A Casebook for Spatial Statistical Data Analysis A Compilation of Analyses of Different Thematic Data Sets

by Griffith, Daniel A.; Layne, Larry J.

• ISBN13: 9780195109580
• ISBN10: 0195109589
• Format: Hardcover
• Copyright: 1999-12-02
• Publisher: Oxford University Press

Summary

This volume compiles geostatistical and spatial autoregressive data analyses involving georeferenced socioeconomic, natural resources, agricultural, pollution, and epidemiological variables. Benchmark analyses are followed by analyses of readily available data sets, emphasizing parallels between geostatistical and spatial autoregressive findings. Both SAS and SPSS code are presented for implementation purposes. This informative casebook will serve geographers, regional scientists, applied spatial statisticians, and spatial scientists from across disciplines.

Author Biography

Daniel A. Griffith is Professor of Geography at Syracuse University.

Table of Contents

List of Data Sets Analyzed	xv
Abbreviations	xvi
PART I: THEORETICAL BACKGROUND
Introduction	3	(66)
Parallels between Spatial Autoregression and Geostatistics	6	(3)
The Many Faces of Spatial Autocorrelation	9	(6)
The Moran Coefficient Scatterplot Tool	15	(3)
Multivariate Spatial Association	18	(7)
Heterogeneity and Locational Information	25	(9)
The Semivariogram Plot Tool	34	(5)
Computer Code for Implementing Spatial Statistical Analyses	39	(30)
SAS Code for Computing MC Using Standard Regression Techniques.	62	(2)
Directions for Using ArcInfo to Construct a Thiessen Polygon Surface Partitioning for a Set of Georeferenced Points.	64	(3)
SAS Macros Used for Converting among Degrees, Decimal Degrees and Radians	67	(2)
Important Modeling Assumptions	69	(51)
The Relative Importance of the Principal Assumptions	71	(2)
Variable Transformations	73	(2)
Transforming in Search of Normality	75	(4)
Transforming in Search of Constant Variance	79	(4)
Linearity: Exploitation of Linear Relationships by Linear Statistical Models	83	(3)
An Absence of Independence: The Presence of Spatial Autocorrelation in Georeferenced Data	86	(19)
Exploring Residuals in Spatial Analysis	105	(6)
Other Statistical Frequency Distribution Assumptions	111	(9)
Popular Spatial Autoregressive and Geostatistical Models	120	(85)
Spatial Autoregressive Models	121	(12)
Geostatistical Models	133	(9)
Articulating Relationships Between Spatial Autoregressive and Geostatistical Models	142	(11)
Computer Code for Spatial Autoregressive and Semivariogram Modeling	153	(52)
SAS Code for Estimating Equations (3.7a-c)---the CAR Model	165	(4)
SPSS Code for Estimating Equations (3.7a-c)---the CAR Model	169	(3)
SAS Code for Estimating Equations (3.8) & (3.9)---the SAR & AR Models	172	(3)
SPSS Code for Estimating Equations (3.8) & (3.9)---the SAR & AR Models	175	(3)
SAS Code for Selected Semivariogram Models	178	(16)
SPSS Code for Selected Semivariogram Models	194	(11)
PART II: GEOREFERENCED DATA SET CASE STUDIES	205	(222)
Analysis of Georeferenced Socioeconomic Attribute Variables	207	(28)
The Cliff-Ord Eire Population Data	211	(3)
Urban Population Density	214	(4)
Residential Insurance Coverage in Chicago	218	(4)
Urban Crime in Columbus, Ohio	222	(5)
Geographic Distribution of Minorities across Syracuse, New York	227	(5)
Concluding Comments: Spatial Autocorrelation and Socioeconomic Attribute Variables	232	(3)
Centroids Derived from a Digitized Version of Cliff and Ord's Map of Eire (1981, 207) Using ArcInfo	234	(1)
Analysis of Georeferenced Natural Resources Attribute Variables	235	(70)
Kansas Oil Wells Data	241	(4)
Natural Resources Inventory Data	245	(20)
Island Biogeography: Plant Species Data	265	(7)
Weather Station Rainfall Data	272	(5)
Drainage Basin Runoff Data	277	(7)
Digital Elevation Data	284	(8)
Preclassified Remotely Sensed Image Reflectance Data	292	(10)
Concluding Comments: Spatial Autocorrelation and Natural Resources Attribute Variables	302	(3)
Analysis of Georeferenced Agricultural Yield Variables	305	(48)
The Mercer-Hall Straw Yield Data	311	(5)
The Wiebe Wheat Yield Data	316	(4)
The Broadbalk Wheat and Straw Yield Data	320	(10)
Sugar Cane Production in Puerto Rico	330	(7)
Milk Production in Puerto Rico	337	(13)
Concluding Comments: Spatial Autocorrelation and Agricultural Yield Variables	350	(3)
Analysis of Georeferenced Pollution Variables	353	(43)
Southwestern Pennsylvania Coal Ash	360	(5)
EMAP Indicators of Ecological Condition	365	(8)
Great Smoky Mountains Water pH	373	(5)
Chemical Elements in Northwest Texas Groundwater	378	(12)
Hazardous Waste Contamination of Soil: Dioxin	390	(3)
Concluding Comments: Spatial Autocorrelation and Pollution Variables	393	(3)
Analysis of Georeferenced Epidemiological Variables	396	(31)
Glasgow Standardized Mortality Rates	405	(5)
Pediatric Lead Poisoning in Syracuse, New York	410	(7)
Fox Rabies in Germany	417	(6)
Concluding Comments: Spatial Autocorrelation and Epidemiological Variables	423	(4)
PART III: VISUALIZING WHAT IS NOT OBSERVED	427	(57)
Exploding Georeferenced Data When Maps Have Holes or Gaps: Estimating Missing Data Values and Kriging	428	(29)
An Introduction to EM Estimation	430	(2)
Estimating Missing Values: Two Simplified Georeferenced Data Illustrations	432	(6)
Estimating a Conspicuous Missing Data Value for the Coal-Ash Data Set	438	(3)
Estimating Conspicuous Missing Data Values for an Agricultural Experiment	441	(4)
Estimating Missing Median Family Income Data for Ottawa-Hull	445	(2)
Generalizing a Map Surface with Kriging	447	(4)
A Cross-Validation Example	451	(3)
Concluding Comments: Exploding Georeferenced Data	454	(3)
Concluding Comments	457	(27)
More about the Nature of Georeferenced Data	459	(1)
Reflections on Spatial Data Model Specifications	460	(9)
Implications regarding Relations between Spatial Autoregressive and Geostatistical Models	469	(4)
Reflections on Kriging	473	(2)
Spatial Statistics and GIS	475	(1)
Now Is the Time for All Good Spatial Scientists to	476	(3)
Some Questions Yet Unanswered: Future Research	479	(5)
Epilogue	484	(4)
References	488	(15)
Subject Index	503