Skip to main content

Table 1 Markov chain statistics.

From: Composite likelihood estimation of demographic parameters

Dataset

Statistic

t 1

t 2

h

α 1

α 2

α 3

AA

ρ 50

0.1520

0.1478

0.0353

0.0600

0.0530

0.0099

 

PSRF

1

1

1

1

1

1

 

Upper PSRF

1

1

1

1

1

1

AE

ρ 50

0.1587

0.1382

0.0302

0.0565

0.0349

0.0044

 

PSRF

1

1

1

1

1

1

 

Upper PSRF

1

1

1

1

1

1

AO

ρ 50

0.1617

0.1324

0.0133

0.0497

0.0326

0.0081

 

PSRF

1

1

1

1

1

1

 

Upper PSRF

1

1

1

1

1

1

A

ρ 50

0.1197

0.2637

0.0868

0.0269

0.0538

0.0669

 

PSRF

1

1.02

1

1.02

1

1

 

Upper PSRF

1.01

1.03

1

1.02

1

1

D

ρ 50

0.1217

0.1916

0.0728

0.0232

0.0604

0.0570

 

PSRF

1

1.01

1

1.01

1

1

 

Upper PSRF

1

1.02

1

1.01

1

1

G

ρ 50

0.1066

0.1567

0.1097

0.0938

0.0053

0.0805

 

PSRF

1.13

1.19

1.11

1

1.15

1.27

 

Upper PSRF

1.27

1.39

1.22

1

1.30

1.53

  1. The behavior of the Markov chains for all datasets, as determined by ten independent runs of the chain for each dataset. Three statistics are given: ρ50 is the autocorrelation of each parameter at lag 50 steps, PSRF refers to the potential scale reduction factor for each parameter, and Upper PRSF is the 97.5% quantile of the PRSF. Multivariate PRSF values are not given. The resequencing data sets of Wall et al. [14] are abbreviated as Africa-Asia (AA), Africa-Europe (AE), and Africa-Oceania (AO). Additionally, autocorrelation values are also presented for three representative simulated data sets (A, D, and G).