We all know what empathy is. Or, at least we think we know. But what is it really, how is in expressed in the therapeutic relationship, and just how much difference does it make to the outcome of therapy? An updated meta-analysis of the links between therapist empathy and outcome offers some important perspectives.

Only last night (as I write this) I sat with a client as he described the experience of being with his son and literally ‘feeling his pain.’ For this client, the experience was a profound one. As therapists, I’m sure we’ll know that experience well. The ability to step into the experience of another, and to be helpful from that place, can be truly transformative.

Setting aside the fact that empathy appears not to be a uniquely human characteristic, our ability to mirror the experience of another is a remarkable trait. As therapists, we need to hold this ability lightly and cautiously. At times there can be a fine line between ‘your stuff’ and ‘my stuff’, and I’m sure I don’t need to labour on about the dangers of projective identification.

By way of an amusing detour, I’ll also mention that empathy has nothing to do with anthropomorphism, the attribution of human characteristics or behaviour to a god, animal, or object. If you happen to believe that you are immune to this tendency, dwell for a moment on the image at the top of this blog and see what feelings, if any, it invokes in you. Alternatively, spend two minutes (from one minute in) watching a clip from the Graham Norton show from 2014. If you don’t feel anything for the little robot, you may have a heart of stone.

How do we define and measure empathy?

A paper from 2018 reports an updated meta-analysis on the connection between therapist empathy and client outcome. It builds upon two previous meta-analyses published in 2002 and 2011 that are contained in John Norcross’ two excellent editions of Psychotherapy Relationships that Work.

The paper, by Robert Elliot and colleagues, updates those earlier studies and provides the most up to date evidence for the impact of empathy on client outcomes, which I’ll summarise in a moment. Recognising that there is no one single definition of empathy, it also provides some helpful definitions and a case example. Empathy is described as a “higher order process under which different subtypes, aspects and modes can be nested.” Three main modes of therapeutic empathy are distinguished:

The establishment of empathic rapport and support

in which the therapist “exhibits a benevolent compassionate attitude toward the client and tries to demonstrate that he or she understands the client’s experience, often to set the context for effective treatment.”

Communicative attunement

consisting of an “active effort to stay attuned on a moment-to-moment basis with the client’s communications and unfolding experience….The therapist’s attunement may be expressed in many ways, but most likely in empathic responses”

Person empathy or experience-near understanding of the client’s world

emphasising “a sustained effort to understand the historical and present context or background of the client’s current experiencing. The question is: How have the client’s experiences led him or her to see/feel/think/act as he or she does?”

To what extent does empathy predict outcome in therapy? In its operation, empathy may be viewed from three perspectives: those of the therapist (empathic resonance), the observer (expressed resonance) and the client (received empathy). Most measures of empathy fall into one these categories, to which the authors have added a fourth: empathic accuracy. More recently this latter category appears to focus on assessing communicative attunement, using recordings of sessions to compare the moment to moment perceptions of therapist or observers to those of clients.

….on the other hand, not all clients respond favorably to explicit empathic expressions. One set of reviewers (Beutler, Crago, & Arizmendi, 1986) cited evidence that highly sensitive, suspicious, and oppositional patients perform relatively poorly with therapists who are particularly empathic and involved.

Elliott et al: Therapist Empathy and Client Outcome: An Updated Meta-Analysis

In the next section we highlight the key findings of the study, including the question of which of these four categories of measure seem to be most predictive of outcome.

What the study found

Overall, the study found a significant association between empathy and outcome. The paper states: “Probably the single best summary value…….. is the study-level random effects weighted r of .28 (95% confidence interval [.23, .33]), a medium effect size (equivalent to d = .58).”

Translated into human, this is a medium sized effect in social science research terms and was similar to what was found in the previous reviews. Empathy accounts, therefore, for about 9% of the variance in therapy outcome. This is an effect size of similar magnitude to that found for the relationship between the therapeutic alliance and outcome in individual therapy. It dwarfs the effects found for specific treatment methods by Wampold, thought to account for less than 1% of the variance in outcome.

Key findings

Clear association between empathy and outcome

Weighted r = 0.28; equivalent to d = 0.58, a medium sized effect

Empathy accounts for about 9% of variance in outcome

Similar effect size as found for the alliance / outcome association

Strongest predictor of outcome is client rating of therapist empathy

Client ratings of therapist empathy proved to be the strongest predictors of outcome. Observer rated and therapist measures were also predictive of outcome, though not to the same degree. By contrast, measures of empathic accuracy were not related to outcome.

What does it look like?

The paper offers a helpful case example of different empathic interventions deployed within a fictitious therapeutic conversation with client “Rick”. For those of you who, like me, have forgotten all our empathic terminology, it provides a useful distinction between, for example, the uses of empathic affirmation, empathic conjecture, exploratory reflection, and process reflection.

At the conclusion of the paper the authors draw together a range of recommendations from the research. Copyright considerations prevent me from replicating them all here, but below I’m highlighting a select few that seem particularly valuable to hold in mind.

