What works in therapy? >  Getting better outcomes 

Essential research findings and getting better outcomes


Is therapy effective, and if it is, what makes it so?


Are some therapies more effective than others? What about therapists?


How can the outcomes of therapy be enhanced?


These are questions which are critical to clients and therapists alike, and to which research provides many answers. Many therapists, however, seem either unaware or uninterested in what research can bring to their practice. A study of American psychotherapists, for example, found that only 4% cited research as the most useful resource which informed their practice.


The picture from more than 50 years of therapy outcome research can be summarised as follows:
The following sections explore these areas in more detail and summarise key research findings. The table of contents below outlines the structure and clicking on the links will take you to the relevant section. At the bottom of this page is a list of the key references and a glossary of highlighted terms.


For clients, I hope that the sections may help you to make more informed choices about what to look for from therapy and therapists. For therapists and service managers, my hope is that they provide motivation to explore further what this way of working can bring to your practice. 



Therapy is effective 


Studies of the efficacy of therapy, which are based on tightly controlled trials, show its effect to be considerable. Effect size in social science research is often expressed as Cohen's d. The larger the value of Cohen's d, the larger the effect:


        Small effect :          d = 0.2

        Medium effect :       d = 0.5

        Large effect :          d = 0.8


In one of the most widely respected publications on the subject, counselling psychologist and former statistician Bruce Wampold (2001) summarised the key findings from the meta-analytic literature. He concludes: 

"From the various meta-analyses conducted over the years, the effect size related to absolute efficacy appears to fall within the range .75 to .85. A reasonable and defensible point estimate for the efficacy of psychotherapy would be .80...."  

In other words therapy has a large effect, such that of ten people receiving therapy, the evidence predicts that eight will show an improvement in their psychological wellbeing. 


Effectiveness refers to the difference an intervention makes in routine or 'clinically representative' settings. Studies in these settings have shown levels of effectiveness at least comparable to those found in experimental research designs. These include:


Stiles et al (2007) examined the outcomes of 5,613 clients in primary care. Using the model of Reliable and Clinically Significant Change developed by Jacobson and Truax (1991), 58.3% met the criteria for 'recovery' i.e. their improvement was both reliable and clinically significant. A further 19.4% met the criteria for reliable improvement. In total then, 78% of clients showed either 'recovery or improvement.' The overall treatment effect size was calculated to be 1.39 - a large effect.


Mullin et al (2006) explored the outcomes of 11,953 clients attending therapy in UK primary care. 54% of clients met the criteria for recovery and a further 18% met the criteria for reliable improvement. In all, 72% showed either recovery or improvement. While the effect size was not shown in the published study, using the same method of calculation used by Stiles et al (2007) gives an effect size of 1.43.


Shadish et al (2000) in a meta-analysis of some 90 studies, concluded:

['These results suggest that] psychological therapies are robustly 
effective across conditions that range from research oriented to 
clinically representative....Previous findings that clinical 
representativeness leads to lower effect sizes are probably an 
artifact of other confounding variables.....'
We can confidently say, then, that both in research and routine practice settings that 'therapy works.'                    
                                                                                              back to top


Therapies are broadly equivalent in their outcomes


The question of whether some therapeutic approaches yield outcomes that are superior to others has, for decades, been one of the most hotly contested areas of therapy research. Comparative outcome studies have yielded evidence both for and against the case that some approaches are superior.


As early as 1936, Rosenweig proposed that common factors across therapies were responsible for their apparent efficacy and therefore that therapies would be broadly equivalent in their outcomes. This proposition has become known as the 'Dodo bird effect,' referring to the end of the race in Alice in Wonderland: "At last the Dodo said, 'Everybody has won and all must have prizes.'" (Rosenzweig, 1936, p412)


Comparative outcome research is a complex area, not least because of the many variables (e.g. therapy type, clients characteristics, study design etc.) that impact on outcomes, and which may confuse or 'confound' findings. It seems that many of the earlier research studies comparing two or more therapies were flawed because they failed to control for key variables and thereby inflated the outcomes for one approach. Two critical variables to which this has often applied are therapist allegiance to the therapy delivered, and the skill or competence of the therapist (both explored below).


More recently, as research has become more sophisticated and meta-analysis has been applied to this question, there has emerged a robust body of evidence that bona fide therapies are broadly equivalent in their outcomes. One of the most comprehensive meta-analyses was that conducted by Wampold and colleagues (Wampold et al, 1997), which concluded that the effect size for different treatments was about 0.2 (i.e. a small effect), that would account for no more than 1% of the variance in outcomes. This finding was echoed by Lester Luborsky and colleagues in 2002 (Luborsky et al, 2002), and in practice would mean that 42% of the clients in the 'inferior' treatment would actually do better than the average client in the 'superior' treatment (Cooper, 2008).   


