In the computation of the sum of squares as described, what is done to each score before the squaring step?

• A. The deviation (deviance) for each score is calculated
• B. The score is normalized to a z-score
• C. The score is log-transformed
• D. The score is scaled by the sample size

Answer: (A)

Sum of squares is built from how far each score is from the mean. The step you take before squaring is to compute the deviation of each score from the mean (subtract the mean from the score). Then you square those deviations and add them up. This centered measure ensures the dispersion around the mean is captured, since the deviations around the mean reflect true variability.

If you instead standardized to a z-score, you’d be scaling by the standard deviation, which changes the measure to a different statistic. A log transform changes the data’s distribution, not the basic deviation-from-mean step used in sum of squares. Scaling by the sample size would lead toward other statistics like a mean square or variance, not the raw sum of squares.