Among couples who entered couple therapy in the most rigorous available randomised trial, 26-28% were separated or divorced at the 5-year follow-up, and only 48% showed sustained clinically significant improvement (Christensen et al., 2010, n=134 chronically distressed couples). Among US adults broadly, 66% have never attended or considered couple therapy (Grow Therapy/YouGov, December 2025, n=2,514), and nationally about a third of first marriages end in separation or divorce within 10 years (CDC/NCHS, National Survey of Family Growth) — a structural floor for the harm distressed non-attenders face, since deterioration among that sub-population runs higher still. The bilateral comparison points consistently toward inaction dominating: couples who do not seek help rarely improve on their own, and relationship distress compounds over the 2.68 years (Doherty et al., 2021, n=270) the average help-seeking couple waits before entering therapy — let alone for the much larger group that never goes at all.
The action-side data require careful reading. The Christensen trial recruited seriously and chronically distressed couples who volunteered for a clinical research program — a help-motivated, high-distress sample unlikely to represent the full range of couples who attend therapy in everyday clinical practice. Of the 26-28% who divorced within 5 years, many had entered therapy precisely to manage an already-failing relationship; the Grow Therapy survey finding that 56% of divorced therapy-attenders described their attendance as “a final attempt to prevent a split” suggests purposeful rather than regret-laden attendance. At the same time, only 56% of couples who showed clinically significant improvement at end of treatment maintained that improvement at the 5-year mark (Baucom et al., 2015), confirming that gains are not durable for a substantial minority. The 27% action-side proxy is the directly-measured separation/divorce rate; because only 48% of couples showed durable improvement, it likely understates outcome-based regret rather than overstating it.
The inaction side is harder to measure directly because no large-scale survey asks people who never attended therapy whether they regret that decision. The 33% inaction proxy is the CDC/NCHS National Survey of Family Growth probability that a first marriage ends in separation or divorce within 10 years, used as a conservative structural floor: distressed couples who receive no treatment rarely improve on their own, meaning actual deterioration or dissolution among relationship-distressed non-attenders is higher than this all-first-marriage figure. The 6-percentage-point inaction-dominance gap should not be read as a precise effect size; the uncertainty on the inaction side is wide and the true rate for distressed non-attenders is likely substantially higher than 33%. Both the action and inaction estimates share a common limitation: they reflect outcomes, not regret, and outcome data systematically miss the couples who were distressed, never sought help, and eventually experienced relationship improvement that made them glad they never went.