MSE performance of the weighted average estimators consisting of shrinkage estimators when each individual regression coefficient is estimated

In this paper, we analytically derive the exact formula for the mean squared error (MSE) of two weighted average (WA) estimators for each individual regression coefficient. Further, we execute numerical evaluations to investigate small sample properties of the WA estimators, and compare the MSE performance of the WA estimators with the other shrinkage estimators and the usual OLS estimator. Our numerical results show that (1) the WA estimators have smaller MSE than the other shrinkage estimators and the OLS estimator over a wide region of parameter space; (2) the range where the relative MSE of the WA estimator is smaller than that of the OLS estimator gets narrower as the number of explanatory variables k increases.     

Multiple comparisons in two-factor testing of psychophysical thresholds by the delta test method

We propose an algorithm of multiple comparisons in two-factor testing of psychophysical thresholds. We use logistic regression combined with guessing rate and adopt the step-down procedure with Ryan–Einot–Gabriel–Welsch (REGW) significance levels for multiple test. We test hypotheses of main effects and interaction in two-factor problem by using the delta test statistics.     

MSE performance of the 2SHI estimator in a regression model with multivariate t error terms

t error terms and derive the explicit formula of the mean squared error (MSE) of the two-stage hierarchial information (2SHI) estimator. It is shown by numerical evaluations that the 2SHI estimator has smaller MSE than the positive-part Stein-rule (PSR) estimator over a wide region of the parameter space. Received: November 6, 1998; revised version: October 15, 1999     

Inheritance of sex ratio in the Kanzawa spider mite,Tetranychus kanzawai Kishida

Summary The frequency distributions in sex ratios of offspring from 4 fieldTetranychus kanzawai populations were examined. No significant difference was observed between average sex ratios of the 4 populations, although there was a great variability within each population, especially in the population fromPueraia lobata. Using the population fromP. lobata, inbred lines with high (H) and low (L) sex ratios were selected. Crossing experiments between H and L showed that sex ratio is determined by the genotype of mothers. It seemed that sex ratio is cointrolled by several genes, with no cytoplasmic factor involved.     

Life cycle of a non-diapausing population of the two-spotted spider mite,Tetranychus urticae Koch in a pear orchard

Summary The life cycle of a non-diapausingT. urticae population was studied in a pear orchard in the southern part of Okayama Prefecture, western Honsh?, Japan. The mites overwintered only on biennial weeds in the orchard without entering diapause, started to increase in number in early spring on them and then moved to other weeds as they emerged. The occurrence of the mites on pear began in early May several weeks after the unfolding of pear leaves and the population on pear was initiated by the migrants from the weeds around the base of pear trees. The population on pear showed two distinct peak occurrences, one in June and the other in mid-autumn. After pear leaves dropped in late autumn, the mites returned to biennial weeds for overwintering. This work was supported in part by Grant-in-Aid No. 00556033 from the Ministry of Education, Science and Culture, Japan.     

Application of Human Reliability Analysis to Nursing Errors in Hospitals

Adverse events in hospitals, such as in surgery, anesthesia, radiology, intensive care, internal medicine, and pharmacy, are of worldwide concern and it is important, therefore, to learn from such incidents. There are currently no appropriate tools based on state-of-the art models available for the analysis of large bodies of medical incident reports. In this study, a new model was developed to facilitate medical error analysis in combination with quantitative risk assessment. This model enables detection of the organizational factors that underlie medical errors, and the expedition of decision making in terms of necessary action. Furthermore, it determines medical tasks as module practices and uses a unique coding system to describe incidents. This coding system has seven vectors for error classification: patient category, working shift, module practice, linkage chain (error type, direct threat, and indirect threat), medication, severity, and potential hazard. Such mathematical formulation permitted us to derive two parameters: error rates for module practices and weights for the aforementioned seven elements. The error rate of each module practice was calculated by dividing the annual number of incident reports of each module practice by the annual number of the corresponding module practice. The weight of a given element was calculated by the summation of incident report error rates for an element of interest. This model was applied specifically to nursing practices in six hospitals over a year; 5,339 incident reports with a total of 63,294,144 module practices conducted were analyzed. Quality assurance (QA) of our model was introduced by checking the records of quantities of practices and reproducibility of analysis of medical incident reports. For both items, QA guaranteed legitimacy of our model. Error rates for all module practices were approximately of the order 10(-4) in all hospitals. Three major organizational factors were found to underlie medical errors: "violation of rules" with a weight of 826 x 10(-4), "failure of labor management" with a weight of 661 x 10(-4), and "defects in the standardization of nursing practices" with a weight of 495 x 10(-4).     

Bootstrap test for a structural break under possible heteroscedasticity

In this article, we consider the Wald test statistic for testing equality between the sets of regression coefficients in two linear regression models when the disturbance variances may possibly be unequal. This test can be also used as a test for a structural break. However, it is well known that the test based on the Wald test statistic suffers from severe size distortion in small sample when the disturbance variances of the two regression models are unequal. Our simulation results show that substantial improvements are made when the bootstrap methods are applied.     

Comparison of powers and exact bahadur slopes of some tests for dimensionality in manova models

In this paper we consider test of dimensionality in MANOVA model. For this testing problem, the likelihood ratio (=LR) test, Lawley-Hotelling (=LH) type test and Bartlett-Nanda-Pillai (=BNP) type test are often used. We obtain the asymptotic expansions of powers of these tests under the local alternatives. Also Bahadur exact slopes of these tests are obtained. Based on these results, we obtain a unified opinion concerning comparison of LR test, LH type test and BNP type test.