Nice try though.
Given how sharply the answers shift between individual questions, they used at least two groups. Additionally, a neighbor of mine who is a registered Democrat was NOT asked Question 6, and he participated in this survey. (I live in the most heavily polled and test-marketed neighborhood in the US, since we happen to have a demographic/ideological/income distribution that matches America as a whole.)
The articles says they surveyed "Registered Voters" and "Probable Voters". Since it doesn't say "Republican Registered Voters" or "Republican Probable Voters", I think it's safe to assume they surveyed all voters.
You are, of course, assuming that they're not trying to slant the poll results to what they favor. It would not be the first time the USA Today folks have slanted their polls to generate results they want.