Bias and Causation : Models and Judgment for Valid Comparisons

by Herbert I. Weisberg (Correlation Research, Inc, USA )

• ISBN13: 9780470631096
• ISBN10: 0470631090
• Format: eBook
• Copyright: 2010-07-01
• Publisher: Wiley

Summary

A one-of-a-kind resource on identifying and dealing with bias in statistical research on causal effectsDo cell phones cause cancer? Can a new curriculum increase student achievement? Determining what the real causes of such problems are, and how powerful their effects may be, are central issues in research across various fields of study. Some researchers are highly skeptical of drawing causal conclusions except in tightly controlled randomized experiments, while others discount the threats posed by different sources of bias, even in less rigorous observational studies. Bias and Causation presents a complete treatment of the subject, organizing and clarifying the diverse types of biases into a conceptual framework. The book treats various sources of bias in comparative studies both randomized and observational and offers guidance on how they should be addressed by researchers.Utilizing a relatively simple mathematical approach, the author develops a theory of bias that outlines the essential nature of the problem and identifies the various sources of bias that are encountered in modern research. The book begins with an introduction to the study of causal inference and the related concepts and terminology. Next, an overview is provided of the methodological issues at the core of the difficulties posed by bias. Subsequent chapters explain the concepts of selection bias, confounding, intermediate causal factors, and information bias along with the distortion of a causal effect that can result when the exposure and/or the outcome is measured with error. The book concludes with a new classification of twenty general sources of bias and practical advice on how mathematical modeling and expert judgment can be combined to achieve the most credible causal conclusions.Throughout the book, examples from the fields of medicine, public policy, and education are incorporated into the presentation of various topics. In addition, six detailed case studies illustrate concrete examples of the significance of biases in everyday research.Requiring only a basic understanding of statistics and probability theory, Bias and Causation is an excellent supplement for courses on research methods and applied statistics at the upper-undergraduate and graduate level. It is also a valuable reference for practicing researchers and methodologists in various fields of study who work with statistical data. Send Comment

Table of Contents

What is Bias?

Apples and Oranges

Statistics vs. Causation

Bias in the Real World

Causality and Comparative Studies

Bias and Causation

Causality and Counterfactuals

Why Counterfactuals?

Causal Effects

Empirical Effects

Empirical Effects and Bias

External Validity

Measures of Empirical Effects

The Difference of Means

The Risk Difference and Risk Ratio

Potential Outcomes

Time-Dependent Outcomes

Intermediate Variables

Measurement of Exposure Status

Measurement of the Outcome Value

Confounding Bias

Varieties of Bias

Research Designs

Bias in Biomedical Research

Bias in Social Science Research

Sources of Bias

Selection Bias

Selection Processes and Bias

Traditional Selection Model: Dichotomous Outcome

Causal Selection Model: Dichotomous Outcome

Randomized Experiments

Observational Cohort Studies

Traditional Selection Model: Numerical Outcome

Causal Selection Model: Numerical Outcome

Appendix to Chapter 5

Confounding: An Enigma

What is the Real Problem?

Confounding and Extraneous Causes

Confounding and Statistical Control

Confounding and Comparability

Confounding and the Assignment Mechanism

Confounding and Model Specification

Confounding : Essence, Correction and Detection

Essence: The Nature of Confounding

Correction: Statistical Control for Confounding

Detection: Adequacy of Statistical Adjustment

Appendix to Chapter 7

Intermediate Causal Factors

Direct and Indirect Effects

Principal Stratification

Noncompliance

Attrition

Information Bias

Basic Concepts

Classical Measurement Model: Dichotomous Outcome

Causal Measurement Model: Dichotomous Outcome

Classical Measurement Model: Numerical Outcome

Causal Measurement Model: Numerical Outcome

Covariates Measured with Error

Sources of Bias

Sampling

Assignment

Adherence

Exposure Ascertainment

Outcome Measurement

Contending with Bias

Conventional Solutions

The Standard Statistical Paradigm

Toward a Broader Perspective

Real-World Bias Revisited

Statistics and Causation

Table of Contents provided by Publisher. All Rights Reserved.