Edward Tufte's book The Visual Display of Quantitative Information is a monumental book. He writes not about how to make your graphs look pretty, but how to display vast quantities of data and distill them down into useful graphics that communicate themselves effectively.
He provides examples of good and bad graphs, but more importantly, explains what exactly it is that makes those examples good and bad, and further generalizes it so you understand how to make good visualizations. If you don't want to shell out the money for it, it's probably at your library (remember those?).
Additionally, if I were you, I'd stay way from statistical approaches to displaying information unless you have some background or are willing to learn about it -- it tends to be highly technical and is probably too complex for what you're trying to do. Basic stats might help you, but not as much as Tufte will.
> Edward Tufte's book The Visual Display of Quantitative Information is a monumental book.
Agreed, its absolutely excellent. Thanks to Y Combinator for listing it in the book list.
> Additionally, if I were you, I'd stay way from statistical approaches to displaying information...
Not agreed. In my opinion you might have missed what I felt was a main point of that book: Always learn the appropriate statistics required to understand the data, choose a correct visualization method to communicate those statistics effectively, and once you've understood it fully, confirmed the results, and removed all the cruft, then publish it.
What I meant was to shy away from approaches that are PURELY based on statistics if you have no background in it, because it can get overwhelming quickly.
Of course, if it's worth it to invest the time required to have a fundamental understanding of statistics, by all means do so -- but if this is a one-time or a short-term project, I'm not sure the time commitment is worth it.
For pure data visualization, Tufte is absolutely the brilliant and helps you avoid a whole raft of bad behavior that leads to sloppy, hard to understand graphics. I can't second your suggestion enough.
I would add that his one day course is a great way to get started reading his books (especially if you can get your employer to pay for it). He's a great speaker and you get all the books as part of the course. I wish all the makers of charting libraries, toolkits, and data analysis software were more familiar with his work. It would save us from some truly awful junk.
One area where I disagree with him frequently is when he strays from data visualization into user interface design. In general, I find his user interface preferences result in UI's that are too cluttered. One of his main rules is that data presentations should be very dense. However, I disagree that this approach works as well for user interfaces as it does when visualizing data. If you look at his web site (http://www.edwardtufte.com/) you can see his UI philosophy on display. I find Don Norman to be much better in this area than Tufte.
He provides examples of good and bad graphs, but more importantly, explains what exactly it is that makes those examples good and bad, and further generalizes it so you understand how to make good visualizations. If you don't want to shell out the money for it, it's probably at your library (remember those?).
Additionally, if I were you, I'd stay way from statistical approaches to displaying information unless you have some background or are willing to learn about it -- it tends to be highly technical and is probably too complex for what you're trying to do. Basic stats might help you, but not as much as Tufte will.