RELATION SEX DISTRIBUTION, SIZE OF GROUPS TO INTERMARRIAGE 129
same time. Now it is of prime importance to determine both their combined and several
effects on intermarriage—their combined effect, because if they do not account for the
actual proportions of intermarriage occurring other influences must be at work. The sepa~
rate influence of each is significant because it assists in explaining the present situation
and also constitutes a basis for prediction as to the future. The method of multiple and
partial correlation enables one to generalize on the basis of the experience of the stocks
examined, and the regression equation makes possible prediction of the expected. amount
of intermarriage for each group in terms of the three independent variables, namely, length
of residence, surplus males and size of the “origin” group.
In computing the correlation, the proportion of exogamous marriages among males of
each origin was taken as the dependent variable, Data for all European peoples except
the Bulgarians and Greeks were used, making 20 eases in all. The Greeks and Bulgarians
were omitted, because the extremely large proportions of surplus males would exercise an
undue influence and distort the result. Only white stocks were included, for the colour
barrier places the Negroes, Indians and Orientals in a class by themselves.
The value R = + 76 was obtained for the multiple coefficient and suitable tests were
applied to prove its reliability. The coefficient is quite large and demonstrates that length
of residence, surplus males and size of the population combine to exert a very important
influence on the proportion of males who have intermarried; and what is of equal importance,
it incidentally makes clear that these three factors of themselves are by no means adequate
to account for the entire spread between the figures for the several stocks. There is a
residuum which must be explained in terms of physical, psychological, social and other
peculiarities associated with the various groups. This will be elaborated in due course.
The regression equation is as follows—
X1=1-37Xy 4 0 12X3 — 15: 10X 4 — 46-56.
X1 =the percentage of married males in a given stock who have intermarried.
Xg =the percentage of the stock North American born.
X3 =the surplus males per 100 females (21 years and over).
Xs =the percentage which the adults of each origin constitute of the total
adult population of Canada.
The equation reveals several interesting facts; first, other things being equal, an addition
of one p.c. in the percentage North American born increases the expected proportion of males
intermarrying 1-37 p.c., and an addition of one p.c. in the surplus of adult males increases
intermarriage 0-12 pc. As was suggested above the influence of both increasing length of
residence and an excess of males is to raise the proportion of men marrying outside their
own stock.
A second point of interest is that a difference of one pc. in the percentage of North
American born is between 11 and 12 times more important from the point of view of the
proportion of males intermarrying than a similar percentage difference in the surplus of
males (21 years of age and over).
In the third place it is made clear by the equation that other things being equal the
larger the group the less marked is the tendency to intermarry. An increase of one pe.
in the proportion that the adults of a given origin constitute of the total adult population
of Canadas exerts a negative influence on intermarriage meny times greater than the combined
 influence of an increase of one p.c. in the proportion North American born and one
p.c. in the surplus of males.
Of course the chances of a change of one p.c. are by no means equal in the three cases.
A more definite idea of their actual importance under the conditions existing in 1921 is
obtained by substituting the standard deviations of Xo, X3 and X4, respectively, in the
regression equation.. It is found that fluctuations which actually occurred in the percentage
North American born had an influence on fluctuations in intermarriage over three times
greater than had differences in the proportions of surplus males. Similarly, the size of the
group, though a third less important than length of residence, was twice as potent as sex
distribution in determining the deviations of the proportions of males intermarrvineg from

74499—90