9: Zero-Order, Part and Partial Correlations Flashcards
Zero-order^2
Total variance in y explained by xi, variance explained by each predictor.
Partial^2
Proportion of the remaining variance in y explained by xi, when the other predictors are removed.
Part^2
Unique variance in y explained by xi.
Shared predictor variance
Zero-order correlation^2, minus part correlation^2. Explained variance shared by multiple predictors.
Total explained variance
Sum(part correlations^2) + shared variance.
Variance not explained
1 - R^2
Partial correlation^2
Unique variance
_____________________
(Unexplained variance + unique variance)
Zero-order correlation
How much variance in y could x1 explain if it was the only predictor?
Part correlation
How much variance in y could x1 explain on their own?
Partial correlation
What proportion of variation in y that cannot be predicted by x1 can be explained by x2?
How is the p-value biased?
The p-value is biased by sample size, and doesn’t say anything about the size of the difference or relationship. Increasing sample size results in a smaller p-value.
What is power?
The ability to find a statistically significant effect when one exists, or the ability to reject the null hypothesis when it is false. Measured on a scale of 0 - 1, where 0 = no power.
Factors influencing power
Effect size, criterion significance level, sample size, statistical test, subjects design, and one/two-tailed hypotheses.
