Table of Contents

Table of Contents

Preface

Introduction

Introduction

Main Aims of the Book

Guide for the Reader

Concluding Remarks

Small Area Estimation

Introduction

Small area estimation

Advantages of small area estimation

Why small area estimation techniques?

Applications of small area estimation

Approaches to small area estimation

Direct estimation

Horvitz-Thomposn (H-T) estimator

Generalized regression (GREG) estimator

Modified direct estimator

Design-based model-assited estimators

A comparison of direct estimators

Concluding remarks

Indirect Estimation: Statistical Approaches

Introduction

Implicit models approach

Synthetic estimaton

Composite estimation

Demographic estimation

Comparison of various implicit models based indirect estimation

Explicit models approach

Basic area level model

Basic unit leve model

General linear mixed model

Comparison of various explicit models based indirect estimation

Methods for estimating explicit models

E-BLUP approach

EB approach

HB approach

A comparison of three methods

Concluding remarks

Indirect Estimation: Geographic Approaches

Introduction

Microsimulation modeling

Process of microsimulation

Types of microsimulation models

Advantages of microsimulation modeling

Methodologies in microsimulation modeling technology

Techniques for creating spatial microdata

Statistical data matching or fusion

Iterative proportional fitting

Repeated weighting method

Reweighting

Combinatorial optimisation reweighing approach

The simulated annealing method in CO

An illustration of CO process for hypothetical data

Reweighting: The GREGWT approach

Theoretical setting

How does GREGWT generate new weights?

Explicit numerical solution for a hypothetical data

A comparison between GREGWT and CO

Concluding remarks

Bayesian Prediction-Based Microdata Simulation

Introduction

The basic steps

The Bayesian prediction theory

The multivariate model

The prior and posterior distributions

The linkage model

Prediction for moedling unobserved population units

Concluding remarks

Microsimulation Modelling Technology for Small Area Estimation

Introduction

Data sources and issues

The Census Data

Survey Datasets

Survey Datasets

MMT based Model Specification

Model inputs

Generating small area synthetic weights

Model inputs

Generating small area synthetic weights

Model inputs

Gnerating small area synthetic weights

Model outputs

Housing stress

Definition

Measures of housing stress

A comparison of various measures

Small area estimation of housing stress

Inputs at the second stae model

Final model outputs

Concluding remarks

Applications of the Methodologies

Introduction

Results of the model: A general view

Model accuracy report

Scenarios of housing stress under various measures

Distribution of housing stress estimation

Lorenz curve for housing stress estimates

Proportional cumulative frequency graph and index of dissimilarity

Scenarios of households and housing stress by tenures

Estimation of households in housing stress by spatial scales

Results for different states

Results for various statistical divisions

Results for various statistical subdivisions

Small area estimates: Number of households in housing stress

Estimated numbers of overall households in housing stress

Estimated numbers of buyerhouseholds in housing stress

Estimated numbers of public renter households in housing stress

Estimated numbers of private renter households in housing stress

Estimated numbers of total renter households in housing stress

Small area estimates: Percentage of households in housing stress

Percentage estimates of housing stress for overall households

Percentage estimates of housing stress for buyer households

Percentage estimates of housing stress for public renter households

Percentage estimates of housing stress for private renter households

Percentage estimates of housing stress for total renter households

Concluding remarks

Analysis of Small Area Estimates in Capital Cities

Introduction

Scenarios of the results for major capital cities

Trends in housing stress for some major cities

Mapping the estimates at SLA levels within major cities

Sydney

Housing stress estimates for overall households

Small area estimation by household's tenure types

Melbourne

Housing stress estimates for overall households

Small area estimation by household's tenure types

Brisbane

Housing stress estimates for overall households

Small area estimation by household's tenure types

Adelaide

Housing stress estimates for overall households

Small area estimation by household's tenure types

Canberra

Housing stress estimates for overall households

Small area estimation by household's tenure types

Hobart

Housing stress estimates for overall households

Small area estimation by household's tenure types

Darwin

Housing stress estimates for overall households

Small area estimation by household's tenure types

Concluding remarks

Validation and Measure of Statistical Reliability

Introduction

Some validation methods in the literature

New approaches to validating housing stress estimation

Statistical significance test of the MMT estimates

Results of the statistical significance test

Absolute standardised residual estimate (ASRE) analysis

Results from the ASRE analysis

Measure of statistical reliability of the MMT estimates

Confidence interval estimation

Results from the estimates of confidence intervals

Concluding remarks

Conclusions and Computing Codes

Introduction

Summary of major findings

Limitations

Areas of further studies

Computing codes and programming

The general model file codes

SAS programming for reweithing algorithms

The second stage program file codes

Concluding remarks

Appendices.

