A client calls at five o'clock with a problem that is entirely of their own making, and you are warm and helpful about it. That is a real piece of work, it has a name in the research literature, and it has a measurable cost that depends on how you do it.
Key Takeaway
A meta-analysis of 494 individual correlations from 95 independent studies found substantial relationships of emotion-rule dissonance and surface acting with indicators of impaired well-being, at correlations between .39 and .48, and with job attitudes between -.24 and -.40. By contrast, deep acting displayed weak relationships with indicators of impaired well-being and job attitudes but positive relationships with emotional performance and customer satisfaction, at .18 and .37[1].
A Note On Scope
Stated first.
This article describes occupational research on emotion regulation at work. It is not medical, psychological or human resources advice, and nothing in it should be used to assess anyone's wellbeing, including the reader's.
The research below concerns associations measured across groups. It says nothing about any individual, and this article does not offer a technique for managing your own emotions.
If your work is affecting your wellbeing, speak to a qualified professional. That is the appropriate response and no research summary substitutes for it.
Our Grades For These Claims
Applying the scheme from the first article in this series.
That surface acting is associated with impaired wellbeing is Grade A. Ninety-five independent studies, nearly five hundred correlations, corroborated by at least one further meta-analysis, with effect sizes among the largest in this series.
That surface acting causes impaired wellbeing is Grade C, and the distinction is the subject of its own section. The design is correlational and the arrow could run either way.
That deep acting is comparatively benign is Grade B, from the same analysis, with a caution that a later source disputes reading it as protective.
That the resource-depletion explanation is correct is Grade C; the paper itself records that this assumption has been challenged.
A Note On Method
Everything here is verified to August 2026.
We obtained the published abstract in full from PubMed, corroborated by an institutional repository and a bibliographic record[1][2][3], and fragments of the paper's introduction from the authors' institutional copy[4].
We did not obtain the full text and report nothing beyond the abstract and those fragments, several of which are truncated mid-sentence.
We did not obtain Hochschild (1983), Grandey (2000), or any of the primary studies, and report citations only.
Two qualifying findings, one on cultural variation and one on whether deep acting is protective, reach us through text citing the meta-analysis on publisher pages, not from the studies themselves.
All conversions of correlations into other effect size metrics are ours.
Three Things, Not Two
The distinction the research makes, and popular treatments almost always collapse it to two.
The meta-analysis examines emotion-rule dissonance, surface acting, and deep acting[1].
Describing each, with our own gloss.
Emotion-rule dissonance is the gap itself: the mismatch between what you actually feel and what the role requires you to display. It is a situation, not an action.
Surface acting is closing that gap at the display end. You change your face and your voice and leave the feeling where it was.
Deep acting is closing it at the other end. You work on the feeling itself so that the display becomes accurate.
Two observations, ours.
Separating dissonance from surface acting matters enormously, because dissonance is a property of the job and surface acting is a response to it. One is something an employer creates; the other is something a person does.
And a source notes a fourth strategy appears in some treatments: expression of naturally felt emotion[5], which is the case where no dissonance arises at all.
The Meta-Analysis
The evidence base.
Hülsheger and Schewe published On the costs and benefits of emotional labor: a meta-analysis of three decades of research in the Journal of Occupational Health Psychology, 16(3), 361–389, in July 2011[1].
The scale: 494 individual correlations drawn from a final sample of 95 independent studies[1].
Two observations, ours.
Ninety-five independent studies is a substantial literature, and three decades of it. This is not a fragile finding.
And the paper's acknowledgements thank researchers providing information about unpublished work[4], our source truncating there. Soliciting unpublished data is what you do to address publication bias, which the first article in this series identified as the central threat to any meta-analysis.
The Numbers
The findings, quoted.
"Results revealed substantial relationships of emotion-rule dissonance and surface acting with indicators of impaired well-being (ρs between .39 and .48) and job attitudes (ρs between -.24 and -.40) and a small negative relationship with performance outcomes (ρs between -.20 and -.05)."[1]
"Overall, deep acting displayed weak relationships with indicators of impaired well-being and job attitudes but positive relationships with emotional performance and customer satisfaction (ρs .18 and .37)."[1]
Three observations, ours.
