{"id":4838,"date":"2026-07-25T14:14:26","date_gmt":"2026-07-25T18:14:26","guid":{"rendered":"https:\/\/paulsyng.com\/blog\/?p=4838"},"modified":"2026-07-26T07:31:07","modified_gmt":"2026-07-26T11:31:07","slug":"everything-i-can-prove-about-therapy","status":"publish","type":"post","link":"https:\/\/paulsyng.com\/blog\/everything-i-can-prove-about-therapy\/","title":{"rendered":"Everything I can prove about therapy"},"content":{"rendered":"\n<p class=\"wp-block-paragraph\">Therapy is a product. In the US alone, it sits inside a behavioural health market worth roughly $89 billion a year (<a href=\"https:\/\/www.beckersbehavioralhealth.com\/behavioral-health-capital-investment\/the-u-s-behavioral-health-market-2025-to-2034-8-things-to-know\/\" target=\"_blank\" rel=\"noopener\">Becker&#8217;s<\/a>). It is sold by the hour, week after week, with no label on the box, no score at the end, and no receipt for whether it worked.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">I have spent the last week reading the strongest evidence for it and the strongest evidence against it. Meta-analyses, which are studies that pool many studies into one number. Trial registries. Textbooks. Regulator filings. Company earnings reports. This essay is the whole account. What therapy is, where it came from, what it genuinely does well, where the credit gets misassigned, what the trials say when you read all of them, why practitioners get angry when you say any of this out loud, who gets paid, and how to buy it well anyway.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Every number below links to its source. Where a claim is my own reasoning, I say so.<\/p>\n\n\n\n<iframe loading=\"lazy\" width=\"100%\" height=\"166\" scrolling=\"no\" frameborder=\"no\" allow=\"autoplay; encrypted-media\" src=\"https:\/\/w.soundcloud.com\/player\/?url=https%3A\/\/api.soundcloud.com\/tracks\/soundcloud%253Atracks%253A2368144901&#038;color=%23ff5500&#038;auto_play=false&#038;hide_related=false&#038;show_comments=true&#038;show_user=true&#038;show_reposts=false&#038;show_teaser=true\"><\/iframe><div style=\"font-size: 10px; color: #cccccc;line-break: anywhere;word-break: normal;overflow: hidden;white-space: nowrap;text-overflow: ellipsis; font-family: Interstate,Lucida Grande,Lucida Sans Unicode,Lucida Sans,Garuda,Verdana,Tahoma,sans-serif;font-weight: 100;\"><a href=\"https:\/\/soundcloud.com\/paulsyng\" title=\"Paul Syng\" target=\"_blank\" style=\"color: #cccccc; text-decoration: none;\" rel=\"noopener\">Paul Syng<\/a> \u00b7 <a href=\"https:\/\/soundcloud.com\/paulsyng\/everything-i-know-about\" title=\"Everything I know about therapy\" target=\"_blank\" style=\"color: #cccccc; text-decoration: none;\" rel=\"noopener\">Everything I know about therapy<\/a><\/div>\n\n\n\n<h3 class=\"wp-block-heading\">Where therapy came from<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">Talking to someone about your suffering is ancient. Priests did it. Elders did it. The paid, scheduled, private version is barely 130 years old.<\/p>\n\n\n\n<figure class=\"wp-block-image size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"575\" src=\"https:\/\/paulsyng.com\/blog\/wp-content\/uploads\/2026\/07\/josef-1024x575.png\" alt=\"\" class=\"wp-image-4842\" srcset=\"https:\/\/paulsyng.com\/blog\/wp-content\/uploads\/2026\/07\/josef-1024x575.png 1024w, https:\/\/paulsyng.com\/blog\/wp-content\/uploads\/2026\/07\/josef-300x168.png 300w, https:\/\/paulsyng.com\/blog\/wp-content\/uploads\/2026\/07\/josef.png 1240w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">In the 1890s, a Viennese doctor named Josef Breuer treated a patient who called his method the talking cure. Sigmund Freud built that into psychoanalysis, a theory that buried conflicts drive visible symptoms and that speaking them aloud releases them. The theory was never tested the way a drug gets tested. It spread anyway, because it was a story about yourself, told by an expert, in a room where you were the only subject. Few products in history have flattered the customer that well.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">By the 1950s psychoanalysis owned American psychiatry. Then in 1952 a psychologist named Hans Eysenck published a paper arguing that patients who got therapy recovered at about the same rate as patients who got nothing (<a href=\"https:\/\/hanseysenck.com\/wp-content\/uploads\/2019\/12\/1952_eysenck_-_the_effects_of_psychotherapy_an_evaluation_journal_of.pdf\" target=\"_blank\" rel=\"noopener\">Eysenck 1952<\/a>). The field spent 25 years trying to answer him. In 1977, Smith and Glass pooled hundreds of studies and found treated patients did better than about 3 out of 4 untreated ones, an effect around 0.68 on the standard scale researchers use, where 0.2 is small, 0.5 is medium, and 0.8 is large (<a href=\"https:\/\/www.cambridge.org\/core\/services\/aop-cambridge-core\/content\/view\/AD6D7EDCBA894C295E67503570BF8957\/S2045796018000057a.pdf\" target=\"_blank\" rel=\"noopener\">Barnett et al. reanalysis, which benchmarks the modern numbers against Smith and Glass<\/a>). Therapy declared victory and moved on.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">The decades after gave us Carl Rogers and warm listening, Aaron Beck and cognitive behavioural therapy, insurance billing codes, and finally the phone in your pocket. In 2024, BetterHelp, one app, collected $1.04 billion in revenue from people typing to therapists (<a href=\"https:\/\/ir.teladochealth.com\/news-and-events\/investor-news\/press-release-details\/2025\/Teladoc-Health-Reports-Full-Year-and-Fourth-Quarter-2024-Results\/default.aspx\" target=\"_blank\" rel=\"noopener\">Teladoc annual results<\/a>).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Radio and television ended up counting their audience because the people buying airtime were institutions with budgets to defend, and they kept asking. Therapy scaled to an $89 billion market with no buyer in that position. The customer pays directly, suffers privately, and improves or declines out of sight. Nobody with money on the line ever forced the industry to count.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">What genuinely works<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">Start with what the evidence supports, because a lot of it is real.