About the Author

Associate Professor Azizur Rahman, PhD, is a statistician and data scientist with expertise in both developing and applying novel methodologies, models and technologies. He is the Leader of “Statistics and Data Mining Research Group” at the Charles Sturt University (CSU), and able to assist in understanding multi-disciplinary research issues within various fields including how to understand the individual activities which occur within very complex scientific, behavioural, socio-economic and ecological systems. His research encompasses issues in simple to multi-facet analyses in various fields ranging from the statistical sciences to the law and legal studies. He has more than 100 scholarly publications including a few books. Prof. Rahman’s research is funded by the Australian Federal and State Governments, and he serves on a range of editorial boards including the International Journal of Microsimulation (IJM) and Sustaining Regions. He obtained several awards including the SOCM Research Excellence Award 2018 and the CSU-RED Achievement Award 2019.

Professor Ann Harding, AO, is an Emeritus Professor of Applied Economics and Social Policy at the National Centre for Social and Economic Modelling (NATSEM) of the University of Canberra. She was the founder and inaugural Director of this world class Research Centre for more than sixteen years, and also a co-founder of the International Microsimulation Association (IMA) and served as the inaugural elected president of IMA from 2004 to 2011. She is a fellow of the Academy of the Social Sciences in Australia. She has more than 300 publications including several books in microsimulation modeling.

Reviews

"The book aims at introducing modern statistical small area estimation methodologies into the framework of spatial microsimulation modelling for a comprehensive presentation, providing a novel approach with much potential in comparative social research and regional statistics production. In my opinion, the strongest methodological developments are in the techniques of generating synthetic spatial microdata at small area levels. This book will be attractive for students, in economics, social sciences and statistics in particular. The increasing use of both SAE and microsimulation methods in different areas of society, such as social planning by government institutions and official or public statistics production by national and international statistical agencies. Finally, I want to congratulate the authors for writing a nice and well readable book on a quite complicated topic."
—Prof. Risto Lehtonen, University of Helsinki ". . .an interesting read for both beginning and more experienced microsimulation modellers. The two authors are well known within the microsimulation community. In this book, they share their experiences and insights into both the more theoretical and empirical aspects of microsimulation modelling. Across disciplines, there are several approaches towards the simulation or projection of small area statistics. However, since these different disciplines make use of different terminologies, there is less cross-pollination than expected (or hoped for). The aim of this book is to show and explain different approaches of small area estimation that are used in different research fields. The book gives an extensive theoretical and empirical overview of different microsimulation techniques and can be of relevance to researchers who want to expand their knowledges on ways to estimate small area characteristics."
~International Journal of Microsimulation "The authors begin with a detailed classification tree of small area estimation techniques. The text then proceeds to review and describe these techniques. A familiarity with regression techniques and survey methods is assumed throughout. The text then proceeds to present some new small area estimation techniques, validation methods, and a detailed worked example. The appendices provide further details of the worked example and SAS code for the generalized regression weighting tool (GREGWT) method."
~Douglas Dover, International Society for Clinical Biostatistics"I enjoyed reading this comprehensively written book. I recommend this book to sociologists, economists, geographers, statistics and computing professionals."
-Ramalingam Shanmugam, in Journal of Statistical Computation and Simulation, June 2019

"The book aims at introducing modern statistical small area estimation methodologies into the framework of spatial microsimulation modelling for a comprehensive presentation, providing a novel approach with much potential in comparative social research and regional statistics production. In my opinion, the strongest methodological developments are in the techniques of generating synthetic spatial microdata at small area levels. This book will be attractive for students, in economics, social sciences and statistics in particular. The increasing use of both SAE and microsimulation methods in different areas of society, such as social planning by government institutions and official or public statistics production by national and international statistical agencies. Finally, I want to congratulate the authors for writing a nice and well readable book on a quite complicated topic." ~Prof. Risto Lehtonen, University of Helsinki". . .an interesting read for both beginning and more experienced microsimulation modellers. The two authors are well known within the microsimulation community. In this book, they share their experiences and insights into both the more theoretical and empirical aspects of microsimulation modelling. Across disciplines, there are several approaches towards the simulation or projection of small area statistics. However, since these different disciplines make use of different terminologies, there is less cross-pollination than expected (or hoped for). The aim of this book is to show and explain different approaches of small area estimation that are used in different research fields. The book gives an extensive theoretical and empirical overview of different microsimulation techniques and can be of relevance to researchers who want to expand their knowledges on ways to estimate small area characteristics."
~International Journal of Microsimulation "The authors begin with a detailed classification tree of small area estimation techniques. The text then proceeds to review and describe these techniques. A familiarity with regression techniques and survey methods is assumed throughout. The text then proceeds to present some new small area estimation techniques, validation methods, and a detailed worked example. The appendices provide further details of the worked example and SAS code for the generalized regression weighting tool (GREGWT) method."
~Douglas Dover, International Society for Clinical Biostatistics"I enjoyed reading this comprehensively written book. I recommend this book to sociologists, economists, geographers, statistics and computing professionals."
-Ramalingam Shanmugam, in Journal of Statistical Computation and Simulation, June 2019