“…Fixed-effects models are also inferior to mixed-effect models when the aim is to make inferences about the population. However, several studies have shown that a fixed-effects model results in a smaller root mean square error (RMSE) than using only the fixed part of a mixed-effects model (Temesgen, Monleon, & Hann, 2008;Garber, Temesgen, Monleon, & Hann, 2009;Pukkala et al, 2009;Shater, de-Miguel, Kraid, Pukkala, & Palahí, 2011;Groom, Hann, & Temesgen, 2012;Guzmán, Morales, Pukkala, & de-Miguel, 2012a, Guzmán, Pukkala, Palahí, & de-Miguel, 2012bHeiðarsson & Pukkala, 2012). De-Miguel, Guzmán, & Pukkala (2013) found that, in the absence of calibration data, mixed-effects models should be used with caution.…”