The wellbeing figures and the performance figures are very different in size. Surface acting's association with impaired wellbeing runs .39 to .48; with performance it runs .05 to .20. The personal cost is far better evidenced than the business cost.
Which means an employer motivated purely by output would find weak grounds here for changing anything. The strong finding is about the person doing the work.
And note that dissonance and surface acting are reported together in the first sentence. The abstract does not separate their coefficients, so we cannot tell you which of the two carries more weight, and we do not claim to.
How Large That Is
Putting the figures in context, with our own conversion.
Applying the standard transformation from a correlation to a standardised mean difference: ρ = .39 corresponds to about d = 0.85, and ρ = .48 to about d = 1.09. In shared variance, those are roughly 15 and 23 percent. All four figures are ours and appear in no paper.
Against what this series has reported: brainstorming penalty at 1.39 on our own conversion, the disputed nudge headline at 0.43, feedback interventions at 0.41, growth mindset interventions at 0.080, the actor-observer asymmetry between -0.016 and 0.095.
Two observations.
On our conversion, the surface acting and wellbeing association is the second largest effect reported anywhere in this series, behind only the brainstorming productivity loss.
And it is worth stating what fifteen to twenty-three percent of shared variance is not. It is not most of the variance. Plenty of people surface-act extensively and are fine, and plenty who never do it are not. This is a population-level association, and the next section explains why even that overstates what we can conclude.
The Arrow Problem
The limitation that a careful reader must apply. This section is our own.
These are correlations. The letter is ρ, not d, and the meta-analysis synthesises observed associations rather than experimental manipulations.
Three readings are consistent with the same data.
Surface acting depletes you. The reading everyone assumes.
Depletion produces surface acting. Someone exhausted has less capacity to genuinely summon the required feeling, so they fake it instead. On this reading the acting is a symptom.
Or a third factor drives both. A punishing job with hostile customers and no autonomy would plausibly produce more dissonance, more exhaustion, and worse attitudes, without either causing the other.
Two consequences.
An intervention aimed at getting people to surface-act less may do nothing, if the arrow runs the other way or if both are downstream of something else.
And this matters practically, because it points at a different target. If dissonance is upstream of both, then reducing the gap is more promising than coaching the response to it, which is where this article ends up.
What The Mediation Adds
One structural finding that supports that reading.
The paper reports: "a mediation analysis confirms theoretical models of emotional labor which suggest that surface acting partially mediates the relationship of emotion-rule dissonance with well-being."[1]
Three observations, ours.
Mediation means dissonance is upstream. The chain runs from the gap, through the acting, to the wellbeing outcome.
Partially is the operative word. Some of dissonance's relationship with wellbeing runs through surface acting, and some does not. So the gap itself carries a cost even when it is not being closed by faking.
And a mediation analysis on cross-sectional correlations does not establish the direction either. It shows the pattern is consistent with the model. It cannot rule out the reverse chain, and we did not obtain the paper's own discussion of this.
The Other Column
The comparatively good news.
Deep acting showed weak relationships with indicators of impaired well-being and job attitudes, and positive relationships with emotional performance and customer satisfaction at .18 and .37[1].
Two observations, ours.
The customer satisfaction figure of .37 is not small. On our own conversion that is about d = 0.80, which is a substantial association and the strongest performance-side finding in the abstract.
So the pattern across the two columns is strikingly clean. One strategy carries the wellbeing cost and delivers little; the other carries little cost and delivers the customer outcome. If that reflects causation, it is an unusually actionable result.
Not A Prescription
The caution that must accompany the previous section, and it is important. Ours, with one source.
The obvious reading is: train people to deep-act. We think that is premature, for three reasons.
The arrow problem applies equally here. People who are already fine may be the ones with the capacity to deep-act, rather than deep-acting making them fine.
Text citing this meta-analysis on a publisher page states directly that deep acting is not inherently protective, and that when deep regulation is accompanied by clear boundaries, it may support professional identity; when it serves open-ended pressure for satisfaction, it may become a form of self-depletion[6]. We did not obtain the study behind that and report it as that text describes it.
