A chi-square goodness-of-fit test for non-identically distributed random variables: with application to empirical Bayes
- Texas Tech Univ., Lubbock, TX (United States)
- Rice Univ., Houston, TX (United States)
- Los Alamos National Lab., NM (United States)
When using parametric empirical Bayes estimation methods for estimating the binomial or Poisson parameter, the validity of the assumed beta or gamma conjugate prior distribution is an important diagnostic consideration. Chi-square goodness-of-fit tests of the beta or gamma prior hypothesis are developed for use when the binomial sample sizes or Poisson exposure times vary. Nine examples illustrate the application of the methods, using real data from such diverse applications as the loss of feedwater flow rates in nuclear power plants, the probability of failure to run on demand and the failure rates of the high pressure coolant injection systems at US commercial boiling water reactors, the probability of failure to run on demand of emergency diesel generators in US commercial nuclear power plants, the rate of failure of aircraft air conditioners, baseball batting averages, the probability of testing positive for toxoplasmosis, and the probability of tumors in rats. The tests are easily applied in practice by means of corresponding Mathematica{reg_sign} computer programs which are provided.
- Research Organization:
- Los Alamos National Lab. (LANL), Los Alamos, NM (United States)
- Sponsoring Organization:
- Nuclear Regulatory Commission, Washington, DC (United States)
- DOE Contract Number:
- W-7405-ENG-36
- OSTI ID:
- 645488
- Report Number(s):
- LA-UR-97-4825; ON: DE98004319; TRN: 98:010062
- Resource Relation:
- Other Information: PBD: Dec 1997
- Country of Publication:
- United States
- Language:
- English
Similar Records
Fitting of Failure Rate Data to a Closed Form Solution using Gamma-Poisson Method of Moments
Application of the Bayes equation to predicting reactor system reliability