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Therapy performs about as well as medication. The largest umbrella review ever run, 102 meta-analyses covering 3,782 trials and 650,514 patients, found therapy at 0.34 and drugs at 0.36 across mental disorders, a statistical tie (<a href=\"https:\/\/pubmed.ncbi.nlm.nih.gov\/35015359\/\" target=\"_blank\" rel=\"noopener\">Leichsenring et al. 2022<\/a>). Whatever you think of that number, therapy matches the medically approved standard of care.<\/p>\n\n\n\n<h1 class=\"wp-block-heading alignwide has-source-serif-pro-font-family has-custom-font\" style=\"font-family:source-serif-pro\">Anyone who tells you therapy does nothing has not read these.<\/h1>\n\n\n\n<p class=\"wp-block-paragraph\">For some conditions, the numbers are big. Exposure treatment for OCD shows effects around 1.31 to 1.39 (<a href=\"https:\/\/commonweb.unifr.ch\/artsdean\/pub\/gestens\/f\/as\/files\/4660\/43333_101926.pdf\" target=\"_blank\" rel=\"noopener\">\u00d6st meta-analysis<\/a>). Trauma-focused treatment for PTSD shows before-and-after gains around 1.70 (<a href=\"https:\/\/pubmed.ncbi.nlm.nih.gov\/37257304\/\" target=\"_blank\" rel=\"noopener\">2023 meta-analysis<\/a>). CBT for insomnia reaches 1.2 (<a href=\"https:\/\/pmc.ncbi.nlm.nih.gov\/articles\/PMC10002474\/\" target=\"_blank\" rel=\"noopener\">PMC review<\/a>). These are among the largest effects anywhere in mental health care. Anyone who tells you therapy does nothing has not read these.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">In depression trials, 41% of treated patients respond, against 17% getting usual care and 16% sitting on a waitlist (<a href=\"https:\/\/pmc.ncbi.nlm.nih.gov\/articles\/PMC8457213\/\" target=\"_blank\" rel=\"noopener\">Cuijpers et al. 2021<\/a>). Treatment also appears to cut the risk of getting worse, by somewhere between 39% and 61% relative to controls, though only 6% of trials bothered to measure deterioration at all, so hold the precision loosely (<a href=\"https:\/\/pubmed.ncbi.nlm.nih.gov\/34107050\/\" target=\"_blank\" rel=\"noopener\">Cuijpers &amp; Karyotaki 2021<\/a>).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">And people keep going back for reasons that are real even when they are unmeasured. An hour of undivided attention from a person paid to take your side. A place to say the unsayable without consequence. Structure in a week that has none. The expectation of getting better, which by itself moves outcomes by around 0.24 to 0.36 (<a href=\"https:\/\/academic.oup.com\/book\/30015\/chapter-abstract\/255628487\" target=\"_blank\" rel=\"noopener\">Constantino et al.<\/a>). None of that is fake. The question is what part of it requires a licensed professional at $200 an hour, and the industry has worked hard to make sure that question never gets a clean answer.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">The credit problem<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">The machine that keeps this market running is simple. People cannot tell what caused their own improvement.<\/p>\n\n\n\n<figure class=\"wp-block-image size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"576\" src=\"https:\/\/paulsyng.com\/blog\/wp-content\/uploads\/2026\/07\/Kahneman--1024x576.jpeg\" alt=\"\" class=\"wp-image-4844\" srcset=\"https:\/\/paulsyng.com\/blog\/wp-content\/uploads\/2026\/07\/Kahneman--1024x576.jpeg 1024w, https:\/\/paulsyng.com\/blog\/wp-content\/uploads\/2026\/07\/Kahneman--300x169.jpeg 300w, https:\/\/paulsyng.com\/blog\/wp-content\/uploads\/2026\/07\/Kahneman-.jpeg 1440w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">You start therapy at your worst. That is when people book. And extreme states tend to drift back toward normal on their own, a boring statistical fact called regression to the mean (<a href=\"https:\/\/doi.org\/10.1093\/ije\/dyh299\" target=\"_blank\" rel=\"noopener\">Barnett et al. 2005<\/a>). Kahneman told a story about flight instructors who screamed at cadets after bad landings and watched them improve, then concluded screaming works. The cadets were regressing to their own average. So are you, some of the time. That story is his field anecdote, and I label it as one, but the statistics underneath it are as solid as statistics get.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Untreated depression lifts on its own for about 12.5% of people within 3 months and 53% within a year (<a href=\"https:\/\/pubmed.ncbi.nlm.nih.gov\/34583099\/\" target=\"_blank\" rel=\"noopener\">Mekonen et al. 2022<\/a>, <a href=\"https:\/\/pubmed.ncbi.nlm.nih.gov\/22883473\/\" target=\"_blank\" rel=\"noopener\">Whiteford et al. 2013<\/a>). Every one of those people who happened to be in therapy at the time walked away crediting the therapist.<\/p>\n\n\n\n<h1 class=\"wp-block-heading alignwide has-source-serif-pro-font-family has-custom-font\" style=\"font-family:source-serif-pro\">People cannot tell what caused their own improvement.<\/h1>\n\n\n\n<p class=\"wp-block-paragraph\">Emergency rooms fill up when ice cream sales spike. Nobody thinks the cones did it. Summer did it. Both rose together. Your recovery and your therapy sessions rise together too, and the mind reaches for the visible cause it paid for.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Meanwhile, life keeps moving. You got a new job. You left the bad relationship. You started sleeping. You made a friend. Time passed. The trials handle this by randomizing, which is the whole reason trial numbers beat testimonials. Your story cannot separate the causes. A trial can.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">There is one more mechanism, and it works on memory itself. In lab experiments, researchers can implant entirely false autobiographical memories in roughly 26 to 30% of adults under suggestive questioning (<a href=\"https:\/\/journals.sagepub.com\/doi\/10.1177\/0956797617703667\" target=\"_blank\" rel=\"noopener\">Wade, Garry &amp; Pezdek 2018<\/a>, <a href=\"https:\/\/pubmed.ncbi.nlm.nih.gov\/27892833\/\" target=\"_blank\" rel=\"noopener\">Scoboria et al. 2017<\/a>). I want to be exact about the fence here. Nobody has measured whether ordinary retelling inside real, multi-session therapy rewrites a client&#8217;s history, and memory researchers themselves say memory is malleable under prolonged contamination and fairly reliable otherwise (<a href=\"https:\/\/discovery.ucl.ac.uk\/id\/eprint\/10091017\/3\/Brewin_Regaining%20the%20Consensus%20on%20Memory%20final%20accepted%20version.pdf\" target=\"_blank\" rel=\"noopener\">Brewin, Andrews &amp; Mickes<\/a>). The lab result tells you suggestion is a real force. It does not tell you your therapist used it.<\/p>\n\n\n\n<p class=\"has-large-font-size wp-block-paragraph\">So when someone says therapy saved my life, the honest response is that it might have. And that the same sentence gets said about horoscopes, juice cleanses, and copper bracelets, by people equally sincere, for the same 4 reasons: they started at the bottom, time passed, life changed, and they paid.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">What the trials actually say<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">Now the part that made practitioners angry. The best-studied corner of the whole field is therapy for adult depression. In 2018, Pim Cuijpers, the most published researcher in this literature and a man who runs therapy trials for a living, re-examined the whole pile (<a href=\"https:\/\/pmc.ncbi.nlm.nih.gov\/articles\/PMC6999031\/\" target=\"_blank\" rel=\"noopener\">Cuijpers et al. 2018<\/a>). The raw, uncorrected effect across all studies is 0.70. Restricting to Western countries drops it to 0.63. Drop the trials that compared therapy to sitting on a waitlist, which inflates the gap because waiting for help appears to actively hold people down, and it falls to 0.51. Keep only the well-run trials, the ones with proper randomization and blinded raters, and it falls to 0.38. Adjust for the studies that got run but never published, and it lands at 0.31.<\/p>\n\n\n\n<h1 class=\"wp-block-heading alignwide\"><strong>0.31 is a real effect. <\/strong><\/h1>\n\n\n\n<p class=\"wp-block-paragraph\">It is also less than half the advertised one, and in the small range on the scale the field itself uses. In practical terms, the fully corrected numbers work out to about 1 additional recovery for every 9 people treated, beyond what control conditions produce. An earlier paper comparing therapy against a sugar pill found 1 in 7 (<a href=\"https:\/\/pubmed.ncbi.nlm.nih.gov\/23552610\/\" target=\"_blank\" rel=\"noopener\">Cuijpers et al. 2013<\/a>). Both numbers are for depression specifically. OCD, PTSD, and insomnia run far better, as covered above. And to be precise about what 1 in 7 means: it is the added lift over placebo. It does not mean the other 6 heal on their own.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">The waitlist detail deserves its own sentence. Patients parked on a waitlist improve less than patients told to carry on as usual, 0.37 against 0.64 within their own arms (<a href=\"https:\/\/www.cambridge.org\/core\/journals\/epidemiology-and-psychiatric-sciences\/article\/overestimation-of-the-effect-sizes-of-psychotherapies-for-depression-in-waitlist-controlled-trials-a-metaanalytic-comparison-with-usual-care-controlled-trials\/C3C7E018609626AA01C4A9E0A8367B07\" target=\"_blank\" rel=\"noopener\">Cuijpers et al. 2024<\/a>). Being told to wait for rescue seems to stop people rescuing themselves. A large share of therapy&#8217;s advertised edge comes from being compared to that.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">The file drawer is measurable too. Of 55 US government-funded depression therapy trials, 13 were never published. Add the missing data back, and the effect drops from 0.52 to 0.39, a quarter of the published effect gone from unpublished results alone (<a href=\"https:\/\/pubmed.ncbi.nlm.nih.gov\/26422604\/\" target=\"_blank\" rel=\"noopener\">Driessen et al. 2015<\/a>).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Fairness note, because I hold everything to the same standard. A famous BMJ analysis found exercise beating therapy&#8217;s corrected number, walking and jogging at 0.62 (<a href=\"https:\/\/www.bmj.com\/content\/384\/bmj-2023-075847\" target=\"_blank\" rel=\"noopener\">Noetel et al. 2024<\/a>). Apply the same bias correction I just applied to therapy and the exercise effect collapses to a non-significant 0.18 in Cochrane&#8217;s high-quality subset (<a href=\"https:\/\/www.cochrane.org\/evidence\/CD004366_exercise-effective-treating-depression\" target=\"_blank\" rel=\"noopener\">Cochrane 2013<\/a>). Corrected numbers compare with corrected numbers. Exercise loses that comparison. Therapy survives it, smaller.<\/p>\n\n\n\n<p class=\"has-large-font-size wp-block-paragraph\">Now the misconceptions, one at a time, each with its study.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>The method matters.<\/strong> Barely. In 1936 a researcher named Saul Rosenzweig predicted all therapy schools would produce similar results, and 90 years of trials keep confirming it, with one narrow exception: CBT holds a small, real edge on primary symptoms since the 2014 update (<a href=\"https:\/\/pubmed.ncbi.nlm.nih.gov\/25238455\/\" target=\"_blank\" rel=\"noopener\">Marcus et al. 2014<\/a>). The name on the door moves your outcome less than the person behind it does (<a href=\"https:\/\/eprints.whiterose.ac.uk\/id\/eprint\/138182\/\" target=\"_blank\" rel=\"noopener\">Johns et al. 2019<\/a>).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Experience matters.<\/strong> No. A study tracking 170 therapists across almost 5 years found outcomes drifting slightly downward as therapists gained experience (<a href=\"https:\/\/pubmed.ncbi.nlm.nih.gov\/26751152\/\" target=\"_blank\" rel=\"noopener\">Goldberg et al. 2016<\/a>). A German replication found no relationship at all (<a href=\"https:\/\/bibliographie.ub.uni-mainz.de\/en\/catalog\/publication-131056\" target=\"_blank\" rel=\"noopener\">Erekson et al. 2022<\/a>). Twenty years on the couch-side chair predicts nothing.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Credentials matter.<\/strong> No. Degree type, license type, and institutional pedigree show no relationship to client outcomes (<a href=\"https:\/\/pubmed.ncbi.nlm.nih.gov\/26301425\/\" target=\"_blank\" rel=\"noopener\">Chow et al. 2015<\/a>).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Who matters, then?<\/strong> The person. Therapist identity explains roughly 5% of outcome variance, several times more than method (<a href=\"https:\/\/eprints.whiterose.ac.uk\/id\/eprint\/138182\/\" target=\"_blank\" rel=\"noopener\">Johns et al. 2019<\/a>). Some therapists reliably help. Some reliably do nothing. A few reliably harm. And nothing on a profile page, a diploma wall, or a directory listing tells you which one you booked.<\/p>\n\n\n\n<h1 class=\"wp-block-heading alignwide has-source-serif-pro-font-family has-custom-font\" style=\"font-family:source-serif-pro\">Some therapists reliably help. Some reliably do nothing. A few reliably harm. <\/h1>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Therapists know when clients get worse.<\/strong> They do not. In a study of 550 active cases, therapists were asked to flag which clients were deteriorating. They flagged 3. The real number was 40. Of their 3 flags, 1 was correct (<a href=\"https:\/\/pubmed.ncbi.nlm.nih.gov\/15609357\/\" target=\"_blank\" rel=\"noopener\">Hannan et al. 2005<\/a>).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Therapists know their own results.<\/strong> Read this one twice. Of 129 clinicians surveyed, the average rated their own skill at the 80th percentile. A quarter placed themselves in the top 10%. Zero rated themselves below average. And 47.7% said no client of theirs had ever gotten worse, against a documented deterioration base rate of 5 to 10% (<a href=\"https:\/\/pubmed.ncbi.nlm.nih.gov\/22662416\/\" target=\"_blank\" rel=\"noopener\">Walfish et al. 2012<\/a>).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Clinical intuition beats formulas.<\/strong> Sixty years of evidence says the reverse. Across 136 studies comparing expert judgment to simple statistical rules, the formula won 64 times, tied 64 times, and lost 8 (<a href=\"http:\/\/users.cla.umn.edu\/~nwaller\/prelim\/grovemeehlclinstixprediction.pdf\" target=\"_blank\" rel=\"noopener\">Grove &amp; Meehl<\/a>, <a href=\"http:\/\/zaldlab.psy.vanderbilt.edu\/resources\/wmg00pa.pdf\" target=\"_blank\" rel=\"noopener\">Grove et al. 2000<\/a>).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>It&#8217;s the patient&#8217;s fault when it fails.<\/strong> Randomization kills this one. Unmotivated patients land in both arms of a trial by design. Whatever the treatment effect is, motivation is already priced in.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>The field is improving.<\/strong> In youth therapy, 50 years of trials show effects flat for anxiety and ADHD and declining for depression and conduct problems (<a href=\"https:\/\/muttlab.fiu.edu\/_assets\/docs\/weisz-kuppens-ng-vaughn-coaxum-ugueto-eckshtain-corteselli-in-press-pps-are-youth-therapies-growing-stronger-over-time.pdf\" target=\"_blank\" rel=\"noopener\">Weisz et al. 2019<\/a>). Surgery counts its deaths. Aviation counts its near misses. Therapy counts its bookings.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Harm is rare and tracked.<\/strong> Nobody knows if it is rare, because almost nobody tracks it. Researchers screened 1,430 review-level publications and found only 57 original studies that monitored for harm at all (<a href=\"https:\/\/pubmed.ncbi.nlm.nih.gov\/38266653\/\" target=\"_blank\" rel=\"noopener\">Honkalampi et al. 2025<\/a>). Where someone did count, 85 trials covering 14,420 patients found that more than 1 in 21 patients reported a serious adverse event, with the causal link to treatment unclear for most (<a href=\"https:\/\/pubmed.ncbi.nlm.nih.gov\/38090772\/\" target=\"_blank\" rel=\"noopener\">Klatte et al.<\/a>). For balance: in the largest study with an independent safety board, zero of 260 serious events were attributed to the treatment itself (<a href=\"https:\/\/pmc.ncbi.nlm.nih.gov\/articles\/PMC2756150\/\" target=\"_blank\" rel=\"noopener\">NIDA monitoring study<\/a>). The honest summary is that the field runs blind in both directions. In a UK survey of 14,587 clients, 5.2% reported lasting bad effects (<a href=\"https:\/\/pubmed.ncbi.nlm.nih.gov\/26932486\/\" target=\"_blank\" rel=\"noopener\">Crawford et al. 2016<\/a>).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">A mainstream Wiley psychology textbook puts the label problem in one sentence: &#8220;In the case of pharmacological and other medical interventions, information about possible side effects or adverse reactions is usually a prominent feature of their packaging and marketing material. But no such warnings typically exist for psychological therapies&#8221; (<a href=\"https:\/\/cashmere.io\/v\/dcaiYuBBr\" target=\"_blank\" rel=\"noopener\">Davey, Psychology, ch. 16<\/a>). That is a teaching text, in its own voice, saying the fine print does not exist.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">Where it makes things worse<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">The harm evidence in children is real, and it is not spread evenly. It clusters in one place. Programs delivered to a whole classroom of kids who never asked for one.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">The largest test of this is a British trial called MYRIAD. 84 secondary schools, 8,376 students aged 11 to 13, randomly assigned to a mindfulness course or to normal teaching. The finding, in the authors&#8217; words, is that the course &#8220;resulted in worse scores on risk of depression and well-being in students at risk of mental health problems both at post intervention and 1-year follow-up,&#8221; and the differences were small (<a rel=\"noreferrer noopener\" target=\"_blank\" href=\"https:\/\/pubmed.ncbi.nlm.nih.gov\/35820993\/\">Kuyken et al. 2022<\/a>). More of it made things worse. &#8220;Higher dose and reach were associated with worse social-emotional-behavioural functioning at postintervention.&#8221; The conclusion the team published is that the course is &#8220;not indicated as a universal intervention&#8221; and &#8220;may be contraindicated for students with existing\/emerging mental health symptoms.&#8221;<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">An Australian team ran an 8-session DBT skills course through 4 schools, 1,071 students, 563 of them in the program. The program group got worse. Depression d = -0.22, anxiety d = -0.28, poorer emotion regulation, lower quality of life. 6 months later, they still reported worse relationships with their mothers and fathers than the control group did (Harvey et al. 2023). The authors wrote that the findings &#8220;raise discussion as to the potential for iatrogenic harm,&#8221; which is the clinical term for harm caused by the treatment. Fair warning on this one. It was not randomized; the schools were allocated by class, and the US federal clearinghouse that reviews child programs rated the study low quality because the two groups were not shown to be equivalent at the start (Prevention Services Clearinghouse). Call it a signal and wait for a better trial.<\/p>\n\n\n\n<h1 class=\"wp-block-heading alignwide has-source-serif-pro-font-family has-custom-font\" style=\"font-family:source-serif-pro\">Therapy makes things worse when it is given to people who were going to be fine.<\/h1>\n\n\n\n<p class=\"wp-block-paragraph\">A 2025 review went through 120 studies covering 112 school-based group mental health programs. 