaneous inference has been largely neglected in this domain. A framework for joint inference is indispensable to carry out statistically valid multiple comparisons of parameters of interest between all or several clusters. We therefore develop simultaneous confidence intervals and multiple testing procedures for empirical best predictors under GLMM. In addition, we implement our methodology to study widely employed examples of mixed models, that is, the unit-level binomial, the area-level Poisson-gamma and the area-level Poisson-lognormal mixed models. The asymptotic results are accompanied by extensive simulations. A case study on predicting poverty rates illustrates applicability and advantages of our simultaneous inference tools.

[Abstract]: Under an area-level random regression coefficient Poisson model, this article derives small area predictors of counts and proportions and introduces bootstrap estimators of the mean squared errors (MSEs). The maximum likelihood estimators of the model parameters and the mode predictors of the random effects are calculated by a Laplace approximation algorithm. Simulation experiments are implemented to investigate the behavior of the fitting algorithm, the predictors, and the MSE estimators with and without bias correction. The new statistical methodology is applied to data from the Spanish Living Conditions Survey. The target is to estimate the proportions of women and men under the poverty line by province.

[Abstract] In a complex socio-economic context, policy makers need highly disaggregated poverty indicators. In this work, we develop a methodology in small area estimation to derive predictors of poverty proportions under a random regression coefficient Poisson model, introducing bootstrap estimators of mean squared errors. Maximum likelihood estimators of model parameters and random effects mode predictors are calculated using a Laplace approximation algorithm. Simulation experiments are conducted to investigate the behaviour of the fitting algorithm, the predictors and the mean squared error estimator. The new statistical methodology is applied to data from the Spanish survey of living conditions to map poverty proportions by province and sex, developing a tool to support policy decision making

[Abstract]: This paper investigates the small area estimation of population averages of unit-level compositional data. The new methodology transforms the compositions into vectors of Rm and assumes that the vectors follow a multivariate nested error regression model. Empirical best predictors of domain indicators are derived from the fitted model, and their mean squared errors are estimated by parametric bootstrap. The empirical analysis of the behavior of the introduced predictors is investigated by means of simulation experiments. An application to real data from the Spanish household budget survey is given. The target is to estimate the average of proportions of annual household expenditures on food, housing and others, by Spanish provinces.

[Abstract]: Under a unit-level bivariate linear mixed model, this paper introduces small area predictors of expenditure means and ratios, and derives approximations and estimators of the corresponding mean squared errors. For the considered model, the REML estimation method is implemented. Several simulation experiments, designed to analyze the behavior of the introduced fitting algorithm, predictors and mean squared error estimators, are carried out. An application to real data from the Spanish household budget survey illustrates the behavior of the proposed statistical methodology. The target is the estimation of means of food and non-food household annual expenditures and of ratios of food household expenditures by Spanish provinces.

[Abstract]: The aim of this paper is to present an original approach to estimate the gender pay gap (GPG). We propose a model-based decomposition, similar to the most popular approaches, where the first component measures differences in group characteristics and the second component measures the unexplained effect; the latter being the real gap. The novel approach incorporates model selection and bias correction. The pay gap problem in a small area context is considered in this paper, although the approach is flexible to be applied to other contexts. Specifically, the methodology is validated for analysing wage differentials by economic activities in the region of Galicia (Spain) and by analysing simulated data from an experimental design that imitates the generation of real data. The good performance of the proposed estimators is shown in both cases, specifically when compared with those obtained from the widely used Oaxaca–Blinder approach.

[Abstract]: Over the past decades, linear mixed models have attracted considerable attention in various fields of applied statistics. They are popular whenever clustered, hierarchical or longitudinal data are investigated. Nonetheless, statistical tools for valid simultaneous inference for mixed parameters are rare. This is surprising because one often faces inferential problems beyond the pointwise examination of fixed or mixed parameters. For example, there is an interest in a comparative analysis of cluster-level parameters or subject-specific estimates in studies with repeated measurements. We discuss methods for simultaneous inference assuming a linear mixed model. Specifically, we develop simultaneous prediction intervals as well as multiple testing procedures for mixed parameters. They are useful for joint considerations or comparisons of cluster-level parameters. We employ a consistent bootstrap approximation of the distribution of max-type statistic to construct our tools. The numerical performance of the developed methodology is studied in simulation experiments and illustrated in a data example on household incomes in small areas.

