No study has ever measured how often people regret telling the truth. Kelly and Wang’s ten-week Notre Dame study — the closest thing to an honesty experiment — reports only the opposite direction: in weeks when participants told fewer lies, their close relationships improved and social interactions went more smoothly, and that relational gain statistically accounted for the study’s health benefits. It gives no week-level breakdown and no figure for relationship strain, so there is no measured truth-telling regret rate to report. The action side is held here at parity with the white-lie proxy to signal that ambiguity, not a measured cost. On the other side, DePaulo et al.’s landmark 1996 diary study of 147 Americans found that people tell between one and two lies per day — roughly a quarter of them “other-oriented,” told to spare someone’s feelings. When asked which of their lies they would tell again given a second chance, participants said about 75% — meaning roughly 25% of everyday lies are ones the liar would take back. That 25% is the closest the literature comes to a direct lie-regret rate from a systematic diary method. Interactions containing lies were also rated as less pleasant and less intimate than honest ones, and subsequent work by Cohen and Levine found that people randomly assigned to lie felt less connected to their partners than truth-tellers did — even when the lies were prosocial.
The case for honesty is not as clean as it sounds. Levine and Schweitzer (2015) demonstrated across four trust-game experiments that prosocial lies actually increase benevolence-based trust: participants who were deceived for their own benefit trusted the deceiver more, not less. Kelly and Wang’s 2012 Notre Dame study — where half of 110 participants were instructed to stop lying for ten weeks — found that the honesty group reported fewer mental-health and physical complaints, but the mechanism was relational: health improved only when honesty also improved relationships. When it did not, honesty carried a real interpersonal cost.
The balanced regret pattern reflects a genuine stalemate in the literature. Neither blanket honesty nor reflexive white-lying dominates in long-term regret. The action side carries no measured regret figure at all, so it is shown at parity with the inaction-side proxy (25% of lies people said they would not repeat); the two are close enough to constitute statistical noise given the thin, differently-designed evidence base. The most defensible reading is that context — who you are talking to, what is at stake, and whether you have earned the standing to deliver hard truths — matters more than any universal policy of honesty or diplomacy.