The camera never lies...

...but lots of people do, especially when they’re talking to researchers or otherwise responding to surveys. A part of it might be attributable to the Lake Wobegon effect, from the mythical town of Garrison Keillor, where it is said all the children are above average. More technically, another driver is social desirability bias. This is where the respondent wants to provide an answer that will be looked at by others as favorable.

• A recent poll asked Americans who they voted for in the last election. This poll showed Obama thrashing McCain by more than 20 percentage points -- far greater than the actual Obama margin of victory on Election Day.

• When people are asked if they voted in a presidential election, the percentage of self-reported turnout is inevitably 10-20 percent higher than actual turnout.

• About 40 percent of Americans say that they attend church regularly. Counting and tracking methodologies used to determine true church attendance found that about half that number can actually be found in the pews.

• A number of years ago, a survey found that upwards of five million people claimed to be New Yorker magazine readers—an unlikely number given that circulation was barely above half a million.

People want to be on the winning team, and want to look virtuous and smart. So when we ask them to self-report, we often get responses that are wildly inaccurate. Researchers are exploring tools such as anonymous online polling and expressionless computer avatars in order to obtain more accurate survey results. But no matter how sophisticated surveys become, there is no substitute for the careful capture of actual human behavior, as we do with video-enabled behavioral analytics to see into the realities of shoppers in the shopping aisles.

As Yogi Berra once said, “You can see a lot by observing.”