Final answer:
Canonical variate weights in Canonical Correlation Analysis are optimized to maximize the explained Sum of Squares and Cross-Products (SSCP), which represents the shared variance between two sets of variables.
Step-by-step explanation:
The question relates to the concept of Canonical Correlation Analysis (CCA), which is a multivariate statistical method. Canonical variate weights in CCA are chosen to maximize the explained Sum of Squares and Cross-Products (SSCP) between variables of two sets. The idea is for the linear combinations (canonical variates) of the variables in each dataset to have the highest possible correlation with each other, thereby maximizing the explained variance across the two sets of variables.
So, to answer the student's question directly, canonical variate weights are designed to maximize the b) explained SSCP. They do not aim to maximize power (a statistical term related to hypothesis testing) or error SSCP (which is related to the variability not explained by the model).