Numerical quantities focus on expected values, graphical summaries on unexpected values.
Chapter 6 discusses parallel coordinate plots for studying many continuous variables simultaneously.
Scatterplots are superb for looking at bivariate data, but they are not so effective for exploring in higher dimensions. Matrices of scatterplots are pretty good, but do not really convey what multivariate structure there might be in the data, even with interactive linking. Dimension reduction methods (principal component analysis, factor analysis, or multidimensional scaling) have their supporters, and like all graphics-related approaches can provide insights when they ...