Controlling for the effects of therapist allegiance and therapist performance as variables serves to reduce further the differences between therapies. Luborsky et al (2002) propose an overall effect size of 0.14, while Wampold (2001) proposes 0.0 - 0.2. meaning that type of therapy, at most, accounts for no more than 1% of the variance between therapies.


Similar results have been found in routine practice, for example the studies of therapy in primary care by Stiles et al (2006, 2007), which found that there was no statistically significant difference between person centred, psychodynamic and cognitive behavioural therapies, and also when each of these was delivered alongside one other type of therapy. 


These and other findings, therefore, offer strong support for the Dodo bird proposition of uniform efficacy across therapies. They are not without their critics, however, and it may be the case that while a wide range of interventions are more or less equally efficacious for depression or generalised anxiety disorder, for example, there may be other problems or conditions for which specific interventions are more suited.  
                                                                                            back to top


Specific ingredients or techniques of therapy


Therapeutic models vary widely in their underlying assumptions of human development and change, and in the specific 'ingredients' or techniques employed to bring about change. Alongside these specific ingredients, therapies also share common factors - those which are common to all therapies - for example, the therapist's belief in the efficacy of their model; the quality of the alliance between therapist and client; placebo effects; and the degree of expertise of the therapist.    


What is it that contributes, or contributes most, to the outcomes of therapy? Is it the therapy and its specific ingredients, or the common factors that apply across all? If indeed therapies are uniformly equivalent in their outcomes, this would suggest that, either, the specific ingredients of each are of more or less equal potency, or, that common factors play a larger part in therapy outcomes than specific ingredients.


Asay and Lambert (1999) attempted to attribute the improvements from therapy as a function of four variables, concluding that therapist model and technique factors account for only 15% of the variance outcomes, whereas the non-specific factors - client factors and extra-therapeutic events, therapeutic relationship and expectancy/hope account for 85% (Figure 1).  These were estimates of limited statistical rigour, an issue subsequently addressed by Wampold (2001).  


The research within this area is complex, but some of the key findings from Wampold's (2001) examination of the literature are presented below: 

  • In studies comparing efficacy of a 'complete' treatment to treatments where one or more ingredients believed to be necessary are removed (for example addressing 'dysfunctional' thinking styles in CBT), there is no significant difference in outcomes (e.g. Jacobson et al, 1996), or evidence that in some cases the treatment without the ingredient is more effective (e.g. Ahn and Wampold, 2001)
  • In a meta-analysis of 43 various treatments for depression, cognitive therapy was no more effective in terms of beneficial impact on clients' cognitive (i.e. thinking) processes than non-cognitive therapies (Oei and Free, 1995)
  • A review of eight studies of CBT for depression found that whereas between 60 - 80% of the reduction in depression severity occurred by the fourth session, the specific techniques designed to bring about a reduction in depressive thoughts typically are not introduced until the fourth therapy session." (Ilardi and Craighead, 1994)  
In reviewing the relative contributions of the range of factors that account for the variance in therapy outcomes Wampold (2001) estimates that at least 70% are attributable to general factors incidental to the specific ingredients of therapy, that the specific effects attributable to therapy models account for no more than 8% at most, while the remaining 22% is not fully explained but likely to be due to client factors. (Figure 2).
Wampold concludes with the following summary:
"The history of psychological treatments is littered with examples of treatments that are beneficial to clients but whose psychological explanation for the benefits have failed to be verified......the ingredients of the most conspicuous treatment on the landscape, cognitive-behavioural treatment, are apparently not responsible for the benefits of this treatment." 
                                                                                     back to top

Contextual or common factors


From the evidence reviewed above it is clear the contribution to therapy outcomes which is attributable to factors which are incidental to the specific effects far outweighs that of model and technique factors. Of the non-specific factors estimated by Asay and Lambert (1999) to account for 85% of variance in therapy outcomes, 15% were estimated to be attributable to expectancy or hope factors, 30% to the therapeutic relationship and 40% to client variables and events external to the therapy, e.g. life events (Figure 1).


Wampold (2001) estimates that at least 92% of variance in outcomes is attributable to factors which are not specific the therapy models, of which 70% are common factors that include working alliance, therapist allegiance to the therapy provided, therapist effectiveness, and placebo effects (Figure 2).


It is probably impossible to arrive at precise estimates of the relative contributions to therapy outcomes made by the multitude of contributory factors, not least because of the difficulty of arriving at a common set of definitions and statistical procedures for doing so. Nonetheless there is strong evidence for the contribution of a range of common factors to outcomes, and some of these are explored in more detail below.


