Among the roughly 20 million US adults flagged for problem gambling by the National Council on Problem Gambling’s 2024 NGAGE 3.0 survey, direct regret-about-continuing data simply does not exist. The closest peer-reviewed proxy is harm prevalence: Tulloch and colleagues’ 2024 national Australian study found that 14.7% of all gamblers reported at least one harm-to-self on the Gambling Harms Scale-10, concentrated in the at-risk and problem-gambler subgroups. The same study publishes the any-harm proportion for each PGSI band: 38.04% of low-risk, 75.78% of moderate-risk, and 96.40% of problem gamblers. Weighted across the at-risk-and-problem (PGSI 1+) population this entry targets, those bands combine to about 56% — not the near-universal rate that holds only within the smallest, problem-gambling group. We treat ~56% as the implied action-regret rate for problem and at-risk gamblers, acknowledging that harm is not synonymous with regret.
On the other side, the literature on quitting is striking for what is missing. The Schuler et al. 2016 scoping review of Gamblers Anonymous outcomes — covering 22 studies and 4 RCTs spanning six decades — identifies no study in which regret-about-quitting emerges as a meaningful outcome theme among sustained abstainers. Müller et al.’s 2017 German multicentre follow-up (N = 270) reports that the 41.6% who maintained 12-month abstinence showed the lowest psychopathology and significant personality improvements: decreased neuroticism, increased extraversion and conscientiousness. Neither Müller et al. nor the Schuler review reports any quality-of-life, dissatisfaction, or regret figure for abstainers, so there is no measured or derivable quit-regret rate — the ~5% shown for quitting is a nominal near-floor placeholder standing in for “undocumented and apparently rare,” not a proxy calculation. The direction is consistent across every recovery study identified.
The 58.4% in Müller’s cohort who either kept meeting gambling-disorder criteria or returned to sub-clinical gambling are NOT counted as quit- regret here. Relapse is a clinical event in a chronic-remitting condition; it reflects craving, situational triggers, and treatment adherence, not an expressed preference for the pre-quit state. Conflating relapse with regret would inflate the inaction-side figure by an order of magnitude on definitionally shaky grounds. The Slutske 2006 NESARC analysis adds a complementary signal: ~36–39% of lifetime pathological gamblers were in natural recovery (no past-year symptoms) without ever attending GA or formal treatment — recovery that sticks without external support is hard to reconcile with significant retrospective regret about having stopped.
The Gilovich pattern here is unambiguously action-dominated, and that matches the broader temporal-regret literature: short-horizon behavioural decisions where the action carries ongoing harm tend to generate action-regret that outpaces inaction-regret by wide margins. Where this entry diverges from the cleaner cases (alcohol, smoking) is the proxy depth on both sides — the action rate is a harm proxy, the inaction value is a placeholder, and the absence of a quit-regret literature is itself the strongest argument that the phenomenon is rare rather than well-characterised. A reader at PGSI 1+ should read the gap between the two sides as “every available signal points the same way,” not as a precise calibration of personal odds.