Abstract
The mixed Weibull distribution provides a flexible model to analyze random durations in a possibly heterogeneous population. To test for homogeneity against unobserved heterogeneity in a Weibull mixture model, a dispersion score test and a goodness-of-fit test are investigated. The empirical power of these tests is assessed and compared on a broad range of alternatives. It comes out that the dispersion score test, as it is based on a Weibull-to-exponential transformation, often breaks down. A simple new procedure is introduced for Weibull mixtures in scale, which combines the dispersion score test and the goodness-of-fit test. The new test is compared with several known procedures and shown to have a good overall power. To detect mixtures in shape and scale, a goodness-of-fit test is recommended.
Similar content being viewed by others
References
Abernethy RB (1996) The new Weibull handbook, 2nd edn. Abernethy R.B., North Palm Beach
Albert JRG, Baxter LA (1995) Applications of the EM algorithm to the analysis of life length data. Appl Stat 44:323–341
Andersen PK, Borgan O, Gill RD, Keiding N (1993) Statistical models based on counting processes. Springer, Berlin Heidelberg New York
Charnigo R, Sun J (2004) Testing homogeneity in a mixture distribution via the L 2 distance between competing models. J Am Stat Soc 99:488–498
Chen H, Chen J, Kalbfleisch JD (2001) A modified likelihood ratio test for homogeneity in finite mixture models. J R Stat Soc B 63:19–29
Cheng SW, Fu JC (1982) Estimation of mixed Weibull parameters in life testing. IEEE Trans Reliab R-31(4):377–381
Commenges D, Andersen PK (1995) Score test of homogeneity for survival data. Lifetime Data Anal 1:145–160
D’Agostino RB, Stephens MA (1986) Goodness-of-fit techniques. Dekker, New York
Falls LW (1970) Estimation of parameters in compound Weibull distributions. Technometrics 12(2):399–407
Jaggia S (1997) Alternative forms of the score test for heterogeneity in a censored exponential model. Rev Econ Stat 79:340–343
Jiang R, Murthy DNP (1995) Modeling failure-data by mixture of 2 Weibull distributions: a graphical approach. IEEE Trans Reliab 44(3):477–488
Kao JHK (1959) A graphical estimation of mixed Weibull parameters in life-testing of electron tubes. Technometrics 1:389–407
Kaylan AR, Harris CM (1981) Efficient algorithms to derive maximum-likelihood estimates for finite exponential and Weibull mixtures. Comput Oper Res 8:97–104
Keats JB, Nahar PC, Korbel KM (2000) A study of the effects of mis-specification of the Weibull shape parameter on confidence bounds based on the Weibull-to-exponential transformation. Qual Reliab Eng Int 16:27–31
Lancaster T, (1990) The econometric analysis of transition data. Cambridge University Press, Cambridge
Lindsay BG (1995) Mixture models: theory, geometry and applications. Institute of Mathematical Statistics, Hayward
Marín JM, Rodríguez Bernal MR, Wiper MP (2003) Using Weibull mixture distributions to model heterogeneous survival data. Universidad Carlos III de Madrid, Working Paper 03-32, Statistics and econometrics series 08
Mosler K, Seidel W (2001) Testing for homogeneity in an exponential mixture model. Aust NZ J Stat 43:231–247
Neyman J, Scott EL (1966) On the use of C(α) optimal tests of composite hypotheses. Bull Inst Int Statist 41(I):477–497
Seidel W, Mosler K, Alker M (2000a) Likelihood ratio tests based on subglobal optimization: a power comparison in exponential mixture models. Stat Pap 41:85–98
Seidel W, Mosler K, Alker M (2000b) A cautionary note on likelihood ratio tests in mixture models. Ann Inst Stat Math 52:481–487
Shaked M (1980) On mixtures from exponential families. J R Stat Soc Ser B 42:192–198
Shorack GR, Wellner JA (1986) Empirical processes. Wiley, New York
Sinha SK (1986) Reliability and life testing. Wiley, New York
Tarum CD (1999) Classification and analysis of Weibull mixtures. SAE Technical Paper Series 1999-01-0055
Titterington DM, Smith AFM, Makov UE (1985) Statistical analysis of finite mixture distributions. Wiley, New York
Venables WN, Ripley BD (2002) Modern applied statistics with S-pLUS, 4th edn. Springer, Berlin Heidelberg New York
Yang Z, Xie M (2003) Efficient estimation of the Weibull shape parameter based on a modified profile likelihood. J Stat Comput Simulat 73(2):115–123
Author information
Authors and Affiliations
Corresponding author
Additional information
This research has been partially sponsored by a grant of the Deutsche Forschungsgemeinschaft. We thank Lars Haferkamp for computational assistance and Wilfried Seidel and a referee for their remarks on alternative test procedures.
Rights and permissions
About this article
Cite this article
Mosler, K., Scheicher, C. Homogeneity testing in a Weibull mixture model. Statistical Papers 49, 315–332 (2008). https://doi.org/10.1007/s00362-006-0015-6
Received:
Revised:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00362-006-0015-6
Keywords
- Mixture diagnosis
- Survival analysis
- Hazard models
- Dispersion score test
- Goodness-of-fit
- Weibull-to-exponential transform