The Annals of Statistics

A robust adjustment of the profile likelihood

James E. Stafford

Full-text: Open access


Under mild misspecifications of model assumptions, maximum likelihood estimates often remain consistent and asymptotically normal. Asymptotic normality will often hold for the signed root of the likelihood ratio statistic and the score statistic as well. However, standard estimates of asymptotic variance are usually inconsistent. This occurs when Bartlett's second identity fails. In the manner of McCullagh and Tibshirani, a variance correction may be used to adjust the profile likelihood so this identity obtains. The resulting likelihood yields the robust versions of the signed root, Wald and score statistic suggested by Kent and Royall.

Assuming model correctness, asymptotic expansions for the first three cumulants of each robust statistic are derived. It is seen that bias and skewness are not severely affected by using a robust statistic. An invariant expression derived for the asymptotic relative efficiency of a robust method allows assessment in numerous examples considered. Even for moderately large sample sizes, losses in efficiency are significant, making the misuse of a robust variance estimate potentially costly. Computer algebra is used in many of the calculations reported in this paper.

Article information

Ann. Statist., Volume 24, Number 1 (1996), 336-352.

First available in Project Euclid: 26 September 2002

Permanent link to this document

Digital Object Identifier

Mathematical Reviews number (MathSciNet)

Zentralblatt MATH identifier

Primary: 62F35: Robustness and adaptive procedures 62F05: Asymptotic properties of tests

Asymptotic relative efficiency Bartlett identities computer algebra cumulants profile likelihood robustness


Stafford, James E. A robust adjustment of the profile likelihood. Ann. Statist. 24 (1996), no. 1, 336--352. doi:10.1214/aos/1033066212.

Export citation


  • ANDREWS, D. F. and STAFFORD, J. E. 1993. Tools for the sy mbolic computation of asy mptotic expansions. J. Roy. Statist. Soc. Ser. B 55 613 627. Z. BARNDORFF-NIELSEN, O. E. 1983. On a formula for the distribution of the maximum likelihood estimator. Biometrika 70 343 365. Z. BARNDORFF-NIELSEN, O. E. and SORENSON, M. 1994. A review of some aspects of asy mptotic likelihood theory for stochastic processes. Internat. Statist. Rev. 62 133 165. Z.
  • COX, D. R. and REID, N. 1987. Parametric orthogonality and approximate conditional inference. J. Roy. Statist. Soc. Ser. B 49 1 39. Z.
  • COX, D. R. and REID, N. 1993. A note on the calculation of adjusted profile likelihood. J. Roy. Statist. Soc. Ser. B 55 467 471. Z.
  • DICICCIO, T. J. 1984. On parameter transformations and interval estimation. Biometrika 71 477 485.
  • DICICCIO, T. J. and STERN, S. E. 1994a. Frequentist and Bayesian Bartlett correction of test statistics based on adjusted profile likelihoods J. Roy. Statist. Soc. Ser. B 56 397 408. Z.
  • DICICCIO, T. J. and STERN, S. E. 1994b. Constructing approximately standard normal pivots from signed roots at adjusted likelihood ratio statistics. Scand. J. Statist. 21 447 460. Z.
  • EFRON, B. and HINKLEY, D. V. 1978. Assessing the accuracy of the maximum likelihood estimator: Observed versus expected Fisher information. Biometrika 65 457 481. Z.
  • GOULD, A. and LAWLESS, J. F. 1988. Consistency and efficiency of regression coefficient estimates in location-scale models. Biometrika 75 535 540. Z.
  • HUBER, P. J. 1967. The behaviour of maximum likelihood estimates under nonstandard conditions. Proc. Fifth Berkeley Sy mp. Math. Statist. Probab. 1 221 233. Univ. California Press, Berkeley. Z.
  • KENT, T. J. 1982. Robust properties of likelihood ratio tests. Biometrika 69 19 27. Z.
  • LAWLEY, D. N. 1956. A general method for approximating the distribution of likelihood ratio criteria. Biometrika 43 295 303. Z.
  • LIN, D. Y. and WEI, L. J. 1989. The robust inference for the Cox proportional hazards model. J. Amer. Statist. Assoc. 84 1074 1078. Z.
  • MCCULLAGH, P. and TIBSHIRANI, R. 1990. A simple method for the adjustment of profile likelihoods. J. Roy. Statist. Soc. Ser. B 52 325 344. Z.
  • ROy ALL, R. M. 1986. Model robust confidence intervals using maximum likelihood estimators. Internat. Statist. Rev. 54 221 226. Z.
  • STAFFORD, J. E. and ANDREWS, D. F. 1993. A sy mbolic algorithm for studying adjustments to the profile likelihood. Biometrika 80 715 730. Z.
  • STAFFORD, J. E., ANDREWS, D. F. and WANG, Y. 1994. Sy mbolic computation: A unified approach to studying likelihood. Statistics and Computing 4 235 245. Z.
  • WHITE, H. 1982. Maximum likelihood estimation of misspecified models. Econometrica 82 1 25. Z.
  • WOLFRAM, S. 1988. Mathematica: A Sy stem for Doing Mathematics by Computer. AddisonWesley, Reading, MA.