Skip to content
Dandelion Psychotherapy
DandelionPsychotherapy
All writings
15 min read

Does Transactional Analysis Work? Evidence and Criticisms

By the Dandelion Reflect team

Last updated:

Does transactional analysis work? What the research shows, who wrote the main reviews, the fair criticisms, and what it all means when you choose a therapist.

Goes further than TA 101, lesson 1: What is transactional analysis?.

TL;DR: The research on transactional analysis (TA) mostly finds that clients improve. A 2021 meta-analysis of 75 studies reported moderate to large effects, and practice-based studies in UK clinics found results in line with national benchmarks for anxiety and depression. The evidence also has clear limits: few randomised trials, small samples, many studies without a control group, outcomes measured by questionnaires clients fill in, and reviews written by TA practitioners and published in TA journals. None of that makes the findings false, but TA's case rests on research that is still light on controlled trials, and its main reviews and most recent studies come from TA practitioners and appear in TA journals. When you choose a therapist, research on the therapy relationship suggests that how well you and your therapist work together matters a great deal, whatever the method.

"Does TA work?" sounds like a yes-or-no question, but it hides several smaller ones. Do people who have TA therapy feel better? Would they have felt better anyway? Does TA do better than other therapies, or about the same? And who did the research? The first TA 101 lesson gives a short answer: a real but modest evidence base. This post looks at the studies behind that answer, the people who wrote them, and the criticisms that deserve a hearing.

What "Evidence" Means for a Therapy

Studies of therapy come in different designs, and each one can rule out different explanations for why a client got better.

  • Client surveys ask people afterwards whether therapy helped. They are cheap and can reach thousands of people, but they can't separate the therapy from time passing, from who chose to reply, or from the wish to say something kind about a therapist you liked.
  • Single-case studies follow one client closely. The better designs collect questionnaires every week and then argue both sides: was it the therapy, or was it something else in the person's life?
  • Practice-based studies measure clients before and after therapy in ordinary clinics. They show what happens in real conditions, but without a comparison group you can't tell how much of the change the therapy caused.
  • Matched comparison studies set clients in one therapy against similar clients in another. That is closer to a fair test, though the two groups may still differ in ways nobody measured.
  • Randomised controlled trials (RCTs) assign people by chance to the therapy or to something else. Chance assignment is the main way researchers rule out "they would have improved anyway" and "the groups were different to start with".
  • Reviews and meta-analyses pool many studies. They are only as strong as the studies inside them.
Kinds of evidence for TA, with a rough rating of each designClient surveys: asks Did clients feel helped? TA example: Novey 2002: 932 clients of 27 TA therapists. Rough design rating 1 on a 1-4 scale. Single-case studies: asks Did this person change, and why? TA example: Widdowson 2012-2013: four depression cases, outside judges. Rough design rating 2 on a 1-4 scale. Practice-based studies: asks Do clients improve in real clinics? TA example: van Rijn and colleagues 2011, 2016: UK clinics, no untreated group. Rough design rating 2 on a 1-4 scale. Matched comparisons: asks Better than similar clients elsewhere? TA example: Horn et al. 2015: 67 matched pairs, inpatient programme. Rough design rating 3 on a 1-4 scale. Randomised trials: asks Better than a comparison, with groups set by chance? TA example: Very few. Vos and van Rijn 2025: small pilot. Rough design rating 4 on a 1-4 scale. Reviews and meta-analyses pool the studies above and are only as strong as what goes into them. TA examples: Ohlsson 2010, 60 outcome studies listed (his own non-systematic count); Vos and van Rijn 2021, meta-analysis of 75 studies.Kind of studyTA examplesDesign strength (rough)Client surveysDid clients feel helped?Novey 2002: 932 clientsof 27 TA therapistsSingle-case studiesDid this person change, and why?Widdowson 2012-2013: fourdepression cases, outside judgesPractice-based studiesDo clients improve in real clinics?van Rijn and colleagues 2011,2016: UK clinics, no untreated groupMatched comparisonsBetter than similar clients elsewhere?Horn et al. 2015: 67 matchedpairs, inpatient programmeRandomised trialsBetter than a comparison, withgroups set by chance?Very few. Vos and van Rijn2025: small pilotReviews and meta-analysesPool the studies above: only asstrong as what goes into themOhlsson 2010: 60 outcome studies listed(his own non-systematic count)Vos and van Rijn 2021: meta-analysis of 75 studies
A rough guide, drawn by Dandelion. Designs lower down can rule out more alternative explanations, such as people improving anyway. Some TA researchers, including Ohlsson, argue that randomised trials are one good design among several. The dots rate the design, not the TA study shown. A small pilot can't do what a large, well-run trial can.

