Specification (regression)

From Wikipedia, the free encyclopedia
Jump to: navigation, search

In regression analysis specification is the process of developing a regression model. This process consists of selecting an appropriate functional form for the model and choosing which variables to include. As a first step of regression analysis, a person specifies the model. If an estimated model is misspecified, it will be biased and inconsistent.[1]

Specification error and bias[edit]

Specification error occurs when an independent variable is correlated with the error term. There are several different causes of specification error:

  • incorrect functional form
  • a variable omitted from the model may have a relationship with both the dependent variable and one or more of the independent variables (omitted-variable bias);[2]
  • an irrelevant variable may be included in the model
  • the dependent variable may be part of a system of simultaneous equations (simultaneity bias)
  • measurement errors may affect the independent variables.


The Ramsey RESET test can help test for specification error.

See also[edit]


  1. ^ Lee, Cheng Few; Lee, John C.; Lee, Alice C. (1999). Statistics for Business and Financial Economics (2nd ed.). World Scientific Publishing Company. p. 718. ISBN 981-02-3485-6. 
  2. ^ Untitled
  • MacKinnon, James G. (1992). "Model Specification Tests and Artificial Regressions". Journal of Economic Literature 30 (1): 102–146. JSTOR 2727880.  edit
  • Asteriou, Dimitrios; Hall, Stephen G. (2011). "Misspecification: Wrong Regressors, Measurement Errors and Wrong Functional Forms". Applied Econometrics (Second ed.). London: Palgrave MacMillan. pp. 172–197. 
  • Thursby, Jerry G.; Schmidt, Peter (September 1977). "Some Properties of Tests for Specification Error in a Linear Regression Model". Journal of the American Statistical Association 72 (359): 635–641. JSTOR 2286231. 
  • Sapra, Sunil (2005). "A regression error specification test (RESET) for generalized linear models" (PDF). Economics Bulletin 3 (1): 1–6.