especially considering where some of them were conducted.

http://home.businesswire.com/portal/site/google/index.jsp?ndmViewId=news_view&newsId=20040903005355&newsLang=en

People are notorious for guessing WRONG about what other people think or do.

Sometimes in my classes I ask students to write down (anonymously) what percentage of members of that class had sex last weekend, and how many of them smoked pot last weekend. Then I have them give yes or no answers to the same questions. Turned in an tallied, the gap between what they THINK others do and what they ACTUALLY did is HUGE. They think 20-40% have sex (but only ~5 report having it) and they think 20-30% smoke weed, but less than half that amount report smoking it. The questions don't much matter, people's judgments of others actions are not reliable predictors of the actual actions.

It's interesting, but not very useful in predicting behavior, to ask these types of questions.