And there is something uncomfortable about an employer instructing staff to change what they actually feel rather than merely what they display. Surface acting at least leaves the person's interior alone. We raise that as an ethical observation, not a finding.
The Proposed Mechanism, And Its Challenger
Why surface acting is thought to cost something, and why to hold that loosely.
The paper's introduction records that it has been argued that surface acting, a response-focused form of emotion regulation, requires considerable mental resources, our source truncating there[4].
Crucially, it continues that to the extent that surface acting draws on such resources, various consequences follow, but that recently, this assumption has been challenged by Liu and colleagues[4].
We did not obtain that challenge and report only that the paper records it.
Two observations, ours.
This is a resource-depletion model, and this series has now encountered three. Ego depletion collapsed under a twenty-three laboratory replication in the first article. The bandwidth account of financial worry was separately contested in the seventeenth. And here the paper's own authors note the assumption has been challenged.
Our general position, which the first article set out: a resource metaphor in psychology should be held loosely by default. The association here is well evidenced. The explanation for it is not.
Why Firms Require It
The organisational logic, from the paper.
The introduction states that the management of emotions has become part of organizational rules and occupational norms because organizational decision makers as well as employees believe that the expression as well as suppression of certain emotions helps influence customers and clients to meet higher-order performance goals[4].
Three observations, ours.
Note believe. The paper describes this as a belief held by decision makers, not as an established fact, and the abstract's own performance figures are modest on the surface-acting side.
Note also as well as employees. This is not purely imposed. People in client-facing roles frequently hold the same belief and impose the rule on themselves, which matters for anyone thinking they can fix it by changing a policy.
And suppression is named alongside expression. Not showing irritation is emotional labour in exactly the same sense as manufacturing warmth, and it is the more common form in professional services.
A Cultural Boundary
A limit on generalisation, from text citing the meta-analysis.
Text on a publisher page reports a study finding that in terms of the outcomes of emotional labor, our findings in China are different from the previous research in Western cultures regarding the robust relationship between surface acting and burnout, citing this meta-analysis among others[6].
We did not obtain that study and report only this description.
Two observations, ours.
If it holds, the relationship is not culturally universal, which matters for any Canadian workplace with an internationally diverse team.
And the phrase previous research in Western cultures is a reminder about the sample composition of the underlying literature, which we did not obtain and cannot describe.
Do People Pick One?
A methodological question raised by later work.
A peer-reviewed paper notes that findings on the three strategies are predicated on compartmentalized treatment of the three strategies, and that a fundamental yet unresolved question concerns whether employees adopt more than one type of emotional labor in the workplace[5].
We obtained only this framing and none of that paper's results.
Two observations, ours.
The question is obviously right. Nobody chooses a strategy for the year. A person deep-acts with a client they like, surface-acts with one they do not, and expresses naturally felt emotion with a third, possibly within the same afternoon.
Which means the meta-analytic correlations describe people who report doing more or less of each, not people who do one or the other. That is a weaker construct than the clean two-column story suggests, and it is a reason to treat the practical implications with more caution than the effect sizes alone would warrant.
The Professional Services Case
The application to this publication's readers. Ours, and not tested by anything cited.
Accountancy and advisory work is not usually classified as emotional labour, because the literature grew out of studies of flight attendants and call centres. We think that is a misclassification, for four reasons.
The suppression form dominates. Not visibly reacting to a client's records, a missed deadline, or a decision taken against your advice is squarely within the paper's definition.
The dissonance is frequently high, because the adviser often knows something the client does not want to hear and must deliver it warmly.
The display rules are unwritten, which is arguably worse than a formal policy, because there is no stated standard and therefore no defensible limit.
And it is concentrated in the same weeks as the workload, since filing season combines volume, deadline pressure and the highest density of difficult conversations.
The Owner Who Cannot Stop
A specific structural problem. Our own reasoning.
Three features of owning a client-facing business.