10 of the 112 reported at least one negative outcome, about 9%. Among the 15 studies rated high quality, 5 reported a negative outcome, which is 33% (<a rel=\"noreferrer noopener\" target=\"_blank\" href=\"https:\/\/pubmed.ncbi.nlm.nih.gov\/40101758\/\">Guzman-Holst et al. 2025<\/a>). The better the study, the more likely it found harm. And the harm landed on specific kids. High-risk students, boys, younger children, and children on free school meals.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Three mechanisms are on the table.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">The first is putting struggling kids in a room together. Tracking adolescent friendships showed that peer groups reward and rehearse problem behaviour, and the effect predicts later delinquency, substance use, and violence (<a rel=\"noreferrer noopener\" target=\"_blank\" href=\"https:\/\/pubmed.ncbi.nlm.nih.gov\/10510665\/\">Dishion, McCord &amp; Poulin 1999<\/a>). The clearest example is Scared Straight, the prison-visit program for at-risk teenagers. Cochrane pooled 7 trials and found the odds of offending went up, 1.68 (<a rel=\"noreferrer noopener\" target=\"_blank\" href=\"https:\/\/www.cochrane.org\/evidence\/CD002796_scared-straight-and-other-juvenile-awareness-programs-preventing-juvenile-delinquency\">Cochrane<\/a>). Doing nothing would have been better. Scared Straight is not therapy. It shares the room shape.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">The second is labelling. The idea is that teaching a 12-year-old to scan for symptoms teaches some of them to read ordinary bad days as a disorder, and the reading itself makes the day worse (<a rel=\"noreferrer noopener\" target=\"_blank\" href=\"https:\/\/doi.org\/10.1016\/j.newideapsych.2023.101010\">Foulkes &amp; Andrews 2023<\/a>). Flag this one honestly. It is a hypothesis the authors published to be tested, and it is not settled.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">The third is interrupting a recovery that was already happening. Group debriefing after a trauma was standard practice for years. Cochrane pooled 11 trials, found no reduction in distress or PTSD, and one trial found the debriefed group had higher PTSD risk a year later, odds 2.88 (<a rel=\"noreferrer noopener\" target=\"_blank\" href=\"https:\/\/pubmed.ncbi.nlm.nih.gov\/12076399\/\">Rose et al. 2002<\/a>). That evidence is mostly adults. The child version, 158 kids aged 7 to 18 after road accidents, found no harm and no benefit, with both groups improving on their own (<a rel=\"noreferrer noopener\" target=\"_blank\" href=\"https:\/\/pubmed.ncbi.nlm.nih.gov\/16423143\/\">Stallard et al. 2006<\/a>).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">This is the same trap mass screening ran into in medicine. Screen a whole population for a slow cancer, and you will find things that were never going to hurt anyone, then treat them, and the side effects land entirely on people who had nothing to gain. Give a distress program to 30 kids in period 4, and 27 of them had nothing to gain either. The mechanism is identical. Apply a treatment to a population where most members do not have the problem and the harm has nowhere to go except onto the healthy.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">The balance matters, so here is the other side of it. Aim the same interventions at children who actually screen positive, and the numbers turn. Targeted school programs produce a depression effect of -0.34 and an anxiety effect of -0.49 right after delivery, though the gains fade past 6 months and versions run by school staff instead of clinicians show no effect at all (<a href=\"https:\/\/centaur.reading.ac.uk\/89175\/3\/Revised%20manuscript%20(clean%20version)-%20Gee%20et%20al.%20Practitioner%20Review.pdf\" target=\"_blank\" rel=\"noreferrer noopener\">Gee et al. 2020<\/a>). A Cambridge cohort followed 14-year-olds with a mental health problem and found that those who had contact with services showed a greater drop in depressive symptoms, and the odds of clinical depression at 17 were more than 7 times higher in the group with no contact (<a href=\"https:\/\/www.cam.ac.uk\/research\/news\/teenagers-who-access-mental-health-services-see-significant-improvements-study-shows\" target=\"_blank\" rel=\"noreferrer noopener\">Neufeld et al. 2017<\/a>).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">So the rule that falls out of the evidence is about aim. A child with a diagnosed problem, seeing a trained clinician, is buying something that works. A wellbeing module rolled out to every student in year 8 is a different product with a different risk profile, and the field now has trials big enough to say so.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">The anger<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">I <a href=\"https:\/\/paulsyng.com\/blog\/why-therapy-is-a-roll-of-the-dice\/\" data-type=\"post\" data-id=\"4833\">published<\/a> a shorter version of this argument. Around 90 comments came in. I read every one and kept a ledger. Across the whole thread, 2 comments contained a link to a study. Two. The rest ran on credentials, anecdotes, and adjectives. The loudest objections came from inside the field.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">One commenter, a doctoral candidate working in engagement science, posted 5 comments and supplied one of the thread&#8217;s 2 citations. By Saturday morning, all 5 were gone. Whether he took them down or the platform did is not something I can see from outside. The study he cited reports that therapy and medication perform about the same, which is the post&#8217;s own parity claim, so the one citation offered against me argued my side.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">A certified EMDR therapist called the post an awfully divisive stance for a misleading claim, and disputed no number in it. A psychotherapist of 20 years wrote that very little of what I claimed has any relevance to his practice, his patients, or his success rates, and produced no figure from any of the three.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">One reader checked my sources carefully and landed 2 real corrections, which I conceded in public and have folded into this essay. That is what the process is supposed to look like. It happened once.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">The closest thing to agreement arrived by accident. The same 20-year psychotherapist opened his comment by saying it took him 25 therapists, as a patient, to find one he could work with. Then he closed it by saying none of the essay applies to him. Both halves sit in one comment. The first half is this essay&#8217;s central claim, learned the expensive way, 25 rolls of the dice.