We ran across a good article on BusinessWeek a few weeks ago that shows us how graphs can easily be misused to demonstrate trends which may not exactly be connected.
This highlights the problem when people assume causation when it comes to trends in our societal data.
Click through to the article to see all the graphs.