The owner has no display rule imposed on them and cannot escape one either, because the client relationship is the asset. There is no policy to appeal to and no manager to raise it with.
The dissonance is higher, not lower, because the owner carries the commercial consequence of the conversation as well as its content. Losing the client costs them personally.
And there is no shift end. A flight attendant's display rule stops at the aircraft door. An owner's phone does not.
We are not going to prescribe anything here. What we will say is that if the association reported in this meta-analysis reflects anything real, the owner-operator in a client-facing firm is structurally among the most exposed people it describes, and is the least likely to be counted, because none of this research was conducted on them.
Attack The Dissonance, Not The Acting
Where we think the practical leverage is. Ours, following from the mediation result.
If dissonance is upstream, and surface acting only partially mediates its relationship with wellbeing, then coaching people on how to act is addressing the middle of the chain and leaving the source in place.
Four ways to reduce the gap rather than the response, all ours and none tested.
Change who is in the conversation. Dissonance is specific to a pairing. The person for whom a given client is difficult is not necessarily everyone.
Reduce the frequency of the trigger. A client who generates the same difficult conversation monthly is a process problem before it is an emotional one.
Make the display rule explicit and bounded. An unwritten expectation of unlimited warmth has no ceiling. A stated standard does.
And decline the work. The one intervention that removes the dissonance entirely is not having the client, which is a commercial decision that this literature at least lets you price.
What To Do
Recognise the three constructs. The gap between felt and required emotion, faking the display, and changing the feeling are different things with different reported consequences.
Note that the wellbeing evidence is far stronger than the performance evidence. Surface acting's association with impaired wellbeing runs .39 to .48; with performance it runs .05 to .20.
Do not read the correlations as causation. Exhaustion producing surface acting fits the data as well as the reverse, and a third factor could drive both.
Target the dissonance rather than the acting. The mediation is partial, so the gap itself carries a cost that coaching the response does not remove.
Count suppression as emotional labour. Not showing irritation is within the definition, and it is the dominant form in professional services.
Be wary of training people to deep-act. One source states it is not inherently protective, and instructing staff to change what they feel rather than what they show raises a question the research does not answer.
Watch the calendar. In advisory work the difficult conversations cluster in the same weeks as the volume.
Treat a recurring difficult client as a process problem. The same conversation every month is generated by something upstream of the conversation.
If it is affecting you, speak to a professional. This is research about populations, and nothing here assesses any individual.
The Limits Of This Analysis
Several caveats matter. This article describes occupational research and is not medical, psychological or human resources advice; nothing in it should be used to assess anyone's wellbeing. Anyone affected should speak to a qualified professional. Everything is verified to August 2026. We obtained the published abstract in full and fragments of the introduction, but not the full text, and several fragments are truncated mid-sentence and reported as such. The findings are correlational; the causal direction is not established by the design, and we set out three readings consistent with the same data. The abstract reports emotion-rule dissonance and surface acting together in its wellbeing finding, so we cannot separate their contributions and do not. The mediation analysis rests on cross-sectional correlations and cannot establish direction. We did not obtain Hochschild (1983), Grandey (2000), the Liu and colleagues challenge to the resource account, the study on cultural difference, the study on whether deep acting is protective, or any primary study, and report citations and descriptions only. Two qualifying findings reach us through text citing the meta-analysis on publisher pages rather than from the studies. All conversions of correlations into other metrics are ours and appear in no paper. The professional services and owner-operator sections are our own extension to populations this research did not study, and the four suggestions for reducing dissonance are untested. Roughly fifteen years of subsequent literature was not reviewed.
Frequently Asked Questions
What is emotional labour?
How big is the cost?
Does faking it cause the harm?
Should we train people to deep-act instead?
Does this apply to professional services?
What actually helps?
