Statistical and econometric methods for transportation data analysis / Simon P. Washington, Matthew G. Karlaftis, Fred L. Mannering.

Format:

Book

Author/Creator:

Washington, Simon.

Contributor:

Karlaftis, Matthew G.

Mannering, Fred L.

Language:

English

Subjects (All):

Transportation--Statistical methods.

Transportation.

Transportation--Econometric models.

Physical Description:

425 pages : illustrations ; 24 cm

Place of Publication:

Boca Raton, Fla. : Chapman & Hall/CRC, [2003]

Summary:

Ideal as both a technical reference and as a textbook, this volume makes three unique contributions to the practice and education of transportation professionals: it presents a host of analytical techniques-both common and sophisticated; it provides a wide range of examples and case studies; and it specifically targets students and practitioners of transportation engineering and planning, urban and regional planning, and transportation economics. It will allow them to formulate research hypotheses, identify appropriate statistical and econometric models, avoid common pitfalls and misapplications of statistical methods, and interpret model results correctly.

Contents:

1 Statistical Inference I: Descriptive Statistics 3

1.1 Measures of Relative Standing 3

1.2 Measures of Central Tendency 4

1.3 Measures of Variability 5

1.4 Skewness and Kurtosis 8

1.5 Measures of Association 11

1.6 Properties of Estimators 14

1.6.1 Unbiasedness 14

1.6.2 Efficiency 15

1.6.3 Consistency 16

1.6.4 Sufficiency 17

1.7 Methods of Displaying Data 17

1.7.1 Histograms 17

1.7.2 Ogives 17

1.7.3 Box Plots 18

1.7.4 Scatter Diagrams 19

1.7.5 Bar and Line Charts 19

2 Statistical Inference II: Interval Estimation, Hypothesis Testing, and Population Comparisons 23

2.1 Confidence Intervals 23

2.1.1 Confidence Interval for m with Known [sigma superscript 2] 24

2.1.2 Confidence Interval for the Mean with Unknown Variance 26

2.1.3 Confidence Interval for a Population Proportion 27

2.1.4 Confidence Interval for the Population Variance 27

2.2 Hypothesis Testing 28

2.2.1 Mechanics of Hypothesis Testing 30

2.2.2 Formulating One- and Two-Tailed Hypothesis Tests 32

2.2.3 The p-Value of a Hypothesis Test 34

2.3 Inferences Regarding a Single Population 35

2.3.1 Testing the Population Mean with Unknown Variance 35

2.3.2 Testing the Population Variance 36

2.3.3 Testing for a Population Proportion 37

2.4 Comparing Two Populations 38

2.4.1 Testing Differences between Two Means: Independent Samples 38

2.4.2 Testing Differences between Two Means: Paired Observations 41

2.4.3 Testing Differences between Two Population Proportions 42

2.4.4 Testing the Equality of Two Population Variances 43

2.5 Nonparametric Methods 45

2.5.1 The Sign Test 46

2.5.2 The Median Test 50

2.5.3 The Mann-Whitney U Test 51

2.5.4 The Wilcoxon Signed-Rank Test for Matched Pairs 54

2.5.5 The Kruskal-Wallis Test 55

2.5.6 The Chi-Square Goodness-of-Fit Test 56

Part II Continuous Dependent Variable Models

3 Linear Regression 63

3.1 Assumptions of the Linear Regression Model 63

3.1.1 Continuous Dependent Variable Y 64

3.1.2 Linear-in-Parameters Relationship between Y and X 64

3.1.3 Observations Independently and Randomly Sampled 65

3.1.4 Uncertain Relationship between Variables 65

3.1.5 Disturbance Term Independent of X and Expected Value Zero 65

3.1.6 Disturbance Terms Not Autocorrelated 66

3.1.7 Regressors and Disturbances Uncorrelated 66

3.1.8 Disturbances Approximately Normally Distributed 66

3.2 Regression Fundamentals 67

3.2.1 Least Squares Estimation 68

3.2.2 Maximum Likelihood Estimation 73

3.2.3 Properties of OLS and MLE Estimators 74

3.2.4 Inference in Regression Analysis 75

3.3 Manipulating Variables in Regression 79

3.3.1 Standardized Regression Models 79

3.3.2 Transformations 80

3.3.3 Indicator Variables 82

3.3.3.1 Estimate a Single Beta Parameter 83

3.3.3.2 Estimate Beta Parameter for Ranges of the Variable 83

3.3.3.3 Estimate a Single Beta Parameter for m - 1 of the m Levels of the Variable 84

3.3.4 Interactions in Regression Models 84

3.4 Checking Regression Assumptions 87

3.4.1 Linearity 88

3.4.2 Homoscedastic Disturbances 91

3.4.3 Uncorrelated Disturbances 93

3.4.4 Exogenous Independent Variables 93

3.4.5 Normally Distributed Disturbances 96

3.5 Regression Outliers 99

3.5.1 The Hat Matrix for Identifying Outlying Observations 99

3.5.2 Standard Measures for Quantifying Outlier Influence 101

3.5.3 Removing Influential Data Points from the Regression 102

3.6 Regression Model Goodness-of-Fit Measures 106