Allegiance effects

Allegiance refers to the degree to which the therapist delivering the treatment believes that treatment to be efficacious (Wampold, 2001). It is of particular interest because of its impact on studies comparing the outcomes of two or more different therapies. In essence, there has been strong tendency for researchers to find results that support their own beliefs, expectations or preferences (Cooper, 2008). In other words, to find in favour of their preferred model of therapy.


In a review of 29 studies comparing psychotherapies, Luborsky et al (1999) estimated that the allegiance of the researchers accounted for over two-thirds of the variance in outcomes between therapies. Reviewing the literature Wampold (2001) comes to a similar conclusion i.e. that allegiance effects have generally been found in meta-analyses investigating allegiance, with an upper effect size of 0.65. He concludes by saying:
"That the effects due to allegiance account for dramatically more of the variance in outcomes than does the particular type of treatment implies that therapist attitudes towards therapy is a critical component of effective therapy..."

A range of reasons would appear to produce allegiance affects in comparative studies of therapies that include:

  1. The alternative treatments used as comparison against the experimental treatment may not be bona fide therapies - in other words they are not therapies that are designed to be therapeutic in the same sense e.g. supportive 'counselling'
  2. Both experimental and comparative treatments may be delivered by practitioners whose primary allegiance is to the experimental treatment, and who may not have been effectively trained in delivery of the comparative treatment. Given that the therapist's belief in the efficacy of the therapy being delivered is a critical component in its actual efficacy, it is hardly surprising in these conditions if the comparative treatment produces inferior outcomes
  3. Within many research designs there has been a failure to control for factors that may inflate the apparent effectiveness of the experimental treatment - for example, using measures that may be particularly sensitive to changes in cognitive processes, or failure to consider therapist performance as a variable that impacts on outcomes

Therapeutic alliance and relationship factors

It would seem self evident that if clients are to make effective use of the deeply personal process that is therapy, then a strong relationship between client and therapist is essential. The question is, to what extent are the therapeutic alliance and other relationship factors critical to effective therapy?


The term 'therapeutic alliance' has been used to encompass a range of factors (e.g. Cooper, 2008 and Wampold, 2001) including:

  1. Client and therapist agreement on the goals and tasks of therapy
  2. The client's motivation and ability to accomplish work collaboratively with the therapist
  3. The therapist's empathic responding to and involvement with the client
  4. A positive affective bond between client and therapist, characterised, for example, by mutual trust, acceptance and confidence

A range of meta-analyses show a strong positive correlation between ratings of the therapeutic alliance and outcome, e.g. Horvath and Symonds (1991), Martin et al (2002), Horvath and Bedi (2002). The correlations are in the range of 0.21 - 0.26, which, converted to effect sizes, give effect sizes in the range of d = 0.45 - 0.54. This is a medium size effect that would mean that between 5 - 7% of outcome is associated with alliance - clearly very significantly greater than the  1% contributed by therapy type, as outlined above.


Further findings related to alliance include:

  1. The alliance "appears to be a necessary aspect of therapy, regardless of the nature of therapy" (Wampold, 2001)
  2. In some studies, early ratings of the alliance are highly correlated with outcome (e.g. Krupnick et al, 1996, Horvath and Symonds (1991), and clients who rate the alliance highly early in therapy are less likely to drop out of therapy and have positive outcomes (Horvath and Bedi, 2002, Blatt et al, 1996)
  3. Client ratings of the alliance tend to be more predictive of outcomes than those of therapists (Horvath and Bedi, 2002)
The importance of the therapeutic relationship and the alliance, as part of it, has been acknowledged by American Psychological Association (APA). The APA Division of Psychotherapy established a Task Force on Empirically Supported Therapy Relationships to review research in this area. Its conclusions (Norcross, 2002) identified 11 factors (see panel below) related to therapeutic relationship that were determined, based on available evidence, to be either 'demonstrably effective', or 'promising and probably effective'.

American Psychological Association Division of Psychotherapy Task Force on Empirically Supported Therapy (Norcross, 2002)


Demonstrably effective factors

Goal consensus and collaboration

Therapeutic alliance


Cohesion (in group therapy)


Promising and probably effective factors

Management of countertransference


Positive regard


Self disclosure

Relational interpretations

Repair of alliance ruptures


Two factors determined by the APA to be demonstrably effective - Goal consensus and collaboration and Empathy - have not been covered so far and are considered briefly below. Both areas are considered by Cooper (2008). In reviewing the evidence for goal consensus and collaboration, Cooper concludes that the evidence for goal consensus is "mixed" but cites the work of Tryon and Winograd (2002), which finds that goal consensus and outcomes are significantly and positively correlated on at least one measure in around 65% of studies reviewed. 


