96 Social Environment 



not go farther than adjoining states, where, as 

 may be seen from the tables, the conditions 

 studied do not, on the average, vary so very 

 much. In 1910, 85.5% of the population were 

 found living in the same group of states in 

 which ihey were born. Most migration, even 

 that which travels great distances, follows occu- 

 pational lines; hence the environment is not 

 likely to be essentially changed. It seems there- 

 fore to be justifiable to take the conditions of 

 the state of birth as indicating the general 

 features of the formative environment. And 

 here the principle concerning errors creeping 

 into the data of correlations finds an applica- 

 tion. The cases in which migration has oper- 

 ated to materially change environment can only 

 reduce correlations, not produce them; and it 

 may be taken for granted that any correlation 

 that may be found m the accompanying data is 

 indicative of a really stronger causal relation 

 that in fact exists. 



The next preliminary question concerns the 

 stability of the environment. It is evident that 

 if there is a great rearrangement of the states 

 within a period of two or three censuses when 

 ranked for any given condition, then the meas- 

 urement of the environment at the time of the 



