Summary
The problem considered is sequential estimation of the mean θ of a one-parameter exponential family of distributions with squared error loss for estimation error and a cost c>0 for each of an i.i.d. sequence of potential observations X 1, X 2,...A Bayesian approach is adopted, and natural conjugate prior distributions are assumed. For this problem, the asymptotically pointwise optimal (A.P.O.) procedure continues sampling until the posterior variance of θ is less than c(r0+n), where n is the sample size and r 0 is the fictitous sample size implicit in the conjugate prior distribution. It is known that the A.P.O. procedure is Bayes risk efficient, under mild integrability conditions. In fact, the Bayes risk of both the optimal and A.P.O. procedures are asymptotic to 2V 0 √c, as c→0, where V 0 is the prior expectation of the standard deviation of X 1 given θ. Here the A.P.O. rule is shown to be asymptotically non-deficient, under stronger regularity conditions: that is, the difference between the Bayes risk of the A.P.O. rule and the Bayes risk of the optimal procedure is of smaller order of magnitude than c, the cost of a single observation, as c→0. The result is illustrated in the exponential and Bernoulli cases, and extended to the case of a normal distribution with both the mean and variance unknown.
Article PDF
Similar content being viewed by others
References
Alvo, M.: Bayesian sequential estimates. Ann. Statist. 5, 955–968 (1977)
Alvo, M., Cabilio, P.: On the sequential estimation of the probability of success. Austral. J. Statist., 21, 66–70 (1979)
Barndorff-Nielsen, O.: Information and Exponential Families in Statistical Theory. New York: Wiley 1978
Bickel, O., Yahav, J.: Asymptotically pointwise optimal procedures in sequential analysis. Proc. Fifth Berkeley Sympos. Math. Statist. Prob., 1, 401–413, University of California 1967
Bickel, P., Yahav, J.: Asymptotically optimal Bayes and minimax procedures in sequential estimation. Ann. Math. Statist., 39, 442–456 (1968)
Bickel, P., Yahav, J.: An A.P.O. rule in sequential estimation with quadratic loss. Ann. Math. Statist, 40, 417–426 (1969a)
Bickel, P., Yahav, J.: Some contributions to the asymptotic theory of Bayes solutions. Z. Wahrscheinlichkeitstheorie und verw. Gebiete, 11, 257–276 (1969b)
Chow, Y.S., Robbins, H., Siegmund, D.: Great Expectations: the Theory of Optimal Stopping. Houghton Mifflin Co. 1970
Diaconis, P., Ylvisaker, D.: Conjugate priors for exponential families. Ann. Statist., 6, 269–281 (1979)
Ferguson, T.: Mathematical Statistics: a Decision Theoretic Approach (1967)
Johnson, R.: An asymptotic expansion for posterior distributions. Ann. Math. Statist. 41, 1899–1907 (1970)
Lehmann, E.: Testing Statistical Hypotheses. New York: Wiley 1959
Rasmussen, S.: Bayesian sequential estimation. Ann. Statist. 8, 1229–1234 (1980)
Shapiro, C., Wardrop, R.: Bayesian sequential estimation for one parameter exponential families. J. Amer. Statist. Ass. 75, 984–988 (1980a)
Shapiro, C., Wardrop, R.: Dynkin's identity applied to Bayes' sequential estimation of a Poisson process. Ann. Statist. 8, 171–180 (1980)
Vardi, Y.: Asymptotic optimality of certain sequential estimators. Ann. Statist. 7, 1034–1039 (1979)
Author information
Authors and Affiliations
Additional information
Research supported by the National Science Foundation Grant NSF-MCS-78-02468
Rights and permissions
About this article
Cite this article
Woodroofe, M. A.P.O. rules are asymptotically non deficient for estimation with squared error loss. Z. Wahrscheinlichkeitstheorie verw Gebiete 58, 331–341 (1981). https://doi.org/10.1007/BF00542639
Received:
Issue Date:
DOI: https://doi.org/10.1007/BF00542639