TA has studies of every kind. What it has fewest of are the large randomised trials that health systems tend to ask for.

What the Main Reviews Found

Ohlsson, 2010: a list of everything

Thomas Ohlsson set out to gather all the TA research he could find for the first issue of the International Journal of Transactional Analysis Research (Ohlsson, 2010). His "Big List" held 326 studies from 1963 to 2010. Most came from the 1970s and early 1980s, and much of that early work was doctoral research at universities in the United States. Later studies were fewer, and most were done in Europe.

From the list he picked 88 studies that looked relevant to whether TA psychotherapy works. Fourteen turned out not to be relevant (some used the phrase "transactional analysis" for something else), five were duplicate reports, and he couldn't find nine of them. That left 60 distinct outcome studies. By his own rating, often from abstracts alone, 50 showed positive effects, 10 showed no significant change, and he rated none as harmful. (He counted the one study with a negative finding, discussed below, as positive overall because TA patients still improved.)

Ohlsson was frank about his method. He wrote that his review should not count as scientific research in its own right, and that his comments on each study came from his own limited reading, not a systematic appraisal. The paper is best read as a map of what exists.

He also gave a useful warning about reading research. One study he examines closely, a 1981 randomised trial of an inpatient alcohol programme, had been summarised in an earlier report as showing that TA did worst on every measure. On his reading of the full report, TA was a small add-on to a programme everyone received, the TA group did drink more at one follow-up point, there was no difference at the longest follow-up, and every group improved. The group that had both the behavioural programme and TA ended with the lowest alcohol intake.

Novey, 2002: asking 932 clients

Theodore Novey's international study surveyed 932 clients of 27 certified transactional analysts, in four language groups (Novey, 2002). He compared the results with a well-known 1995 survey by Consumer Reports, a US consumer magazine, and its groups of clients seen by psychiatrists, psychologists, social workers, counsellors and doctors. Clients whose therapy lasted more than six months reported much better results than those in shorter therapy, and the TA clients' results were significantly higher than any of the comparison groups.

The design has limits that are plain from the summary: the outcomes are clients' own ratings, and the comparison groups come from a different survey. The study shows that many TA clients rated their therapy highly, which is worth knowing, but it can't show that TA beats other therapies.

Vos and van Rijn, 2021: a meta-analysis of 75 studies

The most ambitious review is by Joel Vos and Biljana van Rijn (Vos & van Rijn, 2021). They start from a frank point: before their work, there had been no systematic research testing the evidence for TA's effectiveness. They built a model of TA from what TA psychotherapists around the world say they do in practice, then checked each part against the research.

Their meta-analysis of 75 studies found moderate to large positive effects on psychological symptoms, self-efficacy, social functioning and ego states. They link these results to four therapist skills: building a good working relationship, working with what is happening in the present, analysing where problems come from (through life scripts and injunctions), and giving therapy a clear structure through contracts, stages and teaching the client TA ideas. They conclude that TA can be considered a "bona fide and evidence-based treatment" for a wide range of clients. The abstract doesn't say whether these effects compare TA with a control group or measure change from before to after therapy, and that difference matters. We could read only the published abstract of this paper, so we don't go further into which studies it included or how it weighed them.

