Premium
THE SELECTION INDEX AND ITS TEST OF SIGNIFICANCE
Author(s) -
Dice Lee R.
Publication year - 1949
Publication title -
evolution
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.84
H-Index - 199
eISSN - 1558-5646
pISSN - 0014-3820
DOI - 10.1111/j.1558-5646.1949.tb00026.x
Subject(s) - dice , biology , selection (genetic algorithm) , index (typography) , test (biology) , library science , genealogy , statistics , history , computer science , artificial intelligence , ecology , mathematics , world wide web
Under experimental conditions in which equal numbers of individual animals of two types, A and B, are exposed to selection by a predator with the result that the numbers of individuals taken are a and b and the numbers not taken are c and d, respectively, making a combined total of n, then we may measure the strength of selection by the selection index: (a b)/(a + b). Should a greater number of type A than of type B individuals be taken (selection against A), then the selection index will be positive, while if more of B than of A are taken (selection against B), the index will be negative. The formula which I earlier proposed (Dice, 1947: 3) for the calculation of the chi-square of the difference between the numbers of A and of B taken under such conditions, unfortunately, is inappropriate. This has been pointed out to me by Don W. Hayne, who also has given other aid with this problem. The formula for chi-square earlier given, (a-b)2/(a+ b), is correct for a 1:1 ratio, as stated, but that formula takes account only of the successes in each of the two classes and neglects the failures in the same classes. The chi-square appropriate for testing the significance of the deviation index from zero is that for a 2 x 2 table: