Correlation and Partial Correlation Flashcards
What are the uses of correlation?
Reliability
Correlation and Partial Correlation
Underlying Principle of regression
What are the characteristics of correlational designs?
non-experimental - observe how variables change to see if they vary similarly
What are three ways to collect correlational data?
Observation in a natural setting
Use of archival data
Surveys
What is correlation as an analysis technique?
A statistical technique that measures and describes the relationship between two variables
What does correlation tell us about variables?
How one variable systematically changes as scores on another variable change.
The strength of the relationship
The direction of the relationship
The shape of the relationship
What is required to calculate a correlation?
Each individual must have a score for both variables
What is the most commonly reported correlation coefficient?
Pearson’s R (non parametric equivalent available)
What is an alternate name for standardised covariance?
Correlation co-efficient (r)
what is covariance?
A measure of the extent to which variables change consistently
How is covariance calculated?
A measure of the extent to which variables change consistently
how is covariance calculated?
multiplying both variables together
why is covariance standardised to create the correlation co-efficient?
because each unit for the variable is different (each variable is measured on a different scale/in different units)
what does a positive covariance indicate?
both deviations are positive or negative
what does a negative covariance indicate?
deviations go in opposite directions (one is positive, the other is negative)
What are the commonly used effect sizes for pearsons r?
0.1 - small effect
0.3 - medium effect
0.5 - large effect
What is the relationship between the numerator and denominator in the covariance calculation?
low covariance: denominator is greater than the numerator
High covariance: denominator is lower than the numerator
WHat is another term for standardised covariance?
correlation coefficient
What is the coefficient of determination?
r^2
What does the coefficient of determination tell us?
(r x r) * 100 = percentage of (overlapping) covariance between both variables
What is an important caveat of correlation?
COrrelation does not imply causation
what is the ‘third variable’ problem?
the possibility of spurious relationships due to an unmeasured third variable
What is the impact of restricting the range of scores on the correlation?
reduces the reliability of the computed r value
what are the assumptions of correlation?
Interval level data
Approximately normally distributed variables
Minimal/no outliers
Homoscedasticity
What are non-parametric alternatives to Person’s r?
Spearman’s r
what is homoscedasticity?
The spread of scores around the line of best fit is similar as you move along the axis.
What is partial correlation?
A technique that controls for the effect of a third variable (or more) on both variables of interest
How does partial correlation work?
It examines the relationship between the third variable and the two variables of interest, then removes the overlapping variability to keep the third variable constant
What is a confounding variable?
A variable that can explain some of the relationship between two other variables (the third variable)
What happens to the relationship between two variables when the effect of a confounding variable is removed?
The relationship can get stronger, weaker, or not change at all
