Correlations between parameters

#1
Hi

I have a parameter M (x) (ex weight) and one other N (y) (ex height).
We calculate the regression : y = ax+b
We have an other parameter O (y') (ex blood pressure), and always the parameter M (x) (ex weight): y' = a'x+b'

Here are the two :

*(y)........................-..........* (y')
*........................-..........+..*
*......................-.....+.........*
*...................-+.................*
*..............+..-....................*
*........+......-......................*
*...+.........-........................*
*............-..........................*
** * * * * * * * * * * * * * * (x)


Paramètre N : y = ax+b (+)
Paramètre O : y' = a'x+b' (-)

What can I say between N (y) (height) and O (y') (blood pressure) ???
Is it possible to calculate something like a correlation (with the angle between the two curves) ???

Regards
 

Dragan

Super Moderator
#2
Hi

1. y = ax+b (+)

2. y' = a'x+b' (-)

Is it possible to calculate something like a correlation (with the angle between the two curves) ???

Regards

In both regressions your using Weight as the independent variable. Why not consider computing the Pearson correlation between the error terms from both regressions. This is a Partial correlation coefficient between Height (H) and Blood Pressure (BP). Specificially, this correlation is an index of the linear association between H and BP having removed the (linear) effect of Weight from both H and BP.

You might also want to consider the semi-partials as well. For example, the correlation between Height and the error terms from the second regression.