TABLE 7

Probabilities for data from Hudson et al. (1994) (×10−5): c = 5e1 + e5, s = 55, and n = 10

β = 0β = 0.1β = 1β = 10
ρ = 0C16991442 469 3
M 755 635 222 2
K54814702176450
H 781 647 224 2
ρ = 0.01C17221394 460 6
M 752 592 230 3
K56274629176246
H 778 612 234 3
ρ = 0.1C17761381 427 4
M 766 582 217 2
K56724552167252
H 792 602 222 2
ρ = 1C16171289 368 9
M 726 565 182 4
K52584332155436
H 751 581 182 4
ρ = 10C 674 540 193 4
M 316 268  97 2
K24632044 79529
H 320 271  98 2
ρ = 100C   8   1   0 0
M   3   1   0 0
K  24  12   9 1
H   3   1   0 0
  • Values are based on 105 accepted genealogies using a point estimate of θ, estimated from (26) for β = 0 and using 2s/Embedded Image for β > 0, with Embedded Image based on 105 separate simulated genealogies. If the observed configuration is c, the values in the table are (×105): ℙ(Cc), ℙ[MM(c)], ℙ[KK(c)], and ℙ[HH(c)], where probabilities are estimated from the 105 genealogies. If the observed value of C, M, K, or H was located on the boundary of the rejection region, the null hypothesis was not rejected. To reject the null hypothesis at the 0.05 level, numbers shown must be at most 5000 for the test based on C and 2500 for the other (two-tailed) tests.