3.7 Multicollinearity in the Regression 110

3.8 Regression Model-Building Strategies 112

3.8.1 Stepwise Regression 113

3.8.2 Best Subsets Regression 113

3.8.3 Iteratively Specified Tree-Based Regression 113

3.9 Logistic Regression 114

3.10 Lags and Lag Structure 115

3.11 Investigating Causality in the Regression 117

3.12 Limited Dependent Variable Models 118

3.13 Box-Cox Regression 119

3.14 Estimating Elasticities 120

4 Violations of Regression Assumptions 121

4.1 Zero Mean of the Disturbances Assumption 121

4.2 Normality of the Disturbances Assumption 122

4.3 Uncorrelatedness of Regressors and Disturbances Assumption 123

4.4 Homoscedasticity of the Disturbances Assumption 125

4.4.1 Detecting Heteroscedasticity 127

4.4.2 Correcting for Heteroscedasticity 128

4.5 No Serial Correlation in the Disturbances Assumption 132

4.5.1 Detecting Serial Correlation 135

4.5.2 Correcting for Serial Correlation 136

4.6 Model Specification Errors 139

5 Simultaneous Equation Models 143

5.1 Overview of the Simultaneous Equations Problem 143

5.2 Reduced Form and the Identification Problem 144

5.3 Simultaneous Equation Estimation 146

5.3.1 Single-Equation Methods 146

5.3.2 System Equation Methods 147

5.4 Seemingly Unrelated Equations 150

5.5 Applications of Simultaneous Equations to Transportation Data 152

Appendix 5A A Note on Generalized Least Squares Estimation 152

6 Panel Data Analysis 155

6.1 Issues in Panel Data Analysis 155

6.2 One-Way Error Component Models 157

6.2.1 Heteroscedasticity and Serial Correlation 160

6.3 Two-Way Error Component Models 161

6.4 Variable Coefficient Models 166

6.5 Additional Topics and Extensions 168

7 Time-Series Analysis 169

7.1 Characteristics of Time Series 169

7.1.1 Long-Term Movements 170

7.1.2 Seasonal Movements 170

7.1.3 Cyclic Movements 171

7.1.4 Irregular or Random Movements 171

7.2 Smoothing Methodologies 171

7.2.1 Simple Moving Averages 172

7.2.2 Exponential Smoothing 173

7.3 The ARIMA Family of Models 177

7.3.1 The ARIMA Models 180

7.3.2 Estimating ARIMA Models 182

7.4 Nonlinear Time-Series Models 184

7.4.1 Conditional Mean Models 184

7.4.2 Conditional Variance Models 185

7.4.3 Mixed Models 187

7.4.4 Regime Models 187

7.5 Multivariate Time-Series Models 188

7.6 Measures of Forecasting Accuracy 189

8 Latent Variable Models 193

8.1 Principal Components Analysis 194

8.2 Factor Analysis 199

8.3 Structural Equation Modeling 204

8.3.1 Basic Concepts in Structural Equation Modeling 204

8.3.2 The Structural Equation Model 208

8.3.3 Non-Ideal Conditions in the Structural Equation Model 210

8.3.4 Model Goodness-of-Fit Measures 211

8.3.5 Guidelines for Structural Equation Modeling 213

9 Duration Models 217

9.1 Hazard-Based Duration Models 217

9.2 Characteristics of Duration Data 221

9.3 Nonparametric Models 222

9.4 Semiparametric Models 223

9.5 Fully Parametric Models 226

9.6 Comparisons of Nonparametric, Semiparametric, and Fully Parametric Models 229

9.7 Heterogeneity 231

9.8 State Dependence 233

9.9 Time-Varying Covariates 234

9.10 Discrete-Time Hazard Models 235

9.11 Competing Risk Models 236

Part III Count and Discrete Dependent Variable Models

10 Count Data Models 241

10.1 Poisson Regression Model 241

10.2 Poisson Regression Model Goodness-of-Fit Measures 243

10.3 Truncated Poisson Regression Model 247

10.4 Negative Binomial Regression Model 248

10.5 Zero-Inflated Poisson and Negative Binomial Regression Models 250

10.6 Panel Data and Count Models 254

11 Discrete Outcome Models 257

11.1 Models of Discrete Data 257

11.2 Binary and Multinomial Probit Models 258

11.3 Multinomial Logit Model 260

11.4 Discrete Data and Utility Theory 264

11.5 Properties and Estimation of Multinomial Logit Models 266

11.5.1 Statistical Evaluation 269

11.5.2 Interpretation of Findings 271

11.5.3 Specification Errors 273

11.5.4 Data Sampling 278

11.5.5 Forecasting and Aggregation Bias 280

11.5.6 Transferability 282

11.6 Nested Logit Model (Generalized Extreme Value Model) 283

11.7 Special Properties of Logit Models 286

11.8 Mixed MNL Models 287

11.9 Models of Ordered Discrete Data 288

12 Discrete/Continuous

Models 297

12.1 Overview of the Discrete/Continuous Modeling Problem 297

12.2 Econometric Corrections: Instrumental Variables and Expected Value Method 299

12.3 Econometric Corrections: Selectivity-Bias Correction Term 302

12.4 Discrete/Continuous Model Structures 305

Appendix D Variable Transformations 389.

Notes:

Includes bibliographical references (pages 395-412) and index.

ISBN:

1584880309

OCLC:

51923236