<\/p>\n\n\n\n<p class=\"has-large-font-size wp-block-paragraph\">Think about how strange the whole scene is. If I published data suggesting a popular bridge design fails under load, engineers would flood the thread with load calculations. If I claimed a surgical technique had a weak evidence base, surgeons would cite trials at me until I begged for mercy. When I published the therapy numbers, the professionals whose job is examining defensive reactions responded with defensive reactions, and the evidence total across 90 comments was 2 links. 2.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">The reason is structural, and it is worth saying plainly. A therapist&#8217;s income, identity, and 20 years of self-story sit on the premise that their judgment works. The Walfish data above shows the profession self-rates at the 80th percentile with zero members below average. When a number threatens the premise, the number gets treated as an attack. No malice required. It is what any of us do when the mortgage depends on not understanding something.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">But notice what no other field would tolerate. A pilot who said &#8220;I don&#8217;t need the instrument data, I feel the altitude&#8221; would never fly again. A fund manager who said &#8220;my returns can&#8217;t be measured, but my clients feel wealthier&#8221; would be laughed out of the building, then arrested. Therapy is the one profession where &#8220;I see it work every day in my office&#8221; still counts as an argument, from people whose own literature proves they cannot detect deterioration happening in that same office.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">The patient&#8217;s side of the ledger<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">The commenters kept saying the same thing clients say: I got better, therefore it worked. Walk through what actually happened during those months, because several forces were running at once and only one of them sent an invoice.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">You booked at your lowest point, so the drift back toward your own average was already in motion. Within a year, more than half of untreated depression lifts anyway (<a href=\"https:\/\/pubmed.ncbi.nlm.nih.gov\/22883473\/\" target=\"_blank\" rel=\"noopener\">Whiteford et al. 2013<\/a>). Your circumstances changed, because circumstances always change across 40 weeks. And you did work between sessions. You had the conversations, made the changes, sat with the discomfort. Then you handed the credit to the hour on the couch.<\/p>\n\n\n\n<p class=\"has-large-font-size wp-block-paragraph\">There is also a quieter force, and this part is my reasoning rather than a measured finding, so take it as such. A person who has paid $4,000 and told friends about their therapist has a self that needs the purchase to have worked. Doubting the therapy now means doubting their own judgment for 40 consecutive weeks. Minds do not volunteer for that. The more someone has spent, the more certain they become, and the certainty feels exactly like evidence from the inside.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Casinos understand this force better than anyone. The player who is down $4,000 does not conclude the game is rigged. He concludes his system needs one more session. The sunk money argues on behalf of the table. Therapy never rigged anything, but it collects the same defense from the same wiring.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">None of this proves your therapy did nothing. It proves you cannot know from the inside. That is what the trials are for, and the trials say 0.31 for depression once the accounting is honest.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">Who gets paid<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">Follow the money and the design choices explain themselves. The US behavioural health market runs around $89 billion a year (<a href=\"https:\/\/www.beckersbehavioralhealth.com\/behavioral-health-capital-investment\/the-u-s-behavioral-health-market-2025-to-2034-8-things-to-know\/\" target=\"_blank\" rel=\"noopener\">Becker&#8217;s<\/a>). A private session commonly costs $100 to $200, more in big cities (<a href=\"https:\/\/www.therapyroute.com\/article\/how-much-does-therapy-cost-in-the-usa-2025-by-therapyroute\" target=\"_blank\" rel=\"noopener\">TherapyRoute 2025<\/a>). Medicare paid $104.16 for a 45-minute session in 2025 (<a href=\"https:\/\/www.copehousecollective.com\/post\/therapy-rates\" target=\"_blank\" rel=\"noopener\">Copehouse rate guide<\/a>). BetterHelp alone collected $1.04 billion in 2024 from more than 1 million paying users, and refunded $84 million of it, a figure worth sitting with (<a href=\"https:\/\/ir.teladochealth.com\/news-and-events\/investor-news\/press-release-details\/2025\/Teladoc-Health-Reports-Full-Year-and-Fourth-Quarter-2024-Results\/default.aspx\" target=\"_blank\" rel=\"noopener\">Teladoc results<\/a>, <a href=\"https:\/\/bhbusiness.com\/2025\/02\/27\/betterhelp-wilts-while-other-teladoc-mental-health-services-bloom\/\" target=\"_blank\" rel=\"noopener\">Behavioural Health Business<\/a>). Its parent company then wrote down $790 million of BetterHelp&#8217;s value (<a href=\"https:\/\/www.healthcaredive.com\/news\/teladoc-1-billion-net-loss-2024-betterhelp-challenges\/741134\/\" target=\"_blank\" rel=\"noopener\">Healthcare Dive<\/a>).<\/p>\n\n\n\n<h1 class=\"wp-block-heading alignwide has-source-serif-pro-font-family has-custom-font\" style=\"font-family:source-serif-pro\">Now look at the billing model. Weekly. Open-ended. No defined endpoint, no cure milestone, no discharge criteria. Imagine plumbing billed weekly, open-ended, with the leak&#8217;s progress assessed by the plumber.<\/h1>\n\n\n\n<p class=\"wp-block-paragraph\">The revealing fact is that the field&#8217;s own research proposed an endpoint decades ago and the market ignored it. In 1986, researchers mapped the dose-response curve and found about 75% of clients get whatever benefit they will get by session 26, after which the curve flattens. The authors proposed 26 sessions as a rational limit (<a href=\"https:\/\/cashmere.io\/v\/dcaiYuBBr\" target=\"_blank\" rel=\"noopener\">Davey, Psychology, ch. 16<\/a>). No insurer, platform, or professional body ever adopted it. A finding that would cap revenue at 26 units per customer had no buyer. That is a choice, visible in the structure, and it tells you what the structure is for.