[Abstract]: This paper introduces empirical best predictors of small area bivariate parameters, like ratios of sums or sums of ratios, by assuming that the target unit-level vector follows a bivariate nested error regression model. The corresponding means squared errors are estimated by parametric bootstrap. Several simulation experiments empirically study the behavior of the introduced statistical methodology. An application to real data from the Spanish household budget survey gives estimators of ratios of food household expenditures by provinces.

[Abstract]: This paper introduces area-level compositional mixed models by applying transformations to a multivariate Fay–Herriot model. Small area estimators of the proportions of the categories of a classification variable are derived from the new model, and the corresponding mean squared errors are estimated by parametric bootstrap. Several simulation experiments designed to analyse the behaviour of the introduced estimators are carried out. An application to real data from the Spanish Labour Force Survey of Galicia (north-west of Spain), in the first quarter of 2017, is given. The target is the estimation of domain proportions of people in the four categories of the variable labour status: under 16 years, employed, unemployed and inactive.

[Abstract]: Wildfires are considered one of the main causes of forest destruction. In recent years, the number of forest fires and burned area in Mediterranean regions have increased. This problem particularly affects Galicia (north-west of Spain). Conventional modelling of the number of forest fires in small areas may have a high error. For this reason, four area-level Poisson mixed models with time effects are proposed. The first two models contain independent time effects, whereas the random effects of the other models are distributed according to an autoregressive process AR(1). A parametric bootstrap algorithm is given to measure the accuracy of the plug-in predictor of fire number under the temporal models. A significant prediction improvement is observed when using Poisson regression models with random time effects. Analysis of historical data finds significant meteorological and socioeconomic variables explaining the number of forest fires by area and reveals the presence of a temporal correlation structure captured by the area-level Poisson mixed model with AR(1) time effects.

[Abstract] Wildfires are one of the main causes of forest destruction, especially in Galicia (north-west Spain), where the area burned by forest fires in spring and summer is quite high. This work uses two semiparametric time-series models to describe and predict the weekly burned area in a year: autoregressive moving average (ARMA) modelling after smoothing, and smoothing after ARMA modelling. These models can be described as a sum of a parametric component modelled by an autoregressive moving average process and a non-parametric one. To estimate the non-parametric component, local linear and kernel regression, B-splines and P-splines were considered. The methodology and software were applied to a real dataset of burned area in Galicia for the period 1999–2008. The burned area in Galicia increases strongly during summer periods. Forest managers are interested in predicting the burned area to manage resources more efficiently. The two semiparametric models are analysed and compared with a purely parametric model. In terms of error, the most successful results are provided by the first semiparametric time-series model.

[Abstract] The paper studies the applicability of area-level Poisson mixed models to estimate small area counting indicators. Among the available procedures for fitting generalized linear models, the method of moments (MM) and the penalised quasi-likelihood (PQL) method are employed. The empirical best predictor (EBP) of the area mean is derived using MM and compared with plug-in alternatives using MM and PQL. The plug-in estimator using PQL is computationally faster and provides competitive performance with respect to EBP that involves high complex integrals. An approximation to the mean squared error (MSE) of the EBP is given and three MSE estimators are proposed. The first two MSE estimators are plug-in estimators without and with bias correction to the second order and the third one is based on parametric bootstrap. Several simulation experiments are carried out for analysing the behaviour of the EBP and for comparing the estimators of the MSE of the EBP. A good choice in practice is the bootstrap alternative since it performs similarly to the analytical versions and is computationally faster. The developed methodology and software are applied to data from the 2008 Spanish living condition survey. The target of the application is the estimation of poverty rates at province level.