Blood groups in man / by R.R. Race and Ruth Sanger.

  • R.R. Race
Date:
[1950]
Catalogue details

Licence: Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0)

Credit: Blood groups in man / by R.R. Race and Ruth Sanger. Source: Wellcome Collection.

    66/324 page 46
    46 BLOOD GROUPS IN MAN with the numbers actually observed will at least show whether the results are internally consistent. Expected MM = 0*53165^ =0-28265 Expected MN = 0*53165 X 0*46835 X 2 = 0*49800 Expected NN = 0*46835^ = 0*21935 The absolute numbers, expected and observed, in the sample of 1,279 are: expected observed MM 0*28265X1,279= 361*5 363 MN 0*49800 X 1,279 = 636*9 634 NN 0*21935X1,279= 280*5 282 1,278*9 1,279 This is, by chance, a closer agreement than we could reasonably hope for. Fisher's test for MN results is applied as follows : _ [634' - (4 X 363 X 282)]^ [(2 X 363) + 634] [634 + (2 X 282)]^ = 0*027, foi^ degree of freedom, which corresponds to a probability of 0*87. This means that if ten sets of 1,279 random samples were tested w^ith equal accuracy, nine would be expected to fit less well. Had the agreement between the observed and the expected numbers been less close we should have v^anted to know how serious the deviation was; whether it was such as could be explained by chance, or whether there had been something wrong with the techmque of testing. The x^ test goes a long way towards giving the answer. If x^ had been, say, 3*8 for one degree of freedom the corresponding probabiHty is found from the tables to be 0*05. This means that chance alone would be capable of producing such a deviation, or a larger one, only once in testing twenty sets of 1,279 random samples. At about this level of probability we begin to be v^orried about our results and wonder if the disturbance is due to some error, and not wholly to chance. If x^ had been, say, 6*6 for one degree of freedom we should have been seriously worried, for a fit as bad or worse would only arise purely by chance once in a hundred testings of 1,279 samples.
    digitised image 1
    digitised image 2
    digitised image 3