0
$\begingroup$

I have simple question, what is the difference when you use total and percentages when calculating correlation ? For example I want to calculate correlation between Europeans born in USA in each state every year (here I have total amount of people, and percentage of people) and average number of people who change work place every year in each state. So when I use total amount, correlation coefficient is lower than when I am using percentages. So my question is which one I should use ? And can I calculate correlation between average and percentages at all ?

I hope you understand my question.

$\endgroup$
  • $\begingroup$ Are you using percentages in both cases or only for the Europeans born in the USA? I think you should do it in both cases or not at all (perhaps for the second variable use total number of working people to divide by?). Using only percentages for the Europeans is confounded with total population, as is the number of people changing work place presumably. $\endgroup$ – Knarpie Mar 3 '17 at 12:56
  • $\begingroup$ @user151575 how did you compute percentage of people - % of total immigrants % of total (state-wise population ... $\endgroup$ – Subhash C. Davar Mar 3 '17 at 15:32
0
$\begingroup$

I think when doing calculation for populations (no percentages) you account for dynamics of the total (if both populations grow, the correlation between subgroups is higher).

| cite | improve this answer | |
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.