Individual Studies Worth Knowing

TA in UK GP surgeries

Between 2008 and 2010, a London training institute and the local health service offered 12-session TA and integrative therapy for anxiety and depression in family doctors' clinics (GP surgeries) (van Rijn, Wild & Moran, 2011). The study copied the evaluation method the UK's National Health Service (NHS) used for its CBT services. Seventy-eight clients took part, randomly assigned to TA or integrative therapy, and the therapists were senior trainees. Symptoms fell with both approaches, with no significant difference between them, and by amounts comparable to published NHS figures for CBT. There was no untreated control group. Because of unplanned endings, only 55-57% of clients completed the two longer questionnaires at both the start and the end, though the shorter end-of-therapy measures were nearly complete.

Group and individual TA

A later study from two UK public clinics compared TA group therapy with TA individual therapy for depression and anxiety (van Rijn & Wild, 2016). Both showed large drops in symptoms, with no significant difference between the formats. The authors describe it as a naturalistic study with no control group and limited randomisation, and they say a larger controlled trial is needed.

Four detailed case studies of depression

Mark Widdowson studied four clients through short-term TA therapy using a method called Hermeneutic Single-Case Efficacy Design. Each case was argued both ways, for and against the therapy being the cause of change, and then given to independent judges. In the first case, a young man with severe depression, the judges concluded that he had changed considerably to substantially, and that the change was substantially due to therapy (Widdowson, 2012). The fourth case was mixed: a woman with mild depression improved, but the gains didn't hold when hard events piled up after therapy ended (Widdowson, 2013). Widdowson lists his own limits, including that he was the therapist, helped build the arguments and had taught the analysis team, and that the judges were all therapists.

Inpatient TA for personality disorders

A Dutch study published in the Journal of Personality Disorders compared 67 patients in a three-month inpatient TA programme with 67 matched patients in other psychotherapies (Horn et al., 2015). Both groups improved a great deal. The TA group improved more, and at three years 68% were symptomatically recovered against 48% in the comparison group. Patients were matched, not randomised.

A small randomised pilot

Vos and van Rijn later tested a 16-session brief TA therapy for mild to moderate depression. People were assigned by block randomisation to brief TA, brief CBT or usual care (other relational and humanistic therapies), and the usual-care group was matched to the others (Vos & van Rijn, 2025). Twenty-eight people completed the TA therapy and ten completed CBT. The TA group did better than usual care and about as well as CBT. The authors call it a pilot and say further validation is needed. The reported effects are unusually large, another reason to wait for that validation. With only ten people in the CBT arm, it can't settle how the two compare.

Who Wrote the Reviews, and Why It Matters

The main reviews of TA's evidence come from inside TA. Ohlsson's paper describes him as a Teaching and Supervising Transactional Analyst (Psychotherapy) and a researcher attached to the psychology department at Lund University in Sweden. The 2011 primary care paper describes Biljana van Rijn as a Teaching and Supervising Transactional Analyst (Psychotherapy) and Head of Clinical Studies at Metanoia Institute, and describes Metanoia as a TA psychotherapy training institute that also trains in integrative and other humanistic approaches. The 2025 pilot lists both its authors, van Rijn and Joel Vos, at Metanoia. Novey's study measured clients of TA practitioners. Most of these papers appeared in TA's own journals: the Transactional Analysis Journal and the International Journal of Transactional Analysis Research. Several of the recent studies share the same authors.

This matters for a reason psychotherapy researchers have a name for: researcher allegiance. People who believe in a method can design, run and interpret studies in ways that favour it, usually without meaning to. Widdowson, also a Teaching and Supervising Transactional Analyst, was the therapist in his own cases. His case studies name the risk directly and use independent judges to guard against it (Widdowson, 2012). Ohlsson's reading of the 1981 alcohol study shows it can cut the other way too: there, the authors ran the behavioural programme being compared, and the TA work was done by a single outside therapist.

