With the insurance industry so focused on improving the customer experience, many have turned to behavioral economists for insights into how customers actually think, as opposed to how we think they think or how we want them to think. But... wow... behavioral economists sure aren't doing much for their credibility.
Two years ago, evidence surfaced that the data had been faked in the most famous study yet done with insurance customers. The study, conducted with 13,500 actual customers and published in a paper in 2012, supposedly showed that people were more truthful on their applications for auto insurance if they were simply asked to attest to their honesty at the beginning of the application rather than after they had completed it. But challenges eventually led to the publication of the raw data, which was obviously and rather clumsily faked, and the "landmark" paper was retracted in 2021. Blame fell on one of the four co-authors, Dan Ariely, a best-selling author who became chief behavioral officer at Lemonade in 2015 and held that position until 2020.
Now, another of those co-authors has been accused of faking the data in one of the two other studies cited in that 2012 paper and has been placed on leave by Harvard Business School. The much-cited researcher, Francesca Gino, has been accused of manufacturing the data for other studies, too.
Is it too much to ask for some honesty about honesty?
For good measure, the bloggers who exposed the fraud in Ariely's data and whose recent post at Data Colada raised the concerns about the data in Gino's say they suspect the third study cited in the 2012 paper is also based on made-up data. For now, they conclude:
"That’s right: Two different people independently faked data for two different studies in a paper about dishonesty."
Ariely has always denied faking his results, though neither he nor anyone else has ever explained why The Hartford, which let him conduct the test with its customers, would have fed him false data. As I noted in this Six Things when the initial fraud was exposed, I also found it odd that he waited so long to publish his results — he told me, personally, about the study in 2008 when we met at a conference where we both spoke, yet he didn't publish until 2012.
On the recent allegations, Gino declined to comment for this story in the New York Times, which details the challenges to her work.
Where does that leave us? Do we abandon behavioral economics as a tool for improving the customer experience and for encouraging people to buy insurance?
I suggest a distinction: We should rely less on behavioral economists but still think in terms of behavioral economics.
As the Times article shows, it's easy for behavioral economists to fake the data, and it seems they have incentives to do so. Behavior changes when fame and fortune beckon, even for the best-intentioned — as these economists should know better than anyone. So, I'll be more skeptical of the specifics in many of the studies they cite, even as I'll continue to read books like "Nudge" to provoke my thinking.
Even as we perhaps wean ourselves from the pronouncements of the famous behavioral economists, though, we can all see how behavioral psychology shapes consumers. Just think about how social media companies manipulate us to keep control of our eyeballs. They have made a fortune in the "attention economy" by continually tweaking their algorithms to make us mad or sad or intrigued enough to click on just one more thing or to just keep scrolling.
Certainly, the federal government thinks consumers can be manipulated. The Federal Trade Commission won an $18.5 million settlement from Publishers Clearing House this week on the grounds that it used "dark patterns" to trick customers into paying for products or giving up their data. Last week, the FTC sued Amazon, also for allegedly used dark patterns, to trick customers into signing up for Prime and to keep them from dropping the service. (Amazon denies doing anything nefarious.)
There's also a rock solid basis for behavioral economics, anchored in the work of Daniel Kahneman on cognitive biases, that makes clear that we have to take humans as they are, not as the perfectly rational actors that traditional economists want them to be. Nobody actually thinks based on utility functions and indifference curves, even though they look so clean on a whiteboard.
So, insurers should continue to lean in to behavioral psychology as they try to figure out what bothers customers throughout the insurance lifecycle and as they try to find ways to motivate customers to buy the protection they need. Insurers should just do it primarily on their own, using A/B testing.
Did that email trigger a response? How about that one? Or that one? If we change X about the claims process for half our customers, what does that do to customer satisfaction? Does an increase in satisfaction translate into retention? Cross-selling? Up-selling? And so on and on and on, 24/7/365.
You might even try having people attest to their honesty at the start of an application. Just don't expect the results you were promised in that 2012 paper.
Cheers,
Paul