Statistical and Econometric Methods for Transportation Data Analysis

Simon Washington, Matthew G. Karlaftis, Fred Mannering, Panagiotis Anastasopoulos

• Publisher: Chapman and Hall/CRC
• Year: 2020
• Print ISBN: 9780367199029
• eBook ISBN: 9780429534225
• Topic: Technology & Engineering
• Subtopic: Probability & Statistics
• Index: Technology & Engineering

The book's website (with databases and other support materials) can be accessed here.

Praise for the Second Edition:

The second edition introduces an especially broad set of statistical methods … As a lecturer in both transportation and marketing research, I find this book an excellent textbook for advanced undergraduate, Master's and Ph.D. students, covering topics from simple descriptive statistics to complex Bayesian models. … It is one of the few books that cover an extensive set of statistical methods needed for data analysis in transportation. The book offers a wealth of examples from the transportation field. —The American Statistician

Statistical and Econometric Methods for Transportation Data Analysis, Third Edition

offers an expansion over the first and second editions in response to the recent methodological advancements in the fields of econometrics and statistics and to provide an increasing range of examples and corresponding data sets. It describes and illustrates some of the statistical and econometric tools commonly used in transportation data analysis. It provides a wide breadth of examples and case studies, covering applications in various aspects of transportation planning, engineering, safety, and economics. Ample analytical rigor is provided in each chapter so that fundamental concepts and principles are clear and numerous references are provided for those seeking additional technical details and applications.

New to the Third Edition

• Updated references and improved examples throughout.
• New sections on random parameters linear regression and ordered probability models including the hierarchical ordered probit model.
• A new section on random parameters models with heterogeneity in the means and variances of parameter estimates.
• Multiple new sections on correlated random parameters and correlated grouped random parameters in probit, logit and hazard-based models.
• A new section discussing the practical aspects of random parameters model estimation.
• A new chapter on Latent Class Models.
• A new chapter on Bivariate and Multivariate Dependent Variable Models.

Statistical and Econometric Methods for Transportation Data Analysis, Third Edition

can serve as a textbook for advanced undergraduate, Masters, and Ph.D. students in transportation-related disciplines including engineering, economics, urban and regional planning, and sociology. The book also serves as a technical reference for researchers and practitioners wishing to examine and understand a broad range of statistical and econometric tools required to study transportation problems.

Section II

Continuous Dependent Variable Models

3

Linear Regression

Linear regression is one of the most widely studied and applied statistical and econometric techniques, for numerous reasons. First, linear regression is suitable for modeling a wide variety of relationships between variables. In addition, the assumptions of linear regression models are often suitably satisfied in many practical applications. Furthermore, regression model outputs are relatively easy to interpret and communicate to others, numerical estimation of regression models is relatively easy, and software for estimating models is readily available in numerous “non-specialty” software packages. Linear regression can also be overused or misused. In some cases, the assumptions are not strictly met, and suitable alternatives are not known, understood, or applied. Moreover, more advanced techniques may require specialized software and knowledge to estimate.

It should not be surprising that linear regression serves as an excellent starting point for illustrating statistical model estimation procedures. Although it is a flexible and useful tool, applying linear regression when other methods are more suitable should be avoided.

This chapter illustrates the estimation of linear regression models, explains when linear regression models are appropriate, describes how to interpret linear regression model outputs, and discusses how to select among a competing set of linear regression models. Matrix algebra is used throughout the chapter to illustrate the concepts, but only to the extent necessary to illustrate the most important analytical aspects.

3.1 Assumptions of the Linear Regression Model

Linear regression is used to model a linear relationship between a continuous dependent variable and one or more independent variables. Most regression applications seek to identify a set of explanatory variables that are thought to co-vary with the dependent variable. In general, explanatory or “causal” models are based on data obtained from well-controlled experiments (e.g., those conducted in a laboratory), predictive models are based on data obtained from observational studies, and quality control models are based on data obtained from a process or system being controlled. Whether explanatory variables cause or are merely associated with changes in the dependent variable depends on numerous factors and cannot be determined on the basis of statistical modeling alone.

There are numerous assumptions (or requirements) of the linear regression model. When any of the requirements are not met, remedial actions should be taken, and in some cases, alternative modeling approaches adopted. The following are the assumptions of the linear regression model.

3.1.1 Continuous Dependent Variable Y

An assumption in regression is that the dependent or response variable is continuous; that is, it can take on any value within a range of values. A continuous variable is measured on either ...

Table of contents

• Cover
• Half Title
• Series Page
• Title Page
• Copyright Page
• Table of Contents
• Preface
• Authors
• Section I Fundamentals
• Section II Continuous Dependent Variable Models
• Section III Count and Discrete-Dependent Variable Models
• Section IV Other Statistical Methods
• Appendix A: Statistical Fundamentals
• Appendix B: Statistical Tables
• Appendix C: Variable Transformations
• References
• Index