Table 2.

Burst size CDF model-likelihood tests

Infant	Power law p	Power law with cut-off		Log-normal		Stretched exponential		Distribution
Infant	Power law p	LLR	p	LLR	p	LLR	p	Distribution
1	0	−56.2	0	−24.2	1.6 × 10⁻⁶	−28.6	5.0 × 10⁻⁷	Power law with cut-off
2	0.16	−0.157	0.57	−0.112	0.79	−0.0282	0.97	Power law
3	0	−24.3	3.2 × 10⁻¹²	−19.5	6.8 × 10⁻⁵	−21.6	3.5 × 10⁻⁵	Power law with cut-off
4	0	−31.8	1.4 × 10⁻¹⁵	−29.4	5.1 × 10⁻⁷	−31.9	1.8 × 10⁻⁷	Power law with cut-off/stretched exponential
5	0.80	−6.34 × 10⁻³	0.91	−0.746	0.11	0.207	0.50	Power law
6	0.030	−9.05	2.1 × 10⁻⁵	−0.183	0.68	1.25	0.49	Power law with cut-off
7	0	−288	0	−201	2.2 × 10⁻³⁹	−219	1.2 × 10⁻⁴²	Power law with cut-off
8	0.034	−23.4	8.2 × 10⁻¹²	−4.28	0.035	−4.81	0.039	Power law with cut-off
9	0.60	−6.48 × 10⁻³	0.91	−0.393	0.18	0.262	0.44	Power law
10	0.028	−18.6	1.0 × 10⁻⁹	−3.49	0.049	−3.96	0.048	Power law with cut-off
11	0	−65.9	0	−26.1	7.8 × 10⁻⁷	−29.5	3.8 × 10⁻⁷	Power law with cut-off
12	0	−45.7	0	−16.8	2.7 × 10⁻⁵	−19.5	1.4 × 10⁻⁵	Power law with cut-off
13	0.035	−15.6	2.4 × 10⁻⁸	−6.04	0.016	−6.73	0.014	Power law with cut-off

Corresponding fits and data in Figure 4. Negative log-likelihood ratio (LLR) values favor that specific column's distribution compared with a power-law distribution fit; e.g., for subject 1, an LLR of −56.2 favors a power law with exponential cut-off. For the power law column, p value is for the hypothesis that the power-law distribution is a plausible fit. For the other distribution columns, p value is for the hypothesis that the corresponding LLR is zero. For subject 4, both power law with exponential cut-off and stretched exponential are favored, with the hypothesis test unable to decide between these.