The additive model affected by missing completely at random in the covariate (Q1775956)

This paper focuses on a comparison of several imputation procedures within the simple additive model \(y=f(x)+\varepsilon\), where the independent variable \(x\) is affected by missing completely randomly. Such imputation methods as complete case analysis, mean imputation plus random noise, single imputation and two kinds of nearest neighbour imputations are investigated and compared within a simulation experiment based on the sample mean squared error, estimated variances and sample bias of the estimates of \(f(x)\) at the knots.