I teach an introductory statistics class at the college level. I have a newspaper article that states that college graduates make m thousand per year where as non college graduates make n thousand per year. m is greater than n and the point of the article is that college tuition is worth it.
Then we discuss how realistic it is to use the mean income per year to describe ALL college graduates. We discuss variables that might have an effect on the amount of money earned - students usually mention major, career area, etc. I usually have to suggest number of years out of college, and they agree that they expect to make more later on than they do the first couple of years.
I then ask them to fill in the blanks of the following question. In x:____ years I will be making y:_____ thousand dollars a year. I request that they choose x between 1 and 10 and that everyone not have a y that ends in 0 or 5. I then go around the room and everyone calls out their values - this always brings a lot of laughs and provides insight into the goals of fellow students.
A scatterplot, correlation, and regression equation follow. I find it very easy to explain percent of variation explained by x using this example. It is also easy to explain that the predicted value is a better estimate than the mean y bar.
You may have had something more political in mind.