Table 2.

Root mean squared error of fitted values in routine and cross-validated analyses.

Heuristic assumptions	Error of fitted values		Error of fitted values – cross-validated
	Metric	Value	Metric	Value
No heuristic	Mx-S	0.432 (0.450)	Pr-C	0.502 (0.518)
	No alternates		Mx-S	0.507
			Pr-I	0.513
Free heuristic – impervious	Mx-S	0.387 (0.413)	Pr-C	0.478 (0.495)†
	No alternates		Pr-S	0.493
Free heuristic – vulnerable	Mx-S	0.377 (0.402)*	Pr-C	0.467 (0.483)*
	No alternates		Nno alternates
Obligatory heuristic – impervious	Mx-S	6.287 (6.451)	Mx-T	6.267 (6.431)
	All other metrics were within 95% CI		All other metrics were within 95% CI
Obligatory heuristic – vulnerable	Pr-I	0.793 (0.828)	Pr-I	0.857 (0.894)
	Mx-S	0.803	No alternates
	Mx-C	0.828

Numbers in parentheses represent the upper limits of the 95% confidence interval (CI). For each heuristic assumption, the complexity metric that provides the best fit is listed (unshaded) and other metrics that fell within the 95% CI of this metric are listed below (shaded).

*Indicates the best fit (lowest score) across all heuristic assumptions within a column.

^†Indicates a measure that falls within the 95% CI of the lowest error measurement within the same column.