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">The incentive review runs one way at every level. The therapist earns per session, so ending treatment is a pay cut. The platform earns per subscription month, so recovery is churn. The training institutes earn per trainee, so the pipeline needs demand to keep widening. Nobody in that chain is evil. Everybody in that chain is paid to not build the scoreboard.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">No scoreboard<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">The industry&#8217;s revealed preference shows up fastest in a comparison nobody wants to make. We built review systems for burritos and left the mind running on vibes.<\/p>\n\n\n\n<p class=\"has-large-font-size wp-block-paragraph\">Economists have a name for products you cannot evaluate even after buying them: credence goods. Car repairs, legal advice, therapy. You never learn what your outcome would have been without the purchase, so quality signals do the selling. Warmth. Office decor. A confident manner. The market rewards what clients can see, and what clients can see has nothing to do with the 5% therapist effect that actually moves outcomes.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Where counting exists, it embarrasses the sales pitch. England runs the world&#8217;s only national therapy scoreboard, NHS Talking Therapies, 15 years of mandatory outcome measurement. Its recovery rate has sat at roughly 50% since 2017, flat under continuous measurement (<a href=\"https:\/\/digital.nhs.uk\/data-and-information\/publications\/statistical\/nhs-talking-therapies-for-anxiety-and-depression-annual-reports\/2023-24\" target=\"_blank\" rel=\"noopener\">NHS England<\/a>). Count everyone referred rather than everyone who finished, and recovery falls to about 19%, because 63% of referrals never complete treatment (<a href=\"https:\/\/pmc.ncbi.nlm.nih.gov\/articles\/PMC12247062\/\" target=\"_blank\" rel=\"noopener\">BJPsych Open analysis<\/a>). And of those who do recover through the low-intensity route, 53% relapse within a year, 80% of those within 6 months (<a href=\"https:\/\/www.scribd.com\/document\/540748871\/Ali-et-al-2017-How-durable-is-the-effect-of-low-intensity-CBT-for-depression-and-anxiety\" target=\"_blank\" rel=\"noopener\">Ali et al. 2017<\/a>).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">The regulators count paperwork instead. Under 1% of UK-registered practitioners face any complaint in a year, and the complaints that do land are mostly about conduct and record-keeping rather than whether anyone got better (<a href=\"https:\/\/www.hcpc-uk.org\/globalassets\/resources\/reports\/fitness-to-practise\/hcpc-fitness-to-practise-annual-report-2023-24.pdf\" target=\"_blank\" rel=\"noopener\">HCPC annual report<\/a>). The UK&#8217;s national side-effect reporting system, built for drugs, received 15 psychotherapy-related reports in 2019 and 5 in 2020, against hundreds of thousands of medicine reports a year (<a href=\"https:\/\/www.gov.uk\/government\/publications\/freedom-of-information-responses-from-the-mhra-week-commencing-28-november-2022\/freedom-of-information-request-foi-221096-and-foi-221097\" target=\"_blank\" rel=\"noopener\">MHRA FOI<\/a>). The harm reporting channel exists. It just sits empty, because nobody is required to use it and nobody profits from filling it.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">I owe you the honest complication, because I argued for scoreboards and the evidence on scoreboards is mixed. When New York published cardiac surgeons&#8217; death rates, surgeons started refusing the sickest patients, and 41% of the state&#8217;s reported improvement traced to relabeling patients as sicker on paper (<a href=\"https:\/\/www.journals.uchicago.edu\/doi\/abs\/10.1086\/374180\" target=\"_blank\" rel=\"noopener\">Dranove et al. 2003<\/a>). Restaurant grades worked because an inspector can verify a kitchen independently (<a href=\"https:\/\/ideas.repec.org\/a\/oup\/qjecon\/v118y2003i2p409-451..html\" target=\"_blank\" rel=\"noopener\">Jin &amp; Leslie 2003<\/a>). Therapy outcomes have no independent inspector, so a naive public scoreboard would get gamed. What survives this complication is smaller and still damning: measurement inside the room, tracked session by session and visible to the client, is buildable today, cheap, and used by under 20% of US behavioural health clinicians at all and by around 5% at the every-session schedule the evidence is built on (<a href=\"https:\/\/pmc.ncbi.nlm.nih.gov\/articles\/PMC6584602\/\" target=\"_blank\" rel=\"noopener\">JAMA Psychiatry review<\/a>), with an independent pooled benefit around 0.14 to 0.17 (<a href=\"https:\/\/pmc.ncbi.nlm.nih.gov\/articles\/PMC11703940\/\" target=\"_blank\" rel=\"noopener\">De Jong et al. 2021<\/a>). Small, real, and mostly declined.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">What would change my mind<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">I claim the corrected depression effect is small, the person outweighs the method, and the market is built to avoid counting. Each claim has an exit. Watch for these. If someone re-runs the full bias-correction sequence on the newest trial corpus and the corrected number lands well above 0.5, the small-effect claim dies, and I will retract it. Nobody has published that as of this writing. If a large study ties a visible credential to client outcomes, the pick-by-person claim dies with it. The one study that tested credentials directly found nothing (<a href=\"https:\/\/pubmed.ncbi.nlm.nih.gov\/26301425\/\" target=\"_blank\" rel=\"noopener\">Chow et al. 2015<\/a>). If a functioning adverse-event registry for therapy appears anywhere on earth and fills with reports, the nobody-counts claim dies, and I will celebrate its death. And if routine outcome measurement gets adopted by a majority of practitioners in any country outside a government mandate, the incentive argument takes real damage. None of these exits has opened. The essay stands until one does.