People, the computer and honesty

I’m going to take you back a few days to my post asking if American’s are really racist.  That was the one in which I questioned a poll that was struggling to figure out why the Messiah isn’t doing better and that concluded that it’s because Democrats are racists.  (Republicans are racists too, but they wouldn’t vote for the Messiah regardless.)

One of the things that made this poll different from all other polls was that it tried to do away with the Bradley Effect, which presumes that people are embarrassed to admit to pollsters that they might not vote for a black person.  To the end, the poll relied on computer responses, which the pollsters believed would be more honest.  Here’s what I had to say on that point:

[T]he pollsters relied on a “unique methodology,” “pioneered” by one company, Knowledge Networks, that thinks it can delve into people’s psyches.  First, they used online interviews, on the assumption that people will be more honest in the anonymity of cyberspace, which may well be true.

However, it may also be untrue.  Once online, people may feel more playful or obstreperous and be less honest.  Who knows?  You can only prove people’s honesty if you have an objective measure of the absolute truth.  And in opinion polls, what the heck is that measure?  We’re in a gray area that sees the pollsters assuming that their own biases about people’s probable beliefs constitute absolute truth, if only the pollsters can arrive at a methodology to prove that fact (in other words, there’s lots of room for bootstrapping and circular reasoning).

It turns out I may be on to something, and what I’m on to is that, in the vast internet, without social cues, people are often less honest, not more honest — as was shown by a study of business emails.  (And keep in mind that this study has the virtue of being controlled, so that truth and falsity could be objectively determined, which is distinct from the findings on racism, which may reflect the pollster’s own biases.). Here’s a summary of two new studies on the subject:

Office emails are more loaded with lies than traditional written communications like pen and paper, new research suggests. Previous research has supported this notion, also finding that phone calls are even more packed with prevarication.

A pair of new studies indicates email in the workplace is more deceptive than old fashioned writing, and that people feel quite justified in their distortions.

“There is a growing concern in the workplace over email communications, and it comes down to trust,” said Liuba Belkin, co-author of the studies and an assistant professor of management at Lehigh University. “You’re not afforded the luxury of seeing nonverbal and behavioral cues over email. And in an organizational context, that leaves a lot of room for misinterpretation and, as we saw in our study, intentional deception.”

While the wonderful, ground-breaking new polling process may indeed be more accurate, I’d say there’s compelling evidence to suspect that it may also be less accurate, as mischievous poll takers (and I’m sure there are many) use the vacuum created by the lack of human interaction to tell gross falsehoods.