-
作者：Bartlett, MS
摘要：The simultaneous distribution of sample canonical correlations between 2 sets of statistical variates, when the population canonical correlations are zero, gives rise to more than one possible test of significance for the largest correlation. One test depends on the distribution of the largest correlation for given values of the remainder, and the logical position of this test among tests of significance in multivariate analysis is discussed.
-
作者：Thompson, CM
摘要：The incomplete [beta]-function plays an important part in many biological applications of statistical methods. In particular, the significance levels of Fisher''s z-distribution or Snedecor''s F-distribution can be obtained easily from those of the incomplete [beta]-function, as explained in this paper. The significance levels are tabulated to 5 significant figures for 50, 25, 10, 5, 2.5, 1 and 0.5% probabilities. The degrees of freedom for which entries are given are as follows: v 1 = l (1) 1...
-
作者：Neyman, J
作者单位：University of California System; University of California Berkeley
-
作者：Garwood, F
摘要：The usual practical maximum likelihood treatment of dosage-mortality problems (consisting of the transformation of % surviving into probits, adjustment and weighting of the latter, and calculation of successive regression lines) is shown to be equivalent to calculating successive corrections to the regression coefficients. The process is exactly equivalent to the method, given elsewhere by Fisher, of obtaining the maximum likelihood estimates of the parameters defining the distribution. A refi...
-
作者：Thompson, CM
摘要：The percentage points (or significance levels) are given to 6 significant figures for 99.5, 99, 97.5. 95, 90, 75, 50. 25, 10, 5, 2.5, 1 and 0.5% probabilities. Of these the 99.5, 97.5, 75 25 2.5 and 0.5%; levels do not appear in any previous tables. The number of degrees of freedom, v, takes the values: 1 (1) 30 (10) 100. The x 2 distribution occurs frequently in modern statistical techniques: e.g., in K. Pearson''s ''''goodness of fit test and in tests of Mendelian ratios.
-
作者：Simaika, JB
摘要：The power function of Hotelling''s generalization of Student''s t-test is known to depend only on a single function of the population parameters. This test is uniformly most powerful amongst all tests whose power functions depend on the same function of the population parameters. A similar property holds for the test of sig-nificance of the multiple correlation coefficient.
-
作者：Pearson, ES
摘要：This note is introductory to the 2 papers by Hsu and Simaika immediately following. It traces the development of certain ideas involved in the testing of statistical hypotheses, starting from Neyman and Pearson''s conception of a uniformly most powerful test. When a test concerns the values of c [greater than or equal to] 2 parameters a fresh formulation of the requirements of a satisfactory test becomes necessary. Several lines of attack are open; one of these is that followed by Hsu and Sima...
