Specification error
Specification error is a mistake in a regression model's design: omitting relevant variables, including irrelevant ones, or choosing the wrong functional form. It can bias or distort the estimated coefficients.
Also known asmodel misspecification
FrameworkOrdinary least squares (OLS)
See it move
A tree diagram branches from the root label 'Specification error' into three distinct types of model mis-specification, each with its own consequence. Omitting a relevant variable makes OLS estimates inconsistent and biased. Including an irrelevant variable leaves estimates unbiased but inflates standard errors, reducing precision. Using the wrong functional form produces residuals that follow a curved pattern rather than the random scatter that valid OLS inference requires.