What is an assumption of testing differences between to r values?
The correlations are independent and come from different people.
Under the null hypothesis, what happens to the sampling distribution when rho1 = rho2?
The sampling distribution will be skewed. Because rho is bound by 1, so if rho does not equal 0 then some values will become possible but less likely.
Why convert r to r’?
To account for skewed distributions when population correlation does not equal 0
As an r value moves further from 0 how will the r’ transformation be affected?
Greater transformation from original value the further from 0. (Close to 0 requires less transformation)
What is a curvilinear relationship?
Where the direction of relationship eventually reverses
What factors affect r? 4
- Nonlinear relationships 2. Restriction of range 3. Extreme scores 4. Heterogenous sub samples
What are heterogenous sub samples?
Data which can be divided into distinct sets based on some other variable (eg, men and women vs. all people)