A 2007 report prepared for the Berne Institute was blunter. Khalil and colleagues concluded that the TA evidence base was "insular and not subjected to the quality reviews of the wider academic and health communities" (quoted in Ohlsson, 2010). The same report concluded that TA is probably about as effective as other therapies, but that its evidence was scant and of fairly poor quality. Ohlsson thought these conclusions were too harsh, and he pointed out that their review was as loosely built as his own.

None of this suggests the findings are wrong or that anyone has been dishonest. People who practise a therapy often have the strongest reason and the easiest access to study it. The methods are published and can be checked, and some TA studies have appeared in general journals, such as the Dutch inpatient study, by a large team that included university researchers (Erasmus MC, Amsterdam, Leiden), and the 2025 pilot, by the Metanoia team. The fair conclusion is that TA's evidence would be stronger if more of it came from researchers with no stake in the result.

The Fair Criticisms, and TA's Replies

Few randomised trials

This is the biggest gap. TA has very few randomised trials, and the newest one is a small pilot. Ohlsson's reply is that RCTs are one useful design among several, not the only route to valid knowledge, and that the right design depends on the question. He points to Michael Rawlins's 2008 Harveian Oration to the Royal College of Physicians, which questioned the same idea for medicine. The argument has weight, but it still leaves TA short of the kind of evidence health systems tend to ask for. In 2011, van Rijn and colleagues noted that UK clinical guidelines named CBT as the main treatment for anxiety and depression and did not recommend approaches such as TA. Ohlsson describes TA being refused official recognition in Sweden in the early 1990s because of a lack of research.

Small samples and no control groups

Many TA studies follow tens of clients, not hundreds, and the practice-based ones have no untreated or waiting-list comparison group. The reply is that practice-based studies show how therapy works in real clinics with ordinary clients, which tightly controlled trials can miss. Small uncontrolled studies are a reasonable start, but on their own they can't show that TA caused the change.

Self-reported outcomes

Most outcomes are questionnaires the client fills in. In the UK primary care study they were the same standard questionnaires the NHS programme used to evaluate CBT, such as the PHQ-9 for depression, so TA wasn't held to a lower bar. Self-report can still be coloured by hope, politeness or wanting to please, more so when the person asking is your own therapist or clinic.

An older research base

The busiest period of TA research was the 1970s and early 1980s. Ohlsson argues none of that work is out of date simply because it is old. Still, a reader today would want more recent studies alongside them.

A pop-psychology reputation

Games People Play (1964) was written for professionals and became a hugely popular self-help book, selling over two million copies (UCSF Archives). Thomas Harris's I'm OK, You're OK (1969) was another TA book written for general readers. Catchy terms are easy to use loosely, and that can make TA look shallow. How a theory became popular doesn't tell you whether the therapy helps, but it explains why some people meet TA as self-help before they meet it as psychotherapy. You can read more about how Berne built TA, and about his books, in What is transactional analysis?.

Concepts that are hard to test

Ideas like the life script or a psychological game describe patterns over years, which are hard to measure. Parts of TA have been made measurable. The 2021 review describes the functional model of ego states in terms of five behavioural functions (Controlling Parent, Nurturing Parent, Adult, Adapted Child and Free Child) that studies can measure, the same functions an egogram sketches. Other parts remain closer to useful maps than to tested claims.

How TA Compares With Other Therapies

Where TA has been compared directly with something else, the results go both ways. In the UK primary care study, clients randomly assigned to TA or to integrative therapy did about equally well, and both groups' results were close to published NHS figures for CBT. The 2025 pilot found TA about as good as brief CBT and better than usual care (other relational and humanistic therapies), the Dutch inpatient study found TA did better than other psychotherapies, and in the 1981 alcohol study the TA add-on group did worse at one point. None of these studies is big enough to rank TA above or below other approaches.

