Computer intensive methods for inference on parameters of complex models - A Bayesian alternative

Ayala Cohen, Udi Makov, Ruth Boneh

Research output: Contribution to journalArticlepeer-review


Simulation techniques are gaining popularity for drawing statistical inference, particularly in the field of practical Bayesian Statistics (e.g. Ann. Statist. 9 (1981), 130-134; J. Amer. Statist. Assoc. 85 (1987), 470-477; J. Amer. Statist. Assoc. 85 (1990), 398-409; J. Roy. Statist. Soc. 56 (1994), 3-48). Smith and Gelfand (Amer. Statist. 46 (1992), 84-88) discussed two sampling-resampling methods for obtaining posterior samples from prior samples. The first is based on the rejection method, the second (the weighted bootstrap) is the sampling importance resampling (SIR) (J. Amer. Statist. Assoc. 82 (1987), 543-546; Bayesian Statistics, Vol. 3, Oxford University Press, 1988). This paper uses an example to illustrate the applicability of these methods. The example is on cosmological data for which maximum likelihood estimates do not exist. The data were analyzed by Chernoff (Comput. Statist. Data Anal. 12 (1991), 159-178) who introduced several non-Bayesian computer intensive methods for analyzing these data. The cosmological data are used to evaluate the two Bayesian computer intensive methods.

Original languageEnglish
Pages (from-to)79-94
Number of pages16
JournalJournal of Statistical Planning and Inference
Issue number1-2
StatePublished - 1 Oct 1995


  • Bayes estimation
  • Bootstrap
  • SIR
  • Sampling-resampling
  • Simulations

ASJC Scopus subject areas

  • Statistics and Probability
  • Statistics, Probability and Uncertainty
  • Applied Mathematics


Dive into the research topics of 'Computer intensive methods for inference on parameters of complex models - A Bayesian alternative'. Together they form a unique fingerprint.

Cite this