From the same study (Tryon and Winograd, 2002) the findings for the relationship between collaboration (i.e. the active collaboration of client and therapist in the process of therapy) and outcomes are much stronger, with a positive correlation found in around 90% of studies.


Empathy refers to "entering the private perceptual world of the client and having an accurate, felt understanding of their experiencing" (Cooper, 2008), and is found to be one of the most important relational factors that impact on therapy outcomes. On the basis of 47 studies measuring empathy and outcomes, Bohart et al (2002) conclude that the correlation between the two is about 0.32. This is equivalent to an effect size of 0.68, bigger even than that estimated for the therapeutic alliance. Interestingly, the same review also finds that the correlation between empathy and outcomes is significantly stronger in more cognitive-behavioural therapies than in experiential/humanistic therapies (effect sizes of 1.12 and 0.52 respectively).
                                                                                         back to top

Variations across therapists and services


One of the failings of much of the comparative research between competing therapies has been its failure to acknowledge the therapist as a factor contributing to outcomes, and to control for therapist differences when comparing outcomes for different therapies.


Designing experimental studies which can properly identify therapist effects is complex, but a range of studies now exists which highlight therapist skill or competence as a major factor in therapy outcomes. These include both experimental and naturalistic (i.e. routine practice) studies, of which some of the more significant are outlined here.


Crits-Cristoph et al (1991), from data from 15 previous studies covering 27 different treatments, estimated that nearly 9% of the variance in outcome was due to therapists - equivalent to a medium-large effect size of 0.60. In addition, for one dependent variable tested, over 70% of the variance was due to the therapists.


Luborsky et al (1997) reviewed data from seven samples of clients with issues of depression/addiction. Across 22 therapists in total, the percentages for client improvement ranged from slight deterioration to more than 80% improvement.


On the basis of this and other evidence Wampold (2001) concludes that therapist effects account for between 6 - 9% of variance in outcomes, making it a very much larger factor than the individual therapy or the specific ingredients of therapy.


Evidence of very considerable differences in therapist effectiveness also comes from naturalistic studies. The study by Okiishi et el (2003) compared the outcomes of 1,841 clients and 91 therapists of a university counselling service. In all, 56 therapists had more more than 15 clients in the dataset. Figure 3 shows the improvement 'slopes' (i.e. average rates of change per session) for the top and bottom three therapists. Those whose clients showed the fastest rate of improvement (Therapists 1 - 3) had an average rate of change 10 times greater than that of the centre average. By contrast, clients of the therapists with the slowest rates of improvement (Therapists 54 - 56) showed an average increase in symptoms - in other words they deteriorated.
 Figure 3. Improvement rates of top and bottom performing therapists
(Okiishi et al, 2003)

These findings are echoed in data from the CORE National Database for primary care counselling, containing outcome data for nearly 12,000 clients, which indicate a wide range of improvement rates across individual therapists. Taking into account only those therapists with 10 or more clients in the dataset, the range is between 11-94 per cent (McInnes, 2007).


It would be wrong to suggest that therapists, in terms of their outcomes, are evenly distributed along the range of improvement rates. It is more likely that the each end of the range is composed of outlyers i.e. a small proportion of therapists who perform either exceptionally well or badly, with the majority clustering around the mean level of improvement. Nonetheless the very fact that there are such differences should provide a stimulus for reflection on the part of all therapists.


If there are differences between therapists in terms of their effectiveness, is there evidence to suggest that these may be attributable to particular therapist characteristics? Cooper (2008) reviewed evidence across a range of personal and professional characteristics. A brief summary of those characteristics that appear to be significantly and positively associated with outcomes, as well as other key findings, is presented below:

  1. There appears to be a significant association between therapists' own levels of psychological health and outcome, in one study equivalent to an effect size of 0.24 (Beutler et al, 2004)
  2. Clients that hold strong beliefs or values, including religious beliefs, tend to express a preference for therapists with similar values, but there is little evidence that they achieve better outcomes
  3. There is some evidence that female therapists may achieve slightly better outcomes than male therapists, particularly with female clients, although some of the evidence is less conclusive
  4. There is evidence that clients from marginalised social groups show some preference for working with therapists from similar groupings, and limited evidence that matching may lead to better outcomes, but in many cases, that the therapist's attitudes and values may be of greater importance
  5. Professional characteristics such as training, supervision and length of experience have some bearing on outcomes, although effect sizes are small. Additionally, although the evidence is somewhat contradictory, there is some evidence that paraprofessionals may, in some circumstances, achieve outcomes as good as their professionally trained counterparts
In general, it seems that the effectiveness of therapists is related less to their personal and professional characteristics than to how they are able to relate to their clients, and their level of skill as therapists. As has been previously demonstrated, factors such as goal setting and collaborative working, an empathic stance, and the creation and maintenance of a strong therapeutic bond, may be factors that more effectively account for the differences between therapists and their outcomes.  
                                                                                                                back to top

Early predictors of successful outcomes


Two important factors have been shown to be good early indicators of the likelihood of a successful outcome to therapy, namely, signs of improvement early on in the work, and the client's rating of the therapeutic alliance. 