A wider body of research helps make sense of this. Reviews of many studies across many therapies find that shared ingredients, such as the working alliance, empathy and the client's expectations, matter a lot for whether therapy helps (Wampold, 2015). An American Psychological Association task force concluded that the therapy relationship contributes to outcome substantially and consistently, whatever the type of treatment (Norcross & Lambert, 2018). The 2021 review includes the client-therapist relationship among its four skills, though in the small 2011 UK study the working alliance grew over therapy without being directly linked to outcome.

If You Are Choosing a Therapist

The research suggests a few practical things.

  • Fit matters. Research across therapies finds that the working relationship makes a real difference to outcome, so notice how you get on with a therapist and whether you feel understood. A first conversation is a fair way to judge it.
  • You can ask about training and approach. Any therapist should be happy to tell you where they trained, how they work, whether they have regular supervision, and how they would decide with you whether therapy is helping. TA has its own training route, set out by the international TA associations, so a TA therapist can explain where they are on it. Our guide on how to choose a therapist in India has more questions to ask.
  • Look for a clear goal. TA works through an agreed contract, which gives you and your therapist a way to check progress. If you can't tell after a while whether anything is changing, say so. The post on how psychotherapy works covers what progress can look like.
  • For depression, think about more than one kind of help. The TA depression studies are small. For depression that is severe or has lasted a long time, see a psychiatrist as well, because medication can help and can run alongside therapy. Our guide to online therapy for depression explains the options.

If you are thinking about ending your life, call Tele-MANAS on 14416 or 1-800-891-4416 (free, 24/7) or go to the nearest emergency department.

Frequently Asked Questions

Is transactional analysis evidence-based?

Partly. Reviews by TA researchers report mostly positive outcomes, and a 2021 meta-analysis of 75 studies found moderate to large effects. But TA has few randomised controlled trials, and the main reviews and most recent studies come from TA practitioners and appear in TA journals. It meets a loose standard of "evidence-based" better than a strict one.

Is transactional analysis a pseudoscience?

It doesn't fit the usual meaning of the word. TA makes claims that can be studied, and researchers have studied it since the 1960s, with published methods and results. Some of its concepts, like life scripts, are hard to test directly, and its popular books gave it a self-help image. Those are good reasons to ask for better research.

Does TA therapy work for depression?

The studies that exist are encouraging but small: practice-based studies in UK clinics, detailed single-case studies, and a pilot trial in which brief TA did about as well as brief CBT. For severe or long-lasting depression, see a psychiatrist as well as a therapist.

Is TA better than CBT?

There isn't enough research to say. Only one small trial has compared them directly: brief TA did about as well as brief CBT, but only ten people completed the CBT. A UK clinic study found TA results similar to published NHS figures for CBT. A 2011 UK study noted that national clinical guidelines named CBT, not TA, as the main treatment for anxiety and depression. Research across many therapies suggests the relationship with your therapist matters a great deal, whichever method they use.

Why is transactional analysis criticised?

The main criticisms are that its research has few randomised trials and small samples, relies on questionnaires clients fill in, and that the main reviews and most recent studies come from TA practitioners and appear in TA journals. Some people also see it as dated pop psychology because its best-known book, Games People Play (1964), became a bestselling self-help book. TA researchers reply that practice-based research has value of its own and that randomised trials are not the only valid design.

If you want to talk through whether TA suits what you're dealing with, a free 15-minute intro call is a good place to ask. The page on Transactional Analysis therapy describes what the work involves.

Sources

How we write and check our sources

Related writings

Talk to a therapist about this

Transactional Analysis is the main way Dandelion works with individuals. Here is what it looks like in practice.

Transactional Analysis therapy

You can also look at online individual therapy.

Ready to do your own work?

Start with a free 15-minute introductory call.

Book a Free Intro Call