Table 1.

This table gives the percentage of fits that favor the first of the two listed models based on a particular criterion for each study. For example, the number in the upper left indicates that Akaike information criterion (AIC) and Bayesian information criterion (BIC) favored the bi-gamma model over the bi-normal model for 60.5% of the power-law simulations. These numbers are the percentages of each distribution in Fig. 2 below the finely dotted lines (AIC), coarsely dotted lines (BIC), and dashed lines [95% confidence interval (CI)]. In Sec. 3.2, we approximated the values in the asterisk cells as 15.7, 2.6, and 5.0%.

Models	Criteria	${\mathrm{P}}_1$ Sim	${\mathrm{N}}_1$ Sim	VCRS	DBTRS	VDTRS	DRRS	BMRS	MRIRS
${\mathrm{G}}_2\text{-}{\mathrm{N}}_2$	A/BIC	60.5	22.2	12.5	75.0	34.7	53.6	50.0	70.0
${\mathrm{N}}_1\text{-}{\mathrm{P}}_1$	A/BIC	13.3	83.2	50.0	7.1	59.4	3.6	46.7	10.0
${\mathrm{N}}_2\text{-}{\mathrm{P}}_1$	AIC	15.4	71.9	62.5	10.7	38.6	0.0	33.3	0.0
	BIC	2.8	45.8	50.0	0.0	11.9	0.0	6.7	0.0
${\mathrm{G}}_2\text{-}{\mathrm{P}}_1$	AIC	18.5*	71.6	50.0	14.3	35.6	7.1	26.7	0.0
	BIC	3.5*	42.6	25.0	0.0	7.9	3.6	6.7	0.0
	CI	6.0*	52.2	25.0	0.0	9.9	3.6	10.0	0.0
${\mathrm{N}}_2\text{-}{\mathrm{N}}_1$	AIC	76.1	16.8*	37.5	92.9	32.7	64.3	30.0	70.0
	BIC	50.6	2.9*	37.5	42.9	15.8	21.4	13.3	30.0
	CI	59.1	5.4*	37.5	82.1	17.8	35.7	13.3	60.0
${\mathrm{G}}_2\text{-}{\mathrm{N}}_1$	AIC	78.2	10.2	37.5	92.9	26.7	64.3	30.0	80
	BIC	53.2	2.1	37.5	46.4	11.9	25.0	13.3	40
Number of fits		4010	4010	8	28	101	28	30	10