This content is not included in
your SAE MOBILUS subscription, or you are not logged in.
A Simple Method for Comparing Designs…Are Two Data Sets Significantly Different?
Annotation ability available
Sector:
Language:
English
Abstract
Research into data set comparison indicates that the likelihood ratio (LR) test with adjustment provides accurate confidence level results even for very small data sets. Inclusion of an unbiasing factor to modify the standard LR test for sample size appears to be the best available method for determining whether significant difference exists between data sets such as measurements from two separate designs. This modified LR test is tailored for Weibull data sets, but provides relatively unbiased results for normal and lognormal data sets as well. Two other methods, (1) viewing likelihood contour plots and (2) comparing double confidence bounds, also look promising. Further research is outlined.
Recommended Content
Technical Paper | Development of Nickel-Base Superalloy for Exhaust Valves |
Technical Paper | A Study of Technology for Assembling Vehicle Endurance Reliability |
Technical Paper | Vehicle Powertrain Loading Simulation and Variability |
Authors
Topic
Citation
Fulton, W. and Abernethy, R., "A Simple Method for Comparing Designs…Are Two Data Sets Significantly Different?," SAE Technical Paper 960544, 1996, https://doi.org/10.4271/960544.Also In
References
- Abernethy, Robert “The New Weibull Handbook” 1993
- Nelson, Wayne “Weibull Analysis of Reliability Data with Few or No Failures” Journal of Quality Technology 17 3 July 1985
- Nelson, Wayne “Accelerated Testing” John Wiley 1990
- Thoman, Darrel Bain, Lee “ Two Sample Tests in the Weibull Distribution ” Technometrics 11 4 November 1969
- McCool, John “ Analysis of Single Classification Experiments Based on Censored Samples from the Two-Parameter Weibull Distribution ” Journal of Statistical Planning and Inference 3 1979 39 68
- WeibullSMITH™, MonteCarloSMITH™ and Visual*SMITH™ Fulton Findings™