Research shows that it is possible to predict a general 'trajectory of change' in successful therapy, and also that the majority of the improvement that takes place is likely to happen in the early stages. (e.g. Whipple et al, 2003, Howard et al, 1996). Howard et al (1996) found that up to 40% of clients improve in the first three sessions, rising to up to 65% within seven sessions, 75% in around six months and 85% within 12 months (Figure 4).
The principle of diminishing returns with effort applies, so that the degree of improvement between sessions 53 - 104 is more or less the same as between sessions 2 - 4.
Figure 4. Illustration of the 'dose-effect' relationship showing the majority of improvement happens early in therapy (Howard et al, 1996)

The study by Howard (1996) also found that an absence of an early improvement in the client's self-reported wellbeing significantly decreased the likelihood of a subsequent improvement in symptom relief and functioning. In a large study of outcomes in US managed care, Brown et al (1999) found that clients who showed no improvement by the third therapy session did not on average achieve improvement over the entire course of therapy, and further that those who showed deterioration by the third session where twice as likely to drop out than those making progress. In other words, if improvement is going to take place at all, there are likely to be signs of improvement early on in the process, and early deterioration is a potential indicator of drop out.


The importance of the contribution to outcomes made by the alliance has already been acknowledged above. Interestingly, clients' ratings of the therapeutic alliance have been shown to be more accurate predictors of outcome than therapists' ratings (e.g. Horvath and Bedi, 2002, Connors  G.J., & Carroll, K.M. 1997). The value of measuring the alliance in the early stages of therapy is illustrated by the study by Blatt et al (1996) of the data from the National Institute for Mental Health Treatment of Depression Research Collaborative (NIMH TDCRP), possibly the most conclusive and well constructed trial in the history of comparative therapy research. The alliance was found to be significantly related to drop out from treatment, with clients that reported a positive alliance early in treatment less likely to drop out, and more likely to experience reduced levels of depression.


As will be seen in the following section, therapists sometimes struggle to identify which clients will, and will not, benefit from therapy, and may have rather different views from clients on the quality of the alliance. It follows, then, that the use of an alliance measure as a routine part of the therapy process may help practitioners to identify and address problems in the alliance which may not be otherwise evident to them.

                                                                                                                     back to top


The case for monitoring progress and alliance during therapy


This section presents the case for routinely monitoring, during the course of therapy, progress towards a successful outcome and the quality of the therapeutic alliance. This is based on:

  • Evidence that the client's early rating of the therapeutic alliance, and early improvement, are both predictive of a successful outcome to therapy
  • Therapists tend to over-estimate their effectiveness, and fail to recognise clients at risk of deterioration (see below)
  • There is significant variation in effectiveness, as well as rates of drop out, across therapists and within services
  • As will be shown below, feedback from outcome and alliance measures used during therapy has been shown to significantly reduce attrition and improve client outcomes

Despite evidence that therapy is highly effective, we need to recognise that drop out or 'attrition' plagues therapy in both research studies and routine practice. Clients who drop out may have got what they needed from therapy and forgotten to tell us, but more likely they have not and not been able to tell us. A meta-analysis of 125 studies by Wierzbicki & Pekarik (1993) revealed a mean drop out rate of 49%. A more recent study of drop out in routine primary care found a range of drop out rates across services between 4% - 65%, with a mean of 32.7% (Connell et al, 2006).


A parallel concern is the fact that therapists appear to struggle to identify, early in therapy, which clients will benefit and which will not. Hannan et al (2005) used data from over 11,000 clients to generate a linear model for expected treatment response. Based on this model the researchers devised a test to try to predict early in therapy which clients might be at risk of 'treatment failure' (i.e. a reliable deterioration in their levels of distress). 

The reliability of the test in predicting treatment failure was compared with that of the centre's therapists, based solely on their clinical judgement. Therapists rarely predicted deteriorationOf 550 clients attending at least one session, only 3 were predicted by therapists to deteriorate. Outcome data, however, showed that 40 clients deteriorated by the end of therapy, and only one those predicted to deteriorate actually had. The test somewhat over-predicted potential treatment failure, but proved to be far more accurate than the therapists' judgements.