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">What therapy actually is<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">Strip the theory and the invoice, and what remains is real. This is the strongest true version of the product. Therapy is paid, structured, undivided attention from a person trained to tolerate what your friends cannot. For loneliness, that is no small thing. It is a rehearsal room for hard conversations. It is an appointment that forces a depressed person out of bed on a Tuesday, and the trial evidence suggests that scheduled activation is a big share of the active ingredient: behavioural activation alone, the simplest component, matches full CBT at 21% lower cost (<a href=\"https:\/\/pubmed.ncbi.nlm.nih.gov\/27461440\/\" target=\"_blank\" rel=\"noopener\">Richards et al. 2016<\/a>). It is a container for crisis, a place to put things too heavy to carry into a marriage or a friendship without breaking it.<\/p>\n\n\n\n<h1 class=\"wp-block-heading alignwide has-source-serif-pro-font-family has-custom-font\" style=\"font-family:source-serif-pro\">Therapy is paid, structured, undivided attention from a person trained to tolerate what your friends cannot.<\/h1>\n\n\n\n<p class=\"wp-block-paragraph\">For specific disorders with specific protocols, OCD, PTSD, panic, insomnia, it is genuinely effective treatment, among the best medicine has to offer. For the diffuse modern uses, direction-finding, self-understanding, weekly maintenance of an ordinary life, it is an expensive subscription to a kind stranger&#8217;s attention, with an effect around 0.31 where it has been honestly measured at all.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Both things are true at once. The tragedy of the industry is that it sells the second use at the first use&#8217;s reputation, and prices both the same.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">How to buy it well<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">The product is the person. Shop for the person and measure what happens. <\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Buy 3 first sessions with 3 different therapists.<\/strong> Nobody buys a used car without turning the key. The person explains roughly 5% of your outcome, more than the method does (<a href=\"https:\/\/eprints.whiterose.ac.uk\/id\/eprint\/138182\/\" target=\"_blank\" rel=\"noopener\">Johns et al. 2019<\/a>), so the interview is the highest-value hour in the whole process. Total cost of the extra 2 sessions: $200 to $400. Cost of 40 weeks with the wrong person: $4,000 and a year.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Match the protocol to the problem if you have a specific one.<\/strong> OCD gets exposure and response prevention. PTSD gets trauma-focused treatment. Insomnia gets CBT-I. These are the corners of the field with the big effect sizes, and a therapist who does not offer them for those conditions is selling you their comfort zone.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Ask 3 questions in the first session.<\/strong> How will we measure whether this is working? What does finished look like? What is your plan if I am not better by session 12? A good therapist has answers. The dose-response data says most of your gain arrives in the first 26 sessions, and the best predictor of a bad outcome is a client getting worse while nobody notices, which therapists detect at a rate of 1 in 40 (<a href=\"https:\/\/pubmed.ncbi.nlm.nih.gov\/15609357\/\" target=\"_blank\" rel=\"noopener\">Hannan et al. 2005<\/a>). So insist on the instrument: a 2-minute symptom questionnaire every session, with a trend visible to you. If the therapist refuses to measure, that is your answer about everything else.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Set a review date before you start.<\/strong> Session 8 or 10. On that date, compare scores against week 1. Better: continue. Flat or worse: change the person, and pay attention to how the therapist reacts to the question, because a professional who treats your review date as resistance has told you what the next 30 sessions will be.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Keep the cheap alternatives in the comparison.<\/strong> Guided online CBT matches face-to-face therapy in trials (<a href=\"https:\/\/pubmed.ncbi.nlm.nih.gov\/25273302\/\" target=\"_blank\" rel=\"noopener\">Andersson et al.<\/a>). Behavioural activation, which you can start from a workbook, matches full CBT (<a href=\"https:\/\/pubmed.ncbi.nlm.nih.gov\/27461440\/\" target=\"_blank\" rel=\"noopener\">Richards et al. 2016<\/a>). Neither carries a $200 hourly rate.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>And if it is working, stay.<\/strong> Nothing in this essay argues against a therapy that is measurably helping you. The target is the therapy that only feels like it is, at $200 an hour, on the strength of a feeling that cannot tell the difference.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">The industry will build the scoreboard the day customers stop buying without one. You are the customer. Start counting.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>Therapy is a product. In the US alone, it sits inside a behavioural health market worth roughly $89 billion a year (Becker&#8217;s). It is sold by the hour, week after week, with no label on the box, no score at the end, and no receipt for whether it worked. I have spent the last week [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":4840,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_coblocks_attr":"","_coblocks_dimensions":"","_coblocks_responsive_height":"","_coblocks_accordion_ie_support":"","footnotes":"","rank_math_title":"","rank_math_description":"","rank_math_canonical_url":"","rank_math_focus_keyword":""},"categories":[75],"tags":[],"class_list":["post-4838","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-feature"],"_links":{"self":[{"href":"https:\/\/paulsyng.com\/blog\/wp-json\/wp\/v2\/posts\/4838","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/paulsyng.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/paulsyng.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/paulsyng.com\/blog\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/paulsyng.com\/blog\/wp-json\/wp\/v2\/comments?post=4838"}],"version-history":[{"count":14,"href":"https:\/\/paulsyng.com\/blog\/wp-json\/wp\/v2\/posts\/4838\/revisions"}],"predecessor-version":[{"id":4860,"href":"https:\/\/paulsyng.com\/blog\/wp-json\/wp\/v2\/posts\/4838\/revisions\/4860"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/paulsyng.com\/blog\/wp-json\/wp\/v2\/media\/4840"}],"wp:attachment":[{"href":"https:\/\/paulsyng.com\/blog\/wp-json\/wp\/v2\/media?parent=4838"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/paulsyng.com\/blog\/wp-json\/wp\/v2\/categories?post=4838"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/paulsyng.com\/blog\/wp-json\/wp\/v2\/tags?post=4838"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}