highly significantp -value < 2.2 × 10 -16
All fitted models were highly significant ( F 5,122 = 26.1, p -value < 2.2 × 10 -16 with ‖ Δ ^ ̄ − Δ ‖ as response; F 5,122 = 47.7, p -value < 2.2 × 10 -16 with λ ^ ̄ as response; and F 5,122 = 12.1, p -value = 1.6 × 10 -9 with | ε ^ g e n − ε ̃ g e n | ̄ as response); residual plots showed no large deviations from the assumptions of normality of error distribution (an asymptotic normal distribution of the response variables is warranted by the central limit theorem), homoscedasticity, and independent errors (data not shown).