Rosenberg's notion of standardization as statistical control for percentage tables is extended to more than three variables. After the data are arranged as "recursive proportions," simple calculations allow one (1) to find partials, causal effects, spurious portions, and indirect portions for associations, (2) to carry out realistic simulations or projections using variable levels, and (3) to gain increased insight into interaction effects. I argue that the approach is not actually novel, rather it allows one to present well-established statistical matters in a way that is unusually clear and palatable to the non-technical reader.