References
- Hülsheger, U. R., & Schewe, A. F. (2011). On the costs and benefits of emotional labor: a meta-analysis of three decades of research. Journal of Occupational Health Psychology, 16(3), 361–389. DOI 10.1037/a0022876, published abstract via PubMed, on the article providing a quantitative review of the link of emotional labor, comprising emotion-rule dissonance, surface acting and deep acting, with well-being and performance outcomes; on the meta-analysis being based on 494 individual correlations drawn from a final sample of 95 independent studies; on results revealing substantial relationships of emotion-rule dissonance and surface acting with indicators of impaired well-being at correlations between .39 and .48 and job attitudes between -.24 and -.40, and a small negative relationship with performance outcomes between -.20 and -.05; on deep acting displaying weak relationships with indicators of impaired well-being and job attitudes but positive relationships with emotional performance and customer satisfaction at .18 and .37; and on a mediation analysis confirming theoretical models suggesting that surface acting partially mediates the relationship of emotion-rule dissonance with well-being. Note: we obtained the published abstract in full; we did not obtain the full text. All reported figures are correlations, not experimental effects. pubmed.ncbi.nlm.nih.gov
- Maastricht University institutional repository record for Hülsheger and Schewe (2011), confirming the citation, journal, volume, issue, pages and date, and reproducing the abstract identically. Note: an institutional repository record, used as independent corroboration of the abstract text and citation. cris.maastrichtuniversity.nl
- Semantic Scholar record for Hülsheger and Schewe (2011), confirming the citation and reproducing the mediation finding; and carrying indexed descriptions of related work including that emotional labor refers to the process of regulating both feelings and expressions in response to display rules for promoting organizational goals, and that multilevel analyses have shown surface acting to be detrimental to well-being and more strongly so for high than for low punishment-sensitive individuals. Note: a bibliographic record with third-party indexed summaries; we obtained none of the studies described. semanticscholar.org
- Authors' institutional copy of Hülsheger and Schewe (2011), introduction fragments, on the management of emotions having become part of organizational rules and occupational norms because organizational decision makers as well as employees believe that the expression as well as suppression of certain emotions helps influence customers and clients to meet higher-order performance goals; on the argument that surface acting, a response-focused form of emotion regulation, requires considerable mental resources, our source truncating; on the assumption that surface acting draws on such resources having recently been challenged by Liu and colleagues; on felt inauthenticity appearing as a model component; on the model being based on Grandey (2000), Holman and colleagues (2008b) and Rubin and colleagues (2005); on Hochschild (1983) as the originating citation; and on the authors thanking researchers providing information about unpublished work. Note: an institutional PDF; we obtained the first pages only, and several passages are truncated mid-sentence and reported here as fragments without completion. We did not obtain the Liu and colleagues challenge. cris.maastrichtuniversity.nl
- Journal of Individual Differences, 36(1), Emotional Labor and Occupational Well-Being: A Latent Profile Analytic Approach, publisher record, on emotional labor having been conceptualized over three decades as comprising three strategies, namely surface acting, deep acting, and expression of naturally felt emotion; on research suggesting each strategy relates differently to health and job outcomes such as burnout and job satisfaction; on these findings and conclusions being predicated on compartmentalized treatment of the three strategies; and on a fundamental yet unresolved question concerning whether employees adopt more than one type of emotional labor in the workplace. Note: a publisher record; we obtained the framing only and none of that paper's results. econtent.hogrefe.com
- Publisher and citation-index pages for Hülsheger and Schewe (2011) carrying third-party text citing it, including a study reporting that its findings in China differ from previous research in Western cultures regarding the robust relationship between surface acting and burnout; and a separate passage stating that different combinations of emotional labor may have different implications for wellbeing, that deep acting is not inherently protective, and that when deep regulation is accompanied by clear boundaries it may support professional identity while when it serves open-ended pressure for satisfaction it may become a form of self-depletion. Note: third-party text on publisher and citation-index pages, not peer-reviewed in itself. We did not obtain either underlying study and report both only as this text describes them. scite.ai
This article describes occupational research and is not medical, psychological or human resources advice. Nothing in it should be used to assess anyone's wellbeing; anyone affected should speak to a qualified professional. The meta-analysis was not obtained in full and several quoted introduction passages are truncated mid-sentence. All reported figures are correlations and the causal direction is not established. Conversions into other effect size metrics are the authors' own. The professional services and owner-operator sections are extensions to populations this research did not study.