Omitted Variable Bias
This topic teaches students one of the key principles of social science research – that correlation between two variables does not necessarily imply a causal relationship. Students learn about the challenges in generating useful estimates of the causal impact of one variable and another in a regression analysis that flow from the fact that it is usually impossible to include all relevant variables in the analysis. Students learn the difference between correlation and causation, and learn about the strategies researchers use to try to reduce or eliminate omitted variable bias to produce reasonable estimates of the extent of causal relationships.
Topic Learning Outcome: Students will have sophisticated understanding of the difference between correlation and causation, will understand how omitted variable bias can make it difficult to measure causal relationships and understand techniques that can be used to attempt to reduce omitted variable bias.
Core Concepts associated with this Topic: Environmental Validity; External Validity; Hawthorne Effect; John Henry Effect; Natural Experiment; Omitted Variable Bias; Selection Bias.
University of Toronto: PPG-1004
Stock, James H. and Mark W. Watson. 2011. Introduction to Econometrics, 3rd ed. Pearson/Addison-Wesley. Chapters 8 (p. 252-296) and 9 (p. 312-341).
NYU Wagner: GP-1011
Jan Blustein. SPSS The Wagner Way. (This seems to be some internally distributed textbook only available for download off Wagner’s blackboard). Chapters 8 and 9.
George Washington: PPPA-6002
De Veaux, Richard D., Paul F. Velleman, and David E. Bock. Stats: Data and Models. Boston: Pearson/Addison Wesley, 2005. Chapter 25.
Moore, David S (2011). The Basic Practice of Statistics, 5th edition. W.H. Freeman and Company. Chapters 4 and 5.
Wooldridge, Jeffrey. Introductory Econometrics: A Modern Approach. Cengage Learning, 2012. Chapter 15 (the sections “controlling for a third variable” and “where do control variables come from?”) and Chapter 16.
Rutgers: Analytical Methods I and Analytical Methods II
Wang, XiaoHu (201). Performance analysis for public and nonprofit organizations. Jones & Bartlett Learning. Chapter 10.
Berman, Evan, and XiaoHu Wang (2011). Essential statistics for public managers and policy analysts. CQ Press. Chapter 14, 15.
Healey (2011). Statistics: A Tool for Social Research, 9th Edition. Cengage Learning. Chapter 16.
Schroeder, Larry D., Sjoquist, David L., & Stephan, Paula E. (1986). “Understanding Regression Analysis: An Introductory Guide”. Quantitative Applications in the Social Sciences 57. Newbury Park: Sage Publications.
Possible Assessment Questions:
1.) What is meant by the saying “correlation does not equal causation?”
2.) Define omitted variable bias.
3.) Describe two strategies that researchers can use in an attempt to deal with omitted variable bias and produce an estimate of the causal effect of one variable on another.
Page Created By: Ben Eisen and Joshua Tan, last edited 4 November, 2014.