摘要:

* Presidential address, at the 116th Annual Meeting of the American Statistical Association, Detroit, Michigan, September 9, 1956.

ISSN:0162-1459    

DOI:10.1080/01621459.1957.10501361

出版商:Taylor & Francis Group    

年代:1957

数据来源: Taylor

摘要:

A simple technique is presented for obtaining without calculation the best fitting line according to the least squares criterion, provided the points are equally spaced horizontally. A graphic measure of residual variability is also derived.

ISSN:0162-1459    

DOI:10.1080/01621459.1957.10501362

出版商:Taylor & Francis Group    

年代:1957

数据来源: Taylor

摘要:

In many practical applications, especially when a long question schedule is given to two (or more) series of individuals, the computation of a large number of chi-square tests may become burdensome. For certain common types of data the computation can be avoided, or greatly reduced, by the charts presented in this paper. If so desired these charts may be used as a screen for borderline 2 × 2 tables and the occasional borderline case calculated by the usual methods.

ISSN:0162-1459    

DOI:10.1080/01621459.1957.10501363

出版商:Taylor & Francis Group    

年代:1957

数据来源: Taylor

ISSN:0162-1459    

DOI:10.1080/01621459.1957.10501364

出版商:Taylor & Francis Group    

年代:1957

数据来源: Taylor

ISSN:0162-1459    

DOI:10.1080/01621459.1957.10501365

出版商:Taylor & Francis Group    

年代:1957

数据来源: Taylor

摘要:

* Research sponsored by the Office of Naval Research under Contract Nonr 1636(00). Reproduction in whole or in part is permitted for any purpose of the United States Government.

ISSN:0162-1459    

DOI:10.1080/01621459.1957.10501366

出版商:Taylor & Francis Group    

年代:1957

数据来源: Taylor

ISSN:0162-1459    

DOI:10.1080/01621459.1957.10501367

出版商:Taylor & Francis Group    

年代:1957

数据来源: Taylor

ISSN:0162-1459    

DOI:10.1080/01621459.1957.10501368

出版商:Taylor & Francis Group    

年代:1957

数据来源: Taylor

摘要:

Tables are provided covering sample sizes up to 10. Interpretations are made for the variation of the coefficients and variances of the estimates as the number of censored observations vary. An index of experimental efficiency termed “efficiency per unit of waiting time” can be constructed with the tabulations provided. It is shown that (1) For the one-parameter single exponential distribution, the estimate of σ obtained from singly censored samples from the left has a smaller variance than that based upon singly censored samples from the right, provided that the number of missing observations is equal on both sides. (2) For the two-parameter single exponential distribution:a.The minimum value can be estimated more efficiently in samples censored from the right than in samples censored from the left, provided that the number of missing observations is equal;b.the estimate of the standard deviation in singly and doubly censored samples does not depend upon the side from which censoring takes place and depends only on the total number of missing observations.

ISSN:0162-1459    

DOI:10.1080/01621459.1957.10501369

出版商:Taylor & Francis Group    

年代:1957

数据来源: Taylor

摘要:

Letxandrbe respectively the observed mean and range in a random sample of sizenfrom a normal population. Values ofk1are tabulated which ensure a probability β that the pair of limitsx±k1rwill include at least a proportionpof the population. For tolerance limits of the formx±k2r, wherexis the grand mean andrthe mean range inNsamples of sizemeach, necessary tables of the factork2are provided for situations wheremis either 4 or 5, these being common group sizes in control chart analysis.

ISSN:0162-1459    

DOI:10.1080/01621459.1957.10501370

出版商:Taylor & Francis Group    

年代:1957

数据来源: Taylor