What does semi-partial correlation measure?

What does semi-partial correlation measure?

• A. A measure of the relationship between two variables while adjusting for the effect that one or more additional variables have on one of those variables. If we call our variables x and y, it gives us a measure of the variance in y that x alone shares.
• B. The simple correlation between X and Y without corrections.
• C. The correlation between residuals of X after regressing on Y.
• D. The partial correlation controlling for all other variables.

Answer: (A)

Semi-partial correlation looks at the unique relationship between two variables after removing the influence of other variables from only one of them. In practice, you regress the predictor on the control variables and keep the residual (the part of the predictor not explained by the controls), then correlate that residual with the outcome. This tells you how much of the outcome’s variance is uniquely associated with that predictor beyond what the controls explain in the predictor itself. It’s different from the simple correlation (no adjustments) and from the partial correlation (which adjusts both variables). The squared semi-partial correlation shows the extra variance in the outcome explained when adding the predictor to a model that already includes the controls.

Semi-partial correlation looks at the unique relationship between two variables after removing the influence of other variables from only one of them. In practice, you regress the predictor on the control variables and keep the residual (the part of the predictor not explained by the controls), then correlate that residual with the outcome. This tells you how much of the outcome’s variance is uniquely associated with that predictor beyond what the controls explain in the predictor itself. It’s different from the simple correlation (no adjustments) and from the partial correlation (which adjusts both variables). The squared semi-partial correlation shows the extra variance in the outcome explained when adding the predictor to a model that already includes the controls.