Home > Research > Publications & Outputs > Possible biases induced by MCMC convergence.

Research

Links

Text available via DOI:

Keywords

View graph of relations

Possible biases induced by MCMC convergence.

Research output: Contribution to Journal/Magazine › Journal article › peer-review

Published

Standard

Possible biases induced by MCMC convergence. / Cowles, K.; Roberts, G. O.; Rosenthal, S.
In: Journal of Statistical Computation and Simulation, Vol. 64, No. 1, 1999, p. 87-104.

Research output: Contribution to Journal/Magazine › Journal article › peer-review

Harvard

Cowles, K, Roberts, GO & Rosenthal, S 1999, 'Possible biases induced by MCMC convergence.', Journal of Statistical Computation and Simulation, vol. 64, no. 1, pp. 87-104. https://doi.org/10.1080/00949659908811968

APA

Cowles, K., Roberts, G. O., & Rosenthal, S. (1999). Possible biases induced by MCMC convergence. Journal of Statistical Computation and Simulation, 64(1), 87-104. https://doi.org/10.1080/00949659908811968

Vancouver

Cowles K, Roberts GO, Rosenthal S. Possible biases induced by MCMC convergence. Journal of Statistical Computation and Simulation. 1999;64(1):87-104. doi: 10.1080/00949659908811968

Author

Cowles, K. ; Roberts, G. O. ; Rosenthal, S. / Possible biases induced by MCMC convergence. In: Journal of Statistical Computation and Simulation. 1999 ; Vol. 64, No. 1. pp. 87-104.

Bibtex

@article{afc2eb9e27ad418383ec38654e9bf53d,
title = "Possible biases induced by MCMC convergence.",
abstract = "Convergence diagnostics are widely used to determine how many initial “burn-in” iterations should be discarded from the output of a Markov chain Monte Carlo (MCMC) sampler in the hope that the remaining samples are representative of the target distribution of interest. This paper demonstrates that some ways of applying convergence diagnostics may actually introduce bias into estimation based on the sampler output. To avoid this possibility, we recommend choosing the number of burn-in iterations r by applying convergence diagnostics to one or more pilot chains, and then basing estimation and inference on a separate long chain from which the first r iterations have been discarded.",
keywords = "Markov chain Monte Carlo, convergence diagnostic, estimation, bias, batch means",
author = "K. Cowles and Roberts, {G. O.} and S. Rosenthal",
year = "1999",
doi = "10.1080/00949659908811968",
language = "English",
volume = "64",
pages = "87--104",
journal = "Journal of Statistical Computation and Simulation",
issn = "1563-5163",
publisher = "Taylor and Francis Ltd.",
number = "1",

}

RIS

TY - JOUR

T1 - Possible biases induced by MCMC convergence.

AU - Cowles, K.

AU - Roberts, G. O.

AU - Rosenthal, S.

PY - 1999

Y1 - 1999

N2 - Convergence diagnostics are widely used to determine how many initial “burn-in” iterations should be discarded from the output of a Markov chain Monte Carlo (MCMC) sampler in the hope that the remaining samples are representative of the target distribution of interest. This paper demonstrates that some ways of applying convergence diagnostics may actually introduce bias into estimation based on the sampler output. To avoid this possibility, we recommend choosing the number of burn-in iterations r by applying convergence diagnostics to one or more pilot chains, and then basing estimation and inference on a separate long chain from which the first r iterations have been discarded.

AB - Convergence diagnostics are widely used to determine how many initial “burn-in” iterations should be discarded from the output of a Markov chain Monte Carlo (MCMC) sampler in the hope that the remaining samples are representative of the target distribution of interest. This paper demonstrates that some ways of applying convergence diagnostics may actually introduce bias into estimation based on the sampler output. To avoid this possibility, we recommend choosing the number of burn-in iterations r by applying convergence diagnostics to one or more pilot chains, and then basing estimation and inference on a separate long chain from which the first r iterations have been discarded.

KW - Markov chain Monte Carlo

KW - convergence diagnostic

KW - estimation

KW - bias

KW - batch means

U2 - 10.1080/00949659908811968

DO - 10.1080/00949659908811968

M3 - Journal article

VL - 64

SP - 87

EP - 104

JO - Journal of Statistical Computation and Simulation

JF - Journal of Statistical Computation and Simulation

SN - 1563-5163

IS - 1

ER -