What is the main difference between grand mean centring and group mean centring?

What is the main difference between grand mean centring and group mean centring?

• A. Grand mean centring uses the overall mean; group mean centring uses the mean within each group.
• B. Grand mean centring uses the minimum; group mean centring uses the maximum.
• C. They are identical procedures.
• D. Grand mean centring scales scores by standard deviation; group mean centring adds the mean.

Answer: (A)

Centering a predictor involves subtracting a mean from its values, and the reference mean used determines what the centered scores represent. Grand mean centring subtracts the grand mean—the average across all observations—from every value, so the centered score shows how far an observation is from the overall average. Group mean centring subtracts the mean within each group, so each observation is expressed as its deviation from its own group's average. This distinction matters in regression and multilevel models because it changes interpretation: grand-mean centring keeps the intercept tied to the overall mean and makes the slope reflect the effect at the overall average, while group-mean centering isolates within-group variation and helps separate within-group effects from between-group differences. Therefore the main difference is the reference mean used: grand mean versus group mean. The other statements are not accurate: centering does not involve minimum or maximum, the procedures are not identical, and centering does not scale by standard deviation or add a mean.

Centering a predictor involves subtracting a mean from its values, and the reference mean used determines what the centered scores represent. Grand mean centring subtracts the grand mean—the average across all observations—from every value, so the centered score shows how far an observation is from the overall average. Group mean centring subtracts the mean within each group, so each observation is expressed as its deviation from its own group's average. This distinction matters in regression and multilevel models because it changes interpretation: grand-mean centring keeps the intercept tied to the overall mean and makes the slope reflect the effect at the overall average, while group-mean centering isolates within-group variation and helps separate within-group effects from between-group differences. Therefore the main difference is the reference mean used: grand mean versus group mean. The other statements are not accurate: centering does not involve minimum or maximum, the procedures are not identical, and centering does not scale by standard deviation or add a mean.