A significant body of research now demonstrates that using outcome and alliance measures during therapy can support practitioners' clinical judgement by helping to identify clients that may be at risk of dropping out, or simply not improving as we might hope or expect. In this way we can consider alternative strategies, or even alternative therapists. 


Lambert et al (2005) summarised four large scale studies evaluating the effects of providing therapists (and sometimes patients) with feedback about a client's improvement, using progress graphs and warnings for clients who were not showing the expected treatment responses. The four feedback conditions were:

  1. TAU - therapists received no feedback on clients that were not 'on track'.
  2. T-Fb - therapists received feedback on those clients that were not on track and at potential risk of treatment failure
  3. T-Fb+CST - therapists received feedback for not on track clients plus the option of using a range of clinical support tools e.g. an alliance measure, measures of clients' social support
  4. T/P-Fb - both therapists and clients received feedback when clients were not on track and at potential risk of treatment failure
The impact on outcomes using each of these feedback conditions is very clearly evident from the Figure 5 below. Progressive increases in rates of clinical and/or reliable change were achieved from TAU (21%); T-Fb (34.9%); T-Fb+CST (49.1%) to T/P-Fb (56%), along with corresponding decreases in the proportions of clients that deteriorated. The highest rates of improvement were evident when progress feedback was given to both clients and therapists.
Figure 5. The effects of providing progress feedback to therapist, and
client and therapist (Lambert et al, 2005)
The researchers concluded: 

"It seems likely that therapists become more attentive to a patient when they receive a signal that the patient is not progressing. Evidence across studies suggests that therapists tend to keep 'not on track' cases in treatment for more sessions when they receive feedback, further reinforcing the notion that feedback increases interest and investment in a patient."

Similar results have been found in other studies. Whipple et al (2003) found that clients at risk of a negative outcome were less likely to deteriorate, more likely to stay in treatment longer, and twice as likely to achieve clinically significant change when their therapists had access to outcome and alliance information. Where clients were identified as making adequate progress, feedback decreased the number of therapy sessions without affecting final outcome. These findings suggest that outcome is improved and resources are more efficiently allocated when feedback on client progress is provided to therapists. 


A study by Miller et al (2006) explored the impact of introduction of two short measures of outcome and alliance into an international Employee Assistance Programme. During the early phase of the study, 20% of clients at intake had outcome measure but not alliance data. Those clients were three times less likely to return for a second session and had significantly poorer outcomes. Improving a poorly rated alliance early in therapy was correlated with significantly better outcomes by the end of therapy. Over a two-year period of incorporating  outcome and alliance data into routine practice the effect size of the services doubled (0.37 - 0.79).

                                                                                                                back to top


Ahn H, Wampold B E. (2001). Where Oh Where Are the Specific Ingredients? A Meta-Analysis of Component Studies in Counseling and Psychotherapy. Journal of Counseling Psychology, 48, 3, 251-257

Asay T P, Lambert M J. (1999). The Empirical Case for the Common Factors in Therapy: Quantitative Findings. In Hubble, Duncan, Miller (Eds), The Heart and Soul of Change: What works in therapy (23–55). Washington, D.C.: American Psychological Association Press


Beutler L E, Malik M, Alimohamed S, Harwood M T, Talebi H, Noble S, et al. (2004). Therapist Variables. In M. J. Lambert (Ed.), Bergin and Garfield's Handbook of Psychotherapy and Behaviour Change (5th ed, 227-306). Chicago: John Wiley and Sons


Blatt S J, Zuroff D C, Quinlan D M, & Pilkonis P. (1996). Interpersonal factors in brief treatment of depression: Further analyses of the NIMH Treatment of Depression Collaborative Research Program. Journal of Consulting and Clinical Psychology. 64, 162-171


Bohart A, Elliott R, Greenberg L, & Watson J. (2002). Empathy. In J. Norcross et al. (Ed.), Psychotherapy relationships that work. New York: Oxford University Press


Brown J, Dreis S, & Nace D K. (1999). What really makes a difference in psychotherapy outcome? Why does managed care want to know? In M.A. Hubble, B.L. Duncan, and S.D. Miller (Eds.). The heart and soul of change: What works in therapy (pp. 389-406). Washington, D.C.: American Psychological Association Press


Connell J, Grant S, Mullin S. (2006). Client initiated termination of therapy at NHS primary care counselling services. Counselling & Psychotherapy Research, 6, 60-67


Connors G J, DiClemente C C., Carroll K M, et al. (1997) The therapeutic alliance and its relationship to alcoholism treatment participation and outcome. Journal of Consulting and Clinical Psychology, 65(4), 588-598

Cooper M. (2008). Essential research findings in counselling and psychotherapy: the facts are friendly. Sage, London


Crits-Christoph P, Baranackie K, Kurcias J S, et al. (1991). Meta-analysis of therapist effects in psychotherapy outcome studies. Psychotherapy Research 1:81-91.


