We concur that the fitness of a mean is how accurate or true it is given set parameters of data set. We could then use tools like standard deviation to see how skewed the data set is from the mean line. For example a data set with high deviance would have a mean that does not necessarily depict the "halfway" point in the data. Extrapolating this data by using the sum of squares method would further reveal any discretionary variances in relationship to the mean?