Sometimes poll data is, in historical terms, unreliable. That is to say, the polls are intentionally designed to lie.
As Mark Twain and Benjamin Disraeli noted, there are lies, damned lies, and statistics. You frequently see distortions of this sort in educational surveys that are designed to show the DECLENSION! DECLENSION! of today's youth.
Now, I'm all for bashing you all. Stinky juniors. But dude. You ought not to just make stuff up.
Nate Silver over at Fivethirty eights has caught the Startegic Vision pollsters red handed. Not only did they lie, they lied stupidly. In class, I had (or will have) you answer ten questions from the citizenship test. Here's how Startegic Vision said 1000 high school students did:
(click to embiggen)
Really, not a single student got more than 7 right? There have to be at least a few Tueting-style nerds somewhere in the sample set (if you got nine or ten right, you are, sad to say, a Tueting-style nerd. Deal with it).
Check out these charts from the poll "results." Knowing high school students, do you think that, even if a student was unaware of the answer to the question "what are the two major political parties in the United States," that every single wrong answer would be "Republicans and Communists?" or that instead of saying "don't know," at least some would have said at least one of the two parties? Or, for the terms of Senators, every single respondent who was wrong would choose a random number and they were all divisible by two? The poll itself was laughably fake.
Even so, the poll results were published nationally and mentioned in your local DNR. Liberals used the data to laugh at the red states and conservatives used the data to show how bad the public schools are - agendas trumped reasoned analysis of the data.
Historians often have to use statistical analysis when traditional written primary sources are lacking or unreliable. Go read Nate Silver's analytical smack down. (Here is the original Strategic Vision poll).