摘要:

A composite hypothesis asserting some kind of equality about two distributions may be tested by use of a two-sample statistic. It is often true that if one sample size is allowed to become infinite in this statistic then it is interpretable as one for testing a simple hypothesis about the distribution furnishing the finite sample. (The simple hypothesis is furnished by the “infinite sample.”) Wilcoxon's two-sample test studied in this way becomes a useful tool for certain kinds of problems. A rank test of Lehmann's reduces to either Fisher's test for combining independent tests of significance or to one once proposed by Karl Pearson for the same purpose. Two natural one-sample limits of two-sample median tests are also presented.

ISSN:0162-1459    

DOI:10.1080/01621459.1964.10480719

出版商:Taylor & Francis Group    

年代:1964

数据来源: Taylor

摘要:

A simplification using symmetry is applied to the null distribution of the Fisher-Yates test orc1test of Hoeffding and Terry. A computer is used to extend existing tables to combined sample sizeN≤ 20. Thetapproximation of Hoeffding and Terry and the normal approximation are investigated with the former found to be satisfactory at the limit of the tables when the ratio of sample sizes is not too extreme. Selected power calculations forN≤ 10 are given for normal shift alternatives.

ISSN:0162-1459    

DOI:10.1080/01621459.1964.10480720

出版商:Taylor & Francis Group    

年代:1964

数据来源: Taylor

摘要:

The level of ordinary two-sample procedures is not preserved if the two populations differ in dispersion or shape. The effect of such differences, especially differences in dispersion, on thet, median, Mann-Whitney, and normal scores procedures is investigated asymptotically, and tables are given comparing the four procedures.

ISSN:0162-1459    

DOI:10.1080/01621459.1964.10480721

出版商:Taylor & Francis Group    

年代:1964

数据来源: Taylor

摘要:

A statistical theory based on the general linear statistical model is developed for seasonal adjustment of economic (and other) time series. A method for seasonal adjustment may be represented as taking place in two steps. The first step is to estimate the unknown parameters of the seasonal component of the series; the second step is to remove the estimated seasonal component from the set of observations. For the unique minimum variance, linear, unbiased method for seasonal adjustment, estimation is carried out through the unique, minimum variance, linear unbiased estimator. Sampling theory for statistical inference about a method for seasonal adjustment may be derived from normal sampling theory for the general linear statistical model. The properties of minimum variance, linearity, and unbiasedness provide a complete basis for the selection of a method for seasonal adjustment.

ISSN:0162-1459    

DOI:10.1080/01621459.1964.10480722

出版商:Taylor & Francis Group    

年代:1964

数据来源: Taylor

摘要:

A technique is presented for taking a covariable into account when the treatment and response variables are qualitative (i.e., where the familiar technique for the 2×2 contingency table would be used if no adjustment was made for the covariable). The technique is illustrated and exemplified by a clinical trial of chemotherapies for hyaline membrane disease in babies. The scientific rationale for an “exact” statistical test and a simple approximate test (COVAST) is provided. In terms of the mechanics, COVAST is an easy extension of the familiar Sign Test.

ISSN:0162-1459    

DOI:10.1080/01621459.1964.10480723

出版商:Taylor & Francis Group    

年代:1964

数据来源: Taylor

摘要:

To identify the determinants of saving, cross-sectional studies of households usually collect and compare the current values of variables. A frequent difficulty is correlation of the known variables with unknown characteristics of households. Stratification of the sample may not always eliminate such correlation, making for biased estimates.

ISSN:0162-1459    

DOI:10.1080/01621459.1964.10480724

出版商:Taylor & Francis Group    

年代:1964

数据来源: Taylor

摘要:

In this paper we have attempted to test the income-wealth hypothesis originally suggested by Marshall and later enunciated by Hansen. The hypothesis is that both income and wealth are the determinants of the demand for money The data are drawn from the cross-section sample of the 1957–58 re-interview Surveys of Consumer Finances. The numerical elasticities of income and wealth are also derived for both liquid assets as a whole and its various components in order to make a comparison with those of prior studies. Furthermore, the findings in the present analysis have some implications as to the adequacy of the Gurley-Shaw hypothesis that a substantial volume of close substitutes for money created by non-bank financial intermediaries has tended to reduce the demand for money. The principal findings from the particular sample are that (1) not only income but also wealth has a significant positive effect on the demand for money, thereby rejecting the alternative hypotheses that either income or wealth is the unique constraint on money balances, (2) income elasticities are substantially higher than those reported in other studies, whereas wealth elasticities are almost identical, and (3) some aspects of the Gurley-Shaw hypothesis do not appear to be substantiated by this study.

ISSN:0162-1459    

DOI:10.1080/01621459.1964.10480725

出版商:Taylor & Francis Group    

年代:1964

数据来源: Taylor

摘要:

In this paper, regression models with error terms generated by a first order autoregressive scheme are analyzed from a Bayesian point of view. Methods are developed for computing posterior distributions of regression coefficients and the parameter of the autoregressive process. The relationship of our approach to sampling theory approaches is briefly discussed.

ISSN:0162-1459    

DOI:10.1080/01621459.1964.10480726

出版商:Taylor & Francis Group    

年代:1964

数据来源: Taylor

摘要:

In recent years there has been much discussion of the relative merits of several methods of fitting the normal integral dosage response curve. For a three-dose experiment with ten observations at each dose and for a five-dose experiment with four observations, three methods are compared with respect to their root mean square error in estimatingm, σ, and β. Essentially the total population of samples is used. The main result is that the maximum likelihood method is superior to the minimum normit method in estimatingmand σ, but for β the minimum normit method is superior.

ISSN:0162-1459    

DOI:10.1080/01621459.1964.10480727

出版商:Taylor & Francis Group    

年代:1964

数据来源: Taylor

摘要:

Two examples illustrate the use of bivariate extremal distributions of the first type. One is the distribution of oldest ages at death for the two sexes; the other consists of the floods of the same river recorded at two stations located upstream and downstream. The first is an illustration of independence; the second, of dependence. These illustrations were chosen so as to facilitate the decision on the hypothesis of independence. We limit our consideration to the first, double exponential, extremal distribution because it can be written in a form which is parameter-free. Distribution-free tests for the hypothesis of independence and a criterion for independence of bivariate extremal distributions are given. Bivariate density curves of independent largest values are shown.

ISSN:0162-1459    

DOI:10.1080/01621459.1964.10480728

出版商:Taylor & Francis Group    

年代:1964

数据来源: Taylor