2016-Small Area Estimation: New Developments

 

Small Area Estimation: New Developments 
Chair: Karla Fox (Statistics Canada)
Organizer: Mahmoud Torabi (University of Manitoba) 
[PDF]

M. GIOVANNA RANALLI, University of Perugia, Italy
Time Series Small Area Estimation for Unemployment Rates using Latent Markov Models [PDF]
 
In this work we develop a new area-level Small Area Estimation method using Latent Markov Models (LMMs) in a Bayesian setting. LMMs handle longitudinal data in which the characteristics of interest, and their evolution in time, are represented by a latent process that follows a Markov chain. Small areas are allowed to move between latent states over time. LMMs may be seen as an extension of Markov chain models to control for measurement errors and as a dynamic extension of Latent class models for longitudinal data. Estimation is conducted using a Gibbs sampler with data augmentation. The proposed model is applied to estimate quarterly unemployment rates for Italian Local Labour Market Areas using LFS data from 2004 to 2014. 
 
J.N.K. RAO, Carleton University
On Measuring Uncertainty of Small Area Estimators Under Area Level and Unit Level Models  [PDF]
 
Model-based estimators of small area means that borrow strength from related areas through linking models are extensively used because direct area-specific estimators are imprecise due to small sample sizes within areas. We study different methods of estimating mean squared error (MSE) of the model-based estimators and appraise their relative properties. We also study methods of finding confidence intervals on the small area means. We consider both area level and unit level models. 
 
MAHMOUD TORABI, University of Manitoba
Prediction in Small-Area Spatially Correlated Data [PDF]
 
In small area estimation, we need to predict characteristics of the sub-populations (areas) based on the coarse scale data. Small area predictors are improved by using (standard) mixed models. However, there are many situations where the characteristics are related to their locations. For example, it is an interest of policy makers (and public) to know the spatial pattern of a rare disease (e.g., chronic disease or cancer). We propose small area models in the class of spatial mixed models to be able to predict characteristics and also to obtain corresponding mean squared prediction error (MSPE) and MSPE estimate. The performance of our proposed approach is evaluated through simulations and by a real application.