On estimating the covariance matrix of robust regression M-estimates

By analyzing a special class of regression problems we point out that previously suggested estimates of the covariance matrix of regression M-cstimatos are inadequate for certain design matrices. These results confirm the conclusions drawn in several Monte Cailo studies.     

A comparison of confidence intervals generated by the scheffé and bonferroni methods

Simultaneous confidence intervals for the p means of a multivariate normal random variable with known variances may be generated by the projection method of Scheffé and by the use of Bonferroni's inequality. It has been conjectured that the Bonferroni intervals are shorter than the Scheffé intervals, at least for the usual confidence levels. This conjecture is proved for all p≥2 and all confidence levels above 50%. It is shown, incidentally, that for all p≥2 Scheffé's intervals are shorter for sufficiently small confidence levels. The results are also applicable to the Bonferroni and Scheffé intervals generated for multinomial proportions.     

Sample-size requirements for confidence bounds of prescribed precision on multivariate normal quantiles

The problem of setting confidence bounds on a central multivariate normal quantile is considered. It is shown that for the setting of exact confidence bounds of specified closeness to the quantile,the required minimum size of a normal sample is large and rises rapidly with the number of variates considered.     

An upper bound on a ratio of variances

Several processes may be monitored in terms of their variances relative to each other by the maximum ratio of variances. Based on the observed value an upper confidence bound for the population value is derived and tables are supplied for its implementation. Such bounds may be used when the experimenter wishes to establish that the variances are "not too different". The usual testing of variances for equality is noticed to be inappropriate in such cases.     

Some multiple comparison procedures for comparing treatments within several groups

Multiple comparison procedures are extended to designs consisting of several groups, where the treatment means are to be compared within each group. This may arise in two-factor experiments, with a significant interaction term, when one is interested in comparing the levels of one factor at each level of the other factor. A general approach is presented for deriving the distributions and calculating critical points, following three papers which dealt with two specific procedures. These points are used for constructing simultaneous confidence intervals over some restricted set of contrasts among treatment means in each of the groups. Tables of critical values are provided for two procedures and an application is demonstrated. Some extensions are presented for the case of possible different sets of contrasts and also for unequal variances in the various groups.     

Estimating the discovery rate in a continuous time recapture model

We consider a family of marked Poisson process models for the discovery of distinct errors in a computer program and also for sampling, in continu-ous time, a population containing an unknown number of distinct biological species. Captures (selections or discoveries) are assumed to occur at a con-stant rate, each event consisting of the discovery of a distinct process (error or species) or the recurrence of a previously discovered process. Using a generalization of Nayak’s (1988) model we derive confidence limits for the discovery rate. The limits are based on the asymptotic distribution of a scaled logarithmic function of the maximum likelihood estimator.     

A one-sided test for testino homogeneity of scale parameters against simple ordered alternative

In an earlier paper the authors (1997) extended the results of Hayter (1990) to the two parameter exponential probability model. This paper addressee the extention to the scale parameter case under location-scale probability model. Consider k (k≧3) treatments or competing firms such that an observation from with treatment or firm follows a distribution with cumulative distribution function (cdf) F_i(x)=F[(x-μ_i)/Q_i], where F(·) is any absolutely continuous cdf, i=1,…,k. We propose a test to test the null hypothesis H₀:θ₁=…=θ_k against the simple ordered alternative H₁:θ₁≦…≦θ_k, with at least one strict inequality, using the data X_i,j, i=1,…k; j=1,…,n₁. Two methods to compute the critical points of the proposed test have been demonstrated by talking k two parameter exponential distributions. The test procedure also allows us to construct simultaneous one sided confidence intervals (SOCIs) for the ordered pairwise ratios θ_j/θ_i, 1≦i<j≦k. Statistical simulation revealed that: 9i) actual sizes of the critical points are almost conservative and (ii) power of the proposed test relative to some existing tests is higher.