The Goldilocks Principle: Avoiding Pitfalls in Interpretation of Regression Coefficients
Jane E. Miller
Rutgers, The State University of New Jersey - Institute for Health, Health Care Policy and Aging Research
Regression coefficients estimate the effect of a one-unit increase in the independent variable X on the dependent variable Y. Although a one-unit increase is mathematically convenient, it is an arbitrary choice for interpreting the substantive significance of a coefficient. Depending on the type of variable and its level and distribution, a one-unit increase is too big for some independent variables, too small for others, and just right for yet others. This paper discusses theoretical and empirical criteria for choosing the right size contrast for each Xi to convey the size and shape of its relationship with Y. Transformation of variables and model specification are also considered. Example sentences are used to illustrate how to interpret coefficients in the context of the research question and data.
Number of Pages in PDF File: 15
JEL Classification: A23, C10working papers series
Date posted: July 1, 2008
© 2013 Social Science Electronic Publishing, Inc. All Rights Reserved.
This page was processed by apollo3 in 0.390 seconds