How do I know? A fascinating report on new radio ratings systems shows that when people self-report their classical music listening habits, they inflate their estimate by that margin—probably to give the impression that they’re smarter.

The discrepancy revealed itself when radio surveys were switched from paper-based diaries to more accurate pager-like “people meters.”

Classical music listening went down. Soft-rock listening by men went up—by 16%.

The takeaway? Don’t trust what people say. Trust what they do.

Objective data shows that interest in analytics is gaining on interest in surveys

This is particularly important with web marketing. With the ceaseless launch of new websites and technologies, it’s easy to get caught up in hype. And people don’t want to seem like they’re getting left behind.

Online video and social media marketing are some recent examples. People love to say they’re on Facebook, Twitter, LinkedIn and other social networks, and using them extensively. But often they want to appear savvier than they are.

There’s a similar bias with market research and advertising focus groups. Often, ads that work well in the field perform poorly in testing—a phenomenon discovered in the early days of direct marketing. Why? Because people don’t want to admit how they would really respond to more direct pleas rather than those that are more artful.

So beware self-reported survey data. At the very least, temper them with objective analytics.