6 THE RELATIVE STRENGTH OF 



conditions we wish to modify, we must study their 

 correlation with as many factors as we can possibly 

 measure. In the choice of these factors we must 

 of course be guided by the reasonable probability 

 of association and by the limits of human life and 

 energy. The correlations of a multiplicity of factors 

 being known we may justifiably assume that the 

 factors with the highest correlations are, among 

 those dealt with by us, the most important, and then 

 the process of "partial correlation" will guide us still 

 further towards a final judgment of what fundament- 

 ally are social cause and effect. 



We admit to the full that even then we may not 

 have avoided all danger of pitfalls, that we may have 

 overlooked possible factors, and that spurious cor- 

 relation may have arisen from all sorts of disregarded 

 selective processes. The prudent statistician will 

 always advance his conclusions with a word of 

 caution ; he will simply state that they are those which 

 reasonably follow from the data provided. But at 

 the same time he will not hesitate to proclaim that 

 in the present state of our knowledge the calculus of 

 correlation is the sole rational and effective method 

 available for attacking these urgent social problems. 

 If that calculus throws no light, when properly 

 applied, on social dynamics, then the only solution 

 is to develop a finer statistical calculus ; no other 

 instrument, least of all general reasoning with appeal 

 to social or moral prejudgments, can at present aid 



the correlation between right and left femur in man is '96 which is 

 practically unity, i.e. almost perfect dependence as wc should expect. 

 The inheritance between stature of father and son is "51, half-way 

 between absolute dependence and absolute independence. 



