In a discussion about service delivery yesterday, a colleague urged caution about the promotion of self-service transactions on the grounds that people will be less inhibited about lying to a faceless computer than they would to a contact centre agent (and still less so, presumably, to somebody in a face to face conversation).   It’s tempting to dismiss that as an excuse for luddism, but that’s a temptation which should probably be resisted, not least because there’s a possibility that the intuition of heavy net users is different from more tentative and less experienced users.  We know, for example, that some of our customers have greater confidence that a transaction will be acted on if they have spoken to somebody on the phone than if they complete it online.  That seems perverse to me – but I am not my own target market.  But I wasn’t aware of any evidence one way or another on the question of online honesty.  Before I had a chance to look for any, I came across a reference to a new academic study showing that research subjects lie more when emailing than when using pen and paper:

In one study, Belkin and her colleagues handed 48 full-time MBA students $89 to divide between themselves and another fictional party, who only knew the dollar amount fell somewhere between $5 and $100. There was one pre-condition: the other party had to accept whatever offer was made to them.

Using either e-mail or pen-and-paper communications, the MBA students reported the size of the pot—truthful or not—and how much the other party would get.

The results were staggering. Students using e-mail lied more than 92 percent of the time, while those using pen-and-paper lied slightly less than 64 percent. The rate of lying was almost 50 percent greater among those using e-mail than with those using pen-and-paper.

E-mailers also said they felt more justified in awarding the other party just $29 out of what they claimed was a total pot of about $56.

Although deceptive in their own right, pen-and-paper students were a little friendlier. On average, they passed along almost $34 out of a misrepresented pot of about $67.

That’s the language of a press release rather than sober academic enquiry – the paper on which it is based doesn’t seem to have been published yet – and it’s a bit hard to know how to interpret it.  What is being measured is a very stylised game being played with no consequences.  That suggests to me that this can tell us little, if anything, about real-world absolute levels of dishonesty, though it might tell us something about the relative levels of dishonesty between channels. 

There is nothing in this which tells me anything about the question I started with – though there is just enough to suggest that it’s not a completely unreasonable question.  On the assumption that more relevant research is unlikely to fall into my lap quite as easily as this did, I would be interested to hear of any studies of transactional compliance across channels.