日本統計学会誌

INVERSE SAMPLING WITH APPLICATION TO ESTIMATION OF THE RATIO OF TWO MEAN LIFE TIMES

Shunichi Kigawa

1979 年 9 巻 1 号 p. 1-6
発行日: 1979年
公開日: 2009/01/22

DOIhttps://doi.org/10.11329/jjss1970.9.1

ジャーナルフリー

PDF形式でダウンロード (171K)
THE ASYMPTOTIC CUT-OFF POINT AND COMPARISON OF ERROR PROBABILITIES IN COVARIATE DISCRIMINANT ANALYSIS

Mitsuyo Kanazawa

1979 年 9 巻 1 号 p. 7-17
発行日: 1979年
公開日: 2009/01/22

DOIhttps://doi.org/10.11329/jjss1970.9.7

ジャーナルフリー

抄録を表示する抄録を非表示にする

This paper deals with the problem of classifying a multivariate observation into one of two normal populations with covariates having the same mean in both populations and also with a common unknown covariance matrix. We compare three procedures proposed for this problem, the linear classification statistic W^∗ due to Cochran and Bliss and the ML and the modified ML classification statistics Z^∗ and R^∗ both given by Fujikoshi and Kanazawa, by means of the probabilities of a type of misclassification. We set the cut-off point to achieve a specified probability of misclassification by using the asymptotic expansion of the distribution of the each classification statistic.

抄録全体を表示

PDF形式でダウンロード (294K)
OPTIMAL BALANCED FRACTIONAL 2^m FACTORIAL DESIGNS OF RESOLUTION IV DERIVED FROM BALANCED ARRAYS OF STRENGTH FOUR

Teruhiro Shirakura

1979 年 9 巻 1 号 p. 19-27
発行日: 1979年
公開日: 2009/01/22

DOIhttps://doi.org/10.11329/jjss1970.9.19

ジャーナルフリー

抄録を表示する抄録を非表示にする

In earlier works of Shirakura, it has been shown that some balanced fractional 2^m factorial designs of resolution IV, in which the general mean and main effects are estimable ignoring three-factor and higher order interactions, can be obtained from balanced arrays of strength four. Furthermore, in order to compare these designs, two criteria have been introduced. In this paper, all optimal designs with respect to these two criteria will be presented for each m with 4_??_m_??_10, and for any given assemblies N within a certain practical range.

抄録全体を表示

PDF形式でダウンロード (384K)
ESTIMATION OF THE DIFFERENCE BETWEEN TWO PROPORTIONS IN A FINITE MULTINOMIAL POPULATION

Yoshiaki Funatsu

1979 年 9 巻 1 号 p. 29-35
発行日: 1979年
公開日: 2009/01/22

DOIhttps://doi.org/10.11329/jjss1970.9.29

ジャーナルフリー

抄録を表示する抄録を非表示にする

In a finite multinomial population, the method of interval estimation for specified single proportion is already well known. It can be applied to the estimation for sum of specified two or more proportions. However, the joint estimation for specified two proportions or estimation for the difference between the two are more complicated than that for single proportion.
This paper provides the confidence regions for specified two proportions and the difference between the two with a degree of confidence coefficient. The confidence coefficient is given below by the minimum value instead of proper one that can be found by further study.

抄録全体を表示

PDF形式でダウンロード (832K)
NOTE ON DESIGNS IN SEROLOGY

Sanpei Kageyama

1979 年 9 巻 1 号 p. 37-40
発行日: 1979年
公開日: 2009/01/22

DOIhttps://doi.org/10.11329/jjss1970.9.37

ジャーナルフリー

抄録を表示する抄録を非表示にする

Some new series of neighbor designs, which are useful in serology, are presented in this paper. These are based on a principle of Rees [5].

抄録全体を表示

PDF形式でダウンロード (189K)

J-STAGEへの登録はこちら（無料）