q_{c} | Sample size N | q̂_{+}^{a} | q̂_{−}^{a} | Type I
error rate^{b} |
---|---|---|---|---|

0.2 | 1000 | 0.20025 ± 0.02085 | 0.20000 ± 0.01625 | 0.05060 |

0.1 | 1000 | 0.09996 ± 0.01570 | 0.10008 ± 0.01241 | 0.05580 |

0.2 | 200 | 0.19929 ± 0.04719 | 0.19973 ± 0.03654 | 0.05040 |

0.2 | 100 | 0.19987 ± 0.06644 | 0.20118 ± 0.05238 | 0.05640 |

0.4 | 100 | 0.39986 ± 0.08245 | 0.40040 ± 0.06400 | 0.05980 |

0.5 | 100 | 0.50117 ± 0.08471 | 0.49938 ± 0.06418 | 0.05280 |

↵

Mean ± SD of the estimates for two penetrances obtained by analysis under the alternative hypothesis.^{a}↵

The proportion of attempts that yielded values of the statistic over 3.841 (the value that yields the cumulative density function of 0.95 for the χ^{b}^{2}distribution with 1 d.f.).Each simulation was performed with a given penetrance (

*q*_{c}) and a sample size (*N*) under the null hypothesis in the dominant mode as stated in simulation (under methods). After removing the phase information, PENHAPLO was used to estimate*q̂*_{+}and*q̂*_{−}under the alternative hypothesis. The same software also calculates*L*_{max}by analysis under the alternative hypothesis and*L*_{0max}by analysis under the null hypothesis. The statistic −2 log(*L*_{0max}/*L*_{max}) was then calculated. This simulation was repeated 10,000 times for each parameter set.