misspecification
n. the situation in which the number of variables, factors, parameters, or some combination of these was not correctly specified in a statistical model, with the result that the model does not offer a reasonable representation of obtained data. Misspecification is indicated by such things as large residuals between the data and that predicted by the model, large standard errors for the parameter estimates, values for the parameter estimates outside of the normal range (e.g., Heywood cases), or a nonsignificant result from a goodness-of-fit test.