Hannan C, Lambert M J, Harmon C et al. (2006). A Lab Test and Algorithms for Identifying Clients at Risk for Treatment Failure. Journal of Clinical Psychology, Vol. 61(2), 155–163


Horvath A O & Symonds B D. (1991). Relation Between Working Alliance and Outcome in Psychotherapy: A Meta-Analysis. Journal of Counseling Psychology. 38, 2, 139-149

Horvath A O & Bedi R P. (2002). The alliance. In J. C. Norcross (Ed.), Psychotherapy relationships that work: Therapist contributions and responsiveness to patients (pp. 37-69). New York: Oxford University Press

Howard K I, Kopta S M, Krause M S, & Orlinsky D E. (1986). The dose–effect relationship in psychotherapy. American Psychologist, 41, 159-164

Ilardi S, Craighead W E. (1994). The Role of Nonspecific Factors in Cognitive-Behavior Therapy for Depression. Clinical Psychology: Science and Practice. 1, 2, 138 - 155


Jacobson N S, & Truax P. (1991). Clinical significance: A statistical approach to defining meaningful change in psychotherapy research. Journal of Consulting and Clinical Psychology, 59(1), 12-19.


Jacobson N S, Dobson K S, Truax P A, Addis M E, Koerner K, Gollan J K, Gortner E, & Prince S E. (1996). A component analysis of cognitive-behavioral treatment for depression. Journal of Consulting and Clinical Psychology 64 (2): 295–304.


Kazdin A E. (1986). Comparative outcome studies of psychotherapy: methodological issues and strategies. Journal of Consulting and Clinical Psychology, 54, 95 - 105.


Krupnick J L, Sotsky S M, et al. (1996) The role of therapeutic alliance in psychotherapy and pharmacotherapy outcome: findings in the National Institute of Mental Health Treatment of Depression Collaborative Research Program. Journal of Consulting and Clinical Psychology. 64, 532-539.


Luborsky L, Rosenthal R, Diguer L, Andrusyna T P, Berman J S, Levitt J T, Seligman D A, Krause E D. (2002). The Dodo bird verdict is alive and well - mostly. Clinical Psychology: Science and Practice 9: 2–12

Luborsky L, Diguer L, Seligman D A, Rosenthal R, Krause E D, Johnson S, Halperin G, Bishop M, Berman J S, Schweizer E. (1996) The Researcher's Own Therapy Allegiances: A “Wild Card” in Comparisons of Treatment Efficacy. Clinical Psychology, 6, 1 95–106 

Luborsky L, McClellan A T, Diguer L, Woody G, Seligman D A (1997) The psychotherapist matters: comparisons of outcomes across twenty-two therapists and seven patient samples. Clinical Psychology: Science and Practice, 4, 53–65

Lambert M J, Harmon C, Slade K, Whipple J L, Hawkins E J. (2005). Providing Feedback to Psychotherapists on Their Patients’ Progress: Clinical Results and Practice Suggestions. JCLP/In Session, Vol. 61(2), 165–174

Martin D J, Garske J P, and Davis M K. (2000) Relation of the Therapeutic Alliance With Outcome and Other Variables: A Meta-Analytic Review. Journal of Consulting and Clinical Psychology, 68, No. 3, 438-450

McInnes B. (2007). Workplace counselling: building an evidence base from practice. Counselling at Work. February 2007

Miller S D, Duncan D, Brown J, Sorrell R, Chalk M B. (2006). Using formal client feedback to improve retention and outcome: Making ongoing, real-time assessment feasible. Journal of Brief Therapy, 5(2), 5-22

Mullin T, Barkham M, Mothersole G, Bewick BM, Kinder A. (2006). Recovery and improvement benchmarks in routine primary care mental health settings. Counselling & Psychotherapy Research, 6, 68-80  


Norcross J C (Ed.) (2002). Psychotherapy Relationships that Work: Therapists Contributions and Responsiveness to Patients. New York: Oxford University Press.


Oei T P S, Free M L. (1995). Do cognitive behaviour therapies validate cognitive models of mood disorders? A review of the empirical evidence. International Journal of Psychology, 30, 145 - 179


Okiishi J, Lambert M J, Nielsen S L, et al. (2003). Waiting for supershrink: An empirical analysis of therapist effects. Clinical Psychology and Psychotherapy 10:361-73. 


Stiles W B, Barkham M, Mellor-Clark J, Connell J. (2007) Effectiveness of cognitive-behavioural, person-centred, and psychodynamic therapies in UK primary-care routine practice: replication in a larger sample. Psychological Medicine (2007), 38, 1–12.


Rosenzweig S. (1936). Some implicit common factors in diverse methods of psychotherapy. American Journal of Orthopsychiatry 6: 412–415.


Roth A, Fonagy P. (2005). What works for whom? A critical review of psychotherapy research. 2nd edition. Guildford Press, London


Shadish W R, Matt G E, Navarro A M, Phillips G. (2000). The effects of psychological therapies in clinically representative conditions; A meta analysis. Psychological Bulletin, 126, 512 - 529


Stiles W B, Barkham M, Twigg E, Mellor-Clark J, Cooper M. (2006). Effectiveness of cognitive-behavioural, person-centred and psychodynamic therapies as practised in UK National Health Service settings. Psychological Medicine, 36(4):555-66.


Stiles W B, Barkham M, Mellor-Clark J, Connell J. (2007) Effectiveness of cognitive-behavioural, person-centred, and psychodynamic therapies in UK primary care routine practice: replication in a larger sample. Psychological Medicine. 2007 Published online: 10 September 2007. doi:10.1017/S0033291707001511.


Tryon, G S, & Winograd G. (2002). Goal Consensus and Collaboration. In J. C. Norcross (Ed.), Psychotherapy relationships that work (pp. 106-122). New York: Oxford University Press.


Wampold B E. (2001). The great psychotherapy debate: Models, methods, and findings. Hillsdale, NJ: Lawrence Erlbaum.


Wampold B E, Mondin G W, Moody M, Stich F, Benson K, Ahn H. (1997). A meta-analysis of outcome studies comparing bona fide psychotherapies: Empirically, “all must have prizes.” Psychological Bulletin, 122, 203–215.


Wierzbicki M and Pekarik G (1993). A Meta-Analysis of Psychotherapy Dropout Professional Psychology. Research and Practice, 24, 2, 190-195


Whipple J L, Lambert M J, Vermeersch D A, Smart D W, Nielsen S L, Hawkins E J. (2003). Improving the effects of psychotherapy: The use of early identification of treatment and problem-solving strategies in routine practice. Journal of Counseling Psychology, 50, 59– 68.

                                                                                                                    back to top



Allegiance - the degree to which the therapist delivering the treatment believes that treatment to be efficacious

Alliance - the quality and strength of the therapeutic relationship between client and therapist

Bona fide therapies - therapeutic approaches that contain a rational, coherent psychological explanation of the client's problems or concerns, and the specific techniques or procedures for resolving them

Clinical change – describes a change in distress levels between those found in clinically representative settings and those in non-clinical or ‘normal’ populations. Change may be either clinical improvement or clinical deterioration (see also Reliable Change and Reliable and Clinically Significant Change)

Common factors - the factors that account for the majority of change in therapy and apply across therapies irrespective of model or type, such as the therapeutic alliance, the hope or expectation of change, goal consensus and collaboration, and empathy

Dodo bird effect - the proposition that, because it is the common factors that are primarily responsible for change in therapy, bona fide therapies are equivalent in their outcomes

Efficacy - the capacity of an intervention to bring about an effect or desired results, generally determined in highly controlled trail or clinical settings

Effectiveness - the degree to which an intervention brings about a desired effect in routine or more clinically representative settings

Effect size - the degree of relationship between two variables, for example receipt of therapy and improvement in levels of psychological distress

Empathy - the capacity of the therapist to enter the private perceptual world of the client and have an accurate, felt understanding of their experiencing

Goal consensus - agreement between client and therapist about the goals or aims of a therapeutic relationship

Meta analysis - meta-analysis is a quantitative or statistical process that aggregates the findings of similar studies

Randomised controlled trial - an experimental condition designed to test for efficacy of an intervention, under which participants are randomly assigned to treatment groups, and usually involving other tightly controlled treatment conditions such as use of treatment manuals

Recovery - synonymous with Reliable and Clinically Significant Improvement (see below)

Reliable Change - in therapy research terms, improvement is measured by a change towards a more desired state in, for example, a client's wellbeing, symptoms or problems or functioning. For improvement to be reliable it should be to an extent that cannot be attributable to chance or measurement error (see also Clinical Change and Reliable and Clinically Significant Change)

Reliable and Clinically Significant Change – Change which is both clinical and reliable, and may be either Reliable and Clinically Significant Improvement or Deterioration. Reliable and Clinically Significant Improvement (a.k.a. recovery) is generally seen as the gold standard of improvement, in which a client in therapy moves from a clinical to a non-clinical level of distress and the degree of change is also reliable.

                                                                                             back to top
The Dodo Bird effect

Everybody has won, and all must have prizes










Home   I   Therapy   I   Coaching   I   Service improvement   I   About me   I   Contact

© 2011 Barry McInnes