Jim Horton has reminded us PR folk to tread carefully when working with statistics - especially in attempting to interpret the results. "How to Lie with Statistics" is a great primer for anyone interested in how to portray numbers, facts and trends accurately. The little book, first published in 1954, examines how polls, surveys and other data can be manipulated to support an observation or argument. Gee whiz graphs are just one gimmick singled out for criticism.
Edward Tufte continues to comment on how we choose to present data, and how our choices lead to confusion, misunderstanding or outright deceit. He has identified may examples of good information design, but just take a look at the graphs presented in this 1992 NYT article - they are stunning two and three dimensional representations of information.
Back when I was studying economic history, a prof required everyone to read "The Writing of Economics" by Deirdre/Donald McCloskey. Sold to us undergrads as "very short book with very few hard words," it takes less than a hundred pages to explain how to write about difficult subjects, often math-heavy, with clarity and precision.