if i have value of coefficient of correlation (r) of one variable in 100 sample as 0.6 for male (50) and 0.7 for female (50). can i consider average of these (0.6+0.7/2= 0.65) as a value of 'r' when considering for whole sample (100)?
No, why do you want to do so?
http://mathoverflow.net/questions/57908/combining-correlation-coefficients
dear Mehmet thank you for your link information.
actually i want value of 'r' for whole sample but in some research paper combined value is not given rather it is present as value of 'r' for male and female sample separately. for example a study done among 100 people (50 male an 50 female) value of 'r' for two variables is presented as 0.6 for male and 0.7 for female but as a whole combined value is not given. so how can i get value of 'r' for 100 subjects.
No. The basic proposition in correlation coefficient ids to examine the relationship between two variables in a bivariate model. in partial correlation one can control for a variable. If you want ro calculate the r foer males , females , and combined you need to pool the data of the two sexes and calculate the correlation coefficient. simple averaging or mean will yield wrong results.
How to combine the multiple correlation values?
I have five groups of data, let say A1, A2, A3, A4, and A5.
The correlation between A1 and A2 is C12
The correlation between A1 and A3 is C13
The correlation between A1 and A4 is C14
The correlation between A1 and A5 is C15
How to combine all these correlations (C12, C13, C14, and C15 ) values for conveying that A1 is highly correlated with other group elements.
- Badges
- Science topic