Okay furuoshiki, I disagree with what you said. I think if we are going to see how stupid the "average American" is we would have to take people from all classes and demographics. Though you would have to split it even to the populace in that demograpic or class. So that would mean that we would have thugs and yuppies and rednecks and even highschool dropouts in the equation. They are a part of the "average American" since they do live here, it wouldn't be the "average middle class citizen" we are looking for but the whole of America. Now I am going to hit directly on a few points you threw out.
The point is that we cannot possibly take these stats at face value if we DON'T CONFIRM that these Americans being asked these questions are actually among the hard working (and relatively educated) middle class.
These facts alone no, but if you add them in with the middle class and the working class (which by the way is just as big as the middle class or a few percent off) and then add in the enough from the higher class to add a equal grouping from each group then we could, to judge the "average American" for otherwise you won't be looking for the "average American" in the end.
After all you wouldn't take a poor Frenchman off of the streets of Paris and ask him the location of the Eiffel Tower! And even if he actually knows the only reason is because he can beg tourists for money there.
And you would not proceed to base the intelligence of the French off of the opinions and knowledge of people living as drifters in France or those who have not had the educational opportunity the rest of us on this forum have likely been blessed with.
I would, but I know that I wouldn't be using them as the majority of my survey. If you base it off of one group how are we going to know the average between all groups. That would be like saying the other groups aren't a contribution to society, when they are, whether it be a bad or a good they still are a part of the population.
I would like to see that the people being asked these questions about geography and politics are of the majority demographic of this country.
This would be a White male in his late 20's or early 30's with some sort of post-secondary education located in an urban city who is neither a yuppie nor a thug. His father would probably be of Generation X and he will likely be of politically moderate orientation. Let's start with that and maybe I will accept these findings.
Again, if we did it that way how would we find the "average American"? If we aim at just one group what is that going to accomplish in the end for that type of research? Also if we only did white males between twenty and early thirty then what about women, what about the African-Americans and what about the Orientals and the Mexicans? I find as long as they are legally part of our country then aren't they a part of the "average American"? If you aim to specificly you won't accomplish anything with a survey like this.
This means No rednecks (sorry Texas, Kentucky, others), no high school drop-outs, no dead beat dads and no meat-heads! Let's begin there and see what results we get. I assure you they will be drastically different from what they are now. Even in Alabama! 
Yeah they would be drastically different, but it would be a totally different aim, we would be aiming for the "average middle class citizen" which I believe this was about the "average American" was it not? Last I checked high school drop-outs are still an American citizen, same with "meat-heads" and "dead beat dads" and "rednecks". They do contribute to our "average American" and they always will. Same as the lower class the working class and the rich, they all make up America. So to find the "average American" you have to look at all classes equal to the amount they value as. Just because a person is different doesn't make them any less of an American, and if you take that away from a pursuit to find the "average American" your results would be worth a penny in the gutter to me.