Empathic therapists do not parrot clients’ words back or reflect only the content of those words; instead, they understand their clients’ goals and tasks, their moment to moment experiences in the session, and their unspoken nuances and implications.

Empathy is shown as much in how well the therapist receives, listens, respects, and attends to the client as in what the therapist does or says.

Therapists should neither assume that they are mind readers nor that their experience of the client will be matched by the client’s experience. Empathy is best offered with humility and held lightly, ready to be corrected.

Empathy is a co-created experience between a therapist trying to understand the client and a client trying to communicate with the therapist and be understood.

Certain fragile clients may find the usual expressions of empathy too intrusive, whereas hostile clients may find empathy too directive; still other clients may find an empathic focus on feelings too foreign. Effective empathic therapists know when—and when not—to respond with more or less empathically oriented responses


Elliott, Robert and Bohart, Arthur C. and Watson, Jeanne C. and Murphy, David (2018) Therapist empathy and client outcome : an updated meta-analysis. Psychotherapy, 55 (4). pp. 399-410. ISSN 0033-3204. Full text available here: https://strathprints.strath.ac.uk/66200/


Leave a comment

We thrive on feedback, and we’d love yours. Please leave your thoughts on what you’ve read in the comments section below.

Subcribe to the bulletin

If you haven’t already, subcribe to our regular (approximately fortnightly) bulletin for information about our latest blog and other news. You’ll find details here: 

Share with your networks
Posted by:Barry McInnes

8 replies on “The link between empathy and outcome

  1. Hi, that’s really interesting. Have you seen the paper in JAMA about predictors of outcome from 900,000 hours of CBT transcripts, interestingly, empathy wasn’t as high as you might expect. However, certain aspects, like therapeutic praise were high. It’s definitely worth a look. Whilst there are problems with AI searching transcripts for keywords, it’s interesting what it picked up nonetheless and the first of its kind and with a data set we couldn’t ever get close to in years otherwise. It’s by Ewbank et al 2019 – many thanks for another great blog

    1. Hi Marie
      Many thanks for your post. I hadn’t seen the JAMA paper is must admit but I will be sure to look it out. I’ll be interested to hear the conclusions of the study, but I’m not surprised that praise seem to be highly rated. It sound like an interesting read and thanks for the heads-up!

  2. Really enjoyed reading about empathy, how important this is in the counselling relationship, and humility also very important i feel, so nice to see this included too.

    1. Hi Cheryl and many thanks for taking the time to comment. So glad that you enjoyed the blog, and agree with your comment about humility. I’m not sure if you’re familiar with the paper by Helen Nissen-Lie: Love Yourself as a Person, Doubt Yourself as a Therapist? Her study came to the same conclusion. It’s profiled in one of our earlier blogs http://therapymeetsnumbers.com/wake-up-and-smell-the-evidence/

      Nissen-Lie HA et al. Love Yourself as a Person, Doubt Yourself as a Therapist? Clinical Psychology and Psychotherapy 2015; 24(1): 48–60.

  3. Thanks again Barry for a great summary of an up to date research paper.
    I have recently started working with RCN as an affiliate counsellor who offer brief telephone counselling to members. I was unsure how I would take to telephone counselling as I practice somatic ally using Focusing. However it’s going really well and on reading this article I’m so heartened to identify the specific aspects of empathy – ” receives, listens, respects and attends”.
    Marie Mc

    1. Hi Marie and thanks for the thumbs up! I’m not sure if you’re aware that I was with the RCN counselling service between 1994 and 2005, but I too had my doubts when I started. Lots of “it’s not counselling” or “it can’t be done”. Well, it is can it can. And we also found that our outcomes for face to face v. phone counselling were……..wait for it…….equivalent! Phone is a different medium for sure, but we have the same listening bandwidth, and my conclusion is that we learn to listen differently and attend to different things. Seems to work……. Thanks again ?

  4. Hi Barry,
    I’m fascinated by the thoughts around when therapist empathy may be too much! From a theoretical person-centred standpoint such a thing had never occurred to me, but practically I have sensed, rarely but definitely, that my responses were being felt almost as an invasion of privacy by the client – an interpretation I had discounted as unlikely. However if I get that feeling again, that is likely to be my tentative reflection.

    On several occasions, with EAP clients, I had come to feel that therapy wasn’t actually wanted by the client, but they wanted to use my involvement as a strategic pawn in a grievance procedure with management – “see how bad you have made me feel” and further attempts by me to engage in a counselling relationship were rebuffed with some anger. This new possibility, that it may be an unconscious and unwanted defence with some clients, needs consideration. Thanks!

    1. Hi David
      It’s endless learning, this therapy business! I guess that most clients are yearning to be seen and understood, but for some it may be uncomfortable or even threatening…..”not so close, thanks very much!” I’ve had that experience from time to time too, and I too found this angle on empathy very helpful.
      Thanks for your thoughts!

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.