There is no survey that directly asks people whether they regret having persisted through a hard pursuit, nor one that asks a clean “did you give up too soon?” yes/no question. What the data does support is a directional asymmetry. Daniel Pink’s World Regret Survey, which collected more than 26,000 regrets from people in 134 countries, finds that regrets of inaction are about twice as prevalent as regrets of action in the long run — an approximate two-to-one split that widens with age. Quitting a difficult pursuit is structurally an inaction in the domain of continuation (a road not taken), while persisting is an action; so the regret of having quit tends to be the larger, more durable side. The rates shown here (action ~33%, inaction ~67%) are those action/inaction shares, not a measured “X% of persisters regret it” figure.
Roese and Summerville’s 2005 meta-analysis of 11 regret-ranking studies supports the same direction without supplying a ratio: regrets of inaction persist longer than regrets of action, and the biggest life-regret domains are education and career. Action regrets — the cost of persisting through difficulty that ends in failure — are intense in the immediate aftermath but fade as people rationalize and integrate the experience. The counterfactual of having given up remains available as a road not taken and becomes more vivid as opportunities close. Separately, Duckworth, Peterson, Matthews, and Kelly’s 2007 grit research across six populations found that perseverance predicts educational attainment, military retention, and competitive achievement above and beyond IQ and conscientiousness (grit accounted for about 4% of the variance in success outcomes) — evidence that sustained effort tends to pay off, not a regret measure.
The main caveat is the sunk-cost problem. The inaction-dominates asymmetry is a population-level pattern across all regrets; it does not distinguish between someone who rationally exited a genuinely failing effort and someone who gave up on something that would have succeeded with more time. Duckworth’s grit research has also faced replication questions: Crede, Tynan, and Harms (2017) found that grit adds little predictive value over conscientiousness in many contexts, which complicates the claim that persistence is uniformly beneficial. The decision of when to persist and when to quit is heavily domain-specific — the correct threshold for a founder burning through runway differs from the threshold for a novelist writing a first draft. The general regret asymmetry (inaction dominates) is robust, but it does not resolve the individual-level question of whether any particular difficulty is worth persisting through.