Table 1.

Comparison of scale-free and alternative distributions

		Test outcome
Alternative	p(x)∝f(x)	M P L	Inconclusive	M A lt
Exponential	e − λx	33%	26%	41%
Log-normal	$\frac{1}{x}{e}^{-\frac{{\left(logx-\mu \right)}^2}{2{\sigma }^2}}$	12%	40%	48%
Weibull	$e^{-{\left(\frac{x}{b}\right)}^a}$	33%	20%	47%
Power law with cutoff	x − α e − λx	–	44%	56%

The percentage of network data sets that favor the power-law model M_PL, alternative model M_Alt, or neither, under a likelihood-ratio test, along with the form of the alternative distribution f(x)