The abilities of man : their nature and measurement / by C. Spearman.

  • Charles Spearman
Date:
1927
Catalogue details

Licence: In copyright

Credit: The abilities of man : their nature and measurement / by C. Spearman. Source: Wellcome Collection.

    454/474 page 440
    parts of the n t’s. These two conditions can (with large popula¬ tions) always be satisfied. There is no need, be it observed, to construct the parts actually. By virtue of these divisions we get at once, 2rtxtx r, f , where tx stands for either V or t, (35) lxly j y ' // 9 w^/ h- 2 rtxv rtxv n ■XW [! +nx]4 [1 (36) (37) But, by employing these last three equations, we can get without difficulty the three of which we stand in need. That is to say, we find that the three correlations, Vi +... +tn)(t 1 +... +t>?> Vi + ••• +tn)v an(^ y „ n (<1 + ... -\-tn)W are equal respectively to the right-hand sides of the equations (32), (33) and (34). The required pools are, then, (^ +... + t'n) and (t” 4-... + iQ. As an example, we may take the experiments of Davey quoted on page 224. The tests to be examined for specific correlation were Inference and Likelihood, whilst there were six further tests to supply the two pools. Their mean correlation with Inference was -353 ; with Likelihood, -360 ; and with each other, •445. There ensue, using (35), (36) and (37), , _ 3 x '445 9a9b i + 2 x *445 709, and 3^ x *353 (1 + 2 x-445) 3^ x .360 (1 + 2 x-445) | = '445. r-452- 10. To find the value of a specific correlation.—This is a value which has often been quoted in the preceding volume. Since it is the value which the correlation between a and b becomes when the influence of g is eliminated, it can be determined at once by the method of partial correlations, as follows : ^Sa&b ~ • 3 ~ (b<6 ~ ^bg)/{^ — ^ag) ~ ^by) • (38)
    digitised image 1
    digitised image 2
    digitised image 3