Table 1 Accuracy of the χ2-based P-value in estimating the probability of type I error in the likelihood-ratio test
Sampling parametersα
NTLAbsorption probabilityŇ/N0.050.010.0010.0001
104101012.6 × 10−31.31.41.72.33.3
10410010107.9 × 10−41.31.41.72.32.9
1041000101001.2 × 10−31.31.62.44.38.1
10310010102.2 × 10−31.31.72.54.58.3
1041000101001.2 × 10−31.31.62.44.38.1
105100001010001.3 × 10−21.31.21.52.33.6
1041005207.7 × 10−41.72.02.95.28.5
10410010107.9 × 10−41.31.41.72.32.9
10410010018.0 × 10−41.01.11.11.21.4
  • Columns 1–4 show simulation and sampling parameters (see text for notations). Column 5 shows the probability that the allele fixes or goes extinct within the sampling period. Column 6 shows the ratio of the population size most likely under the neutral null hypothesis Ň to the true population size N. Columns 7–10 show the ratio of the true fraction of false positives in the likelihood-ratio test to the fraction α expected under the assumption that the LRS is distributed as χ2 with 1 d.f., across a range of α-values. We performed 106 neutral Wright–Fisher simulations with the initial allele frequency ν0 = 0.5. See Table S1 for results for other initial frequencies. Simulations with absorption events were excluded from the analysis.