The Journal of Things We Like (Lots)
Select Page
Ethan LaMothe & Donna Bobek, Are Individuals More Willing to Lie to a Computer or a Human? Evidence from a Tax Compliance Setting, 167 J. Business Ethics 157 (2020), available at ResearchGate.

Imagine your accountant asks you if you earned any income that wasn’t reported on a 1099 or W-2 this year, and you know that you have an extra $5000 of such income. Do you tell her? Probably. For starters, you might be worried that she is going to be suspicious if you lie to her. Something in your voice might give it away, or perhaps your income this year is lower than last year and she wants to know why. Further, you might have developed a rapport with your accountant, and lying to her might cause psychological discomfort.

Now imagine that you are debating whether to report the same income without an accountant’s help, using tax return preparation software. The software isn’t suspicious of your omission and doesn’t harbor any ill feelings about whether you are telling it the truth. In that case, you might be more likely to lie and not report the income. A fascinating new study by Ethan LaMothe and Donna Bobek confirms this intuition. In a survey of 211 participants, LaMothe and Bobek find that individuals may be more willing to lie to tax preparation software than they are to a human tax return preparer.

Survey respondents were presented with a hypothetical scenario where a taxpayer had earned $5000 of unreported income and was asked, by either a tax software program or a human preparer, if she had any additional income to report. The subjects were then asked how likely they thought it would be for the taxpayer to report all or none of the income. Those presented with the tax software scenario were significantly more likely to think the income would go unreported as compared to the human tax return preparer scenario.

LaMothe and Bobek posit two explanations for this increased willingness to lie to tax preparation software. First, people may perceive that human tax return preparers are more likely to detect a lie than tax software, noting that tax software programs are not designed to try to detect deception. Second, the authors hypothesize that individuals will feel greater “social presence” if they are dealing with a human, which may deter them from lying. The idea of social presence is that the salience of the relationship with the other person may make the psychological cost of lying higher. In other words, it may be harder to lie to a human in a social interaction than it is to lie to a cold, unfeeling machine. Further survey questions confirmed that participants generally believed that a human tax preparer would be more likely to detect deception than software, and there was higher social presence in the interaction with the human preparer as compared to the software. Social presence was gauged by asking participants Likert-scale questions rating whether the interaction with the preparer/software was cold-warm, impersonal-personal, passive-active, as well as asking whether there was rapport with the preparer/software and whether there was anonymity.

The study, like any survey, is not without limitations (as acknowledged by the authors). Survey participants answered questions about a hypothetical taxpayer, which may not accurately reflect real world behavior. It’s possible, for example, that many taxpayers report cash income on tax software regardless of whether the software can detect the omission or regardless of personal interaction. On the other hand, the effects of human interaction on taxpayer’s willingness to lie may be even stronger than the survey results suggest. This is because survey participants read about a fictional email interaction with a return preparer, whereas actual taxpayers often have live conversations with their tax return preparers.

This article should be of great interest to tax compliance scholars and those interested in the impact of technology on the tax system. As LaMothe and Bobek note, policymakers “should consider the tradeoffs inherent to tax software usage and perhaps view preparation method as a relevant risk factor when developing enforcement strategies.” The study also has interesting implications for tax software design. While it may be impossible to replicate true human interaction, software designers could consider ways to compensate. For example, the author’s findings arguably support proposals to make ethics more salient when individuals use tax software (such as those found here, here, and here). While automated tax preparation has tremendous efficiency advantages, the potential compliance costs of using tax software should not be overlooked.

Download PDF
Cite as: Kathleen DeLaney Thomas, Your Tax Software Doesn’t Know You’re Lying, JOTWELL (March 11, 2021) (reviewing Ethan LaMothe & Donna Bobek, Are Individuals More Willing to Lie to a Computer or a Human? Evidence from a Tax Compliance Setting, 167 J. Business Ethics 157 (2020), available at ResearchGate),