Reduced χ² using the errors in the χ² definition (χ²(error)/d.o.f., see Eq. (A.1)), versus that using the model in the χ² definition (χ²(model)/d.o.f., see Eq. (A.2)). Filled and empty symbols show the values when foreground extinction is included and neglected in the spectral fitting procedure, respectively. The dot-dashed line shows the locus expected if both measurements agree with each other. The continuous lines show a factor of five of agreement between both χ² definitions. An object located in the blank area shows complexity not well fitted by the model. The dashed area shows the locus when the model is too complex for the data (χ²/d.o.f. < 0.5). The dark-gray shaded area shows the locus where both reduced χ² indicate that the model provides a statistically acceptable fit (i.e., χ²/d.o.f.(error) < 2 and χ²/d.o.f.(model) < 1.5 × 10⁻³). The light-gray shaded area shows the region where the object is expected to locate if the error bars of the spectra are overestimated and underestimated although the spectral fit is good. Each panel shows the results for one of the models. We show the percentage of good spectral fits per model, considering those objects in the gray areas (the result without foreground extinction is shown next to it). The percentage of objects statistically requesting foreground extinction (FG extinct